aiLL 7/4/23 - The Future of AI: Optimism vs Fear

Description

Chapters

Transcript

0:01 foreign
0:26 welcome to the AI learning lab
0:29 um
0:30 let's see can I get that to look more
0:33 straight I don't want that side to come
0:36 up all right I don't know this looks
0:39 crooked to me
0:43 yeah that looks better
0:45 welcome welcome welcome my name is Kyle
0:47 Shannon this is the AI learning lab
0:50 we've got some excitement in the house
0:53 tonight
0:54 open AI
0:56 looks like they had a boo-boo they got
0:59 rid of there they got rid of their
1:01 internet so uh we'll we'll wait for some
1:04 some people to join my name is Kyle
1:07 Shannon
1:08 I'm a long time entrepreneur I've
1:10 started a bunch of companies in my day
1:13 I was there for the early days of the
1:15 World Wide Web I started one of the
1:17 first digital agencies in the mid 90s
1:18 and uh the parallels between what's
1:21 going on with generative Ai and what was
1:24 happening in the early days of the web
1:25 is nuts and so that's what this channel
1:28 is all about it's just talking about
1:29 generative Ai and answering questions
1:32 and doing what I can to
1:35 to share I don't know I suppose my
1:38 passion for what's going on but it's I
1:40 don't know it's more than just passion
1:42 I'm not just an Enthusiast I feel that
1:45 it's really important anyway
1:48 um welcome all uh if you have any
1:50 questions about AI pop them in the
1:51 comments below we're just getting
1:52 started so I'm waiting for some folks to
1:54 come in it looks like the numbers are
1:56 ticking up now so it looks like The Tick
1:58 Tock has has tossed us into the
2:00 algorithm so if you would go ahead and
2:02 hit
2:03 um hit double tap the screen hit the
2:05 like things that'll let people know
2:06 we're here
2:07 what's generative mean to AI well the
2:10 way the tools work if
2:13 if you're brand new to it and if you've
2:15 heard of chat GPT
2:17 the G and GPT is generative so
2:20 generative pre-trained Transformers what
2:22 that stands for and I'll talk about all
2:25 these URLs in a second uh but back to
2:28 the generative thing it's actually
2:29 generating original content so that's
2:32 what it means
2:34 um and there's a lot of tools out there
2:37 that are using the the the T in GPT
2:40 which is the Transformer model so
2:43 everything from mid journey and stable
2:45 diffusion that make those really pretty
2:46 images to chat gbt to the the music
2:50 tools that make music to the animation
2:52 tools that do all that all of that stuff
2:54 is based on the same basic uh
2:57 Transformer architecture and then the
2:59 generative piece means it's generating
3:01 original stuff
3:04 is a class of things called large
3:06 language models where basically the the
3:09 p in GPT stands for pre-trained meaning
3:12 it was pre-trained you don't have to
3:14 bring your data it's this really amazing
3:16 machine learning algorithm that's been
3:18 pre-trained what's it been pre-trained
3:20 on essentially all of the public
3:22 internet
3:24 so think about the last 35 or 40 Years
3:26 of crap that's been uploaded to the
3:29 internet you know brilliant horrible you
3:31 know good bad biased unbiased and
3:33 everything in between all of that is
3:36 what it's been trained on
3:38 tapping into the output of humanity
3:41 that's made it to the public internet
3:42 now there's you know I think that's five
3:45 or ten percent of what you know is out
3:47 there in the world you know most of the
3:49 data we can't see uh but it's still a
3:52 remarkably
3:53 massive set of data that that uh
3:57 attached apt has been trained on anyway
4:00 if you're new here welcome my name is
4:01 Kyle Shannon I'm a long time
4:03 entrepreneur and this is the AI learning
4:06 lab so if you've got questions about AI
4:08 pop them in the comments below I'll do
4:09 my best to answer them the URLs I have
4:11 on screen here's what they're there for
4:13 if you haven't played with chat GPT that
4:15 top URL that's the official chat GPT
4:18 website so go there start playing
4:21 um
4:22 the next one down is Bing so that's
4:25 Microsoft search engine and uh if you
4:29 click on the chat button there that's
4:31 actually gpt4 connected to the internet
4:33 which is now a bigger deal than it was
4:35 last night and I'll talk about that and
4:37 then you've got poe.com which gives you
4:40 access to six different large language
4:42 model models to play with three from uh
4:46 anthropic three of the Claude models two
4:48 from open Ai and one from quora from Poe
4:51 the the company behind Poe and then
4:54 prompts.chat that little URL down there
4:56 will teach you about how to use these
4:59 tools it'll teach you about what
5:00 prompting is so if you don't know what
5:01 prompting is go to prompts.chat it'll
5:03 teach you and then it'll give you lots
5:05 of ideas about
5:07 how you can sort of Bend chat GPT to
5:10 your will you can make it act like a
5:12 certain kind of assistant or
5:14 collaborator with you so act like an
5:17 accountant act like a mathematician act
5:19 like a creative writing coach whatever
5:21 it might be
5:22 um so prompts.chat gives you lots of
5:24 ideas for that and then futurepedia.io
5:26 that is a directory of uh uh sorry the
5:34 brain just shut down uh futurepedia.io
5:38 is a directory of AI tools so go play
5:42 with those so if you have questions uh
5:44 pop them in the comments below I'll do
5:46 my best to answer them uh if you've got
5:48 trolls please be creative about it like
5:50 boring trolls don't interest me creative
5:53 ones are good ideally a troll should be
5:55 one that I can't tell if it's a
5:56 compliment or a put down those are the
5:58 solid ones
6:00 um
6:01 what else happy Fourth of July happy
6:03 Independence Day uh looks like it's
6:05 about to do some major storming here in
6:08 Denver Colorado so uh maybe we'll get
6:11 some different kinds of fireworks some
6:13 some coming down from the sky as well
6:15 and as well as going up into the sky
6:17 all right let me see what questions we
6:20 have here if you were in sales what type
6:22 of AI would you sell now
6:24 or in the future to make the most money
6:28 um
6:30 that's a good question I would
6:34 I would investigate companies that are
6:37 going to make that are making it easy
6:41 for organizations to make their own data
6:45 gptable right so you've got chat GPT
6:48 that's been trained up on all of the
6:49 internet
6:50 one of the next Trends is going to be
6:52 well I want to be able to do that for
6:54 all of my company data so there are a
6:56 bunch of companies out there that are
6:58 starting to play in that Microsoft is
7:00 going to be doing a lot of that because
7:01 these generative AI tools are going to
7:04 be built into Azure cloud services
7:06 they're going to be built into windows
7:08 they're going to be built into Office
7:09 365 Dynamics they're already built into
7:13 GitHub GitHub with copilot it's all
7:15 going to be called copilot across the
7:17 Microsoft Suite
7:19 um
7:19 that feels like the next big Trend to me
7:22 um
7:25 yeah and then and then short term
7:28 I I think that you could probably sell
7:30 there's going to be a lot of agencies
7:33 and Consultants uh putting their putting
7:36 their hat in the ring going I can help
7:37 you figure out AI because because at
7:40 some point here shortly if it's not
7:41 already started there's going to be a
7:43 bit of a panic amongst companies going
7:45 we gotta figure this [ __ ] out
7:50 a lot of company companies out there
7:51 going we can help you so I think short
7:53 term probably consulting or
7:57 ai-centric agencies is probably a good
8:00 sell and then I think longer term
8:03 any companies or technologies that make
8:07 it easy for other companies to GPT eyes
8:10 their data because that's where my head
8:13 is right now
8:15 how do I generate videos on mid-journey
8:17 oh
8:22 you type
8:24 slash video I forget hang on
8:27 we can go that's something we can go
8:29 Google
8:31 oh I did a cool thing today in sheets
8:34 with GPT
8:36 that I can show you in a bit
8:39 um but let me let me go Google this
8:41 um I'm actually going to Bing it because
8:43 I have Bing as my default search engine
8:45 now
8:45 which I can't believe I've been a
8:47 Microsoft hater for three and a half
8:48 decades
8:50 um let's see um
8:53 show
8:57 videos in mid Journey
9:16 to create a mid-journey process video
9:19 you must add dash dash video parameter
9:21 to your prompt
9:23 then once the image is binging it
9:25 generated react to the Bots message by
9:28 selecting the envelope so let's go try
9:30 there's a mid-journey image that I made
9:32 for oh that's something I want to talk
9:34 about something I want to share with
9:36 everyone
9:37 so we're gonna go here we're going to go
9:38 slash imagine so if you haven't seen mid
9:40 Journey this is mid Journey you go into
9:42 uh
9:43 you go into Discord
9:46 and then you type in command line
9:48 command line
9:50 uh interfaces to make images and you
9:53 know why do you have to do that you ask
9:55 oh you know because so that's good uh
9:58 let's so let's see uh fireworks
10:01 over Manhattan
10:05 with Brooke Lynn Bridge
10:08 in for ground oh it's starting to rain
10:13 here in Denver I would think there's
10:15 going to be some Thunder Boomers soon so
10:17 I'm gonna go dash dash video
10:20 and I think that should be enough to
10:22 just do something
10:25 [Music]
10:32 all right so
10:34 here come the images if you haven't seen
10:35 mid-journey before any of these
10:37 generative AI things generate someone
10:38 asked earlier what's generative about AI
10:41 so it's actually generating these images
10:43 you notice how they start as kind of
10:45 just these blurry noise noisy kind of
10:48 images and then they get increasingly
10:50 less blurry
10:52 um that's the nature of the diffusion
10:54 model they start out as kind of a
10:56 mathematical average of a bunch of
10:59 noised images and then
11:01 it gets denoised over 50 or 70 steps
11:05 and generates an original image which is
11:09 super cool
11:10 the rain sheet is coming down good
11:15 is it rain I hope it doesn't hail too
11:17 big but you know okay so this is almost
11:21 done
11:23 and then
11:25 reply
11:28 and I think I have to do reply with the
11:30 envelope envelope
11:33 and you're asking why do you have to
11:34 reply with the envelope to get a video I
11:37 don't know because that's the way they
11:40 they made it
11:42 uh
11:44 that does not seem to be
11:46 doing it
11:48 um I don't know I don't know you have to
11:51 Google it I can't figure it out now
11:53 it'll drive people crazy if I just sit
11:54 here trying to figure it out uh anyway
11:57 uh welcome if you're new here my name is
12:00 Kyle Shannon this is the AI learning lab
12:02 um
12:06 a couple of things to talk about tonight
12:07 happy Fourth of July to you too uh if
12:10 you've got questions about AI pop them
12:12 in the comments below
12:14 uh sorry I couldn't figure that out with
12:16 videos on mid-journey I'll figure it out
12:17 at some point here tonight if you're if
12:19 you're still around
12:21 um I appreciate all your help man you're
12:23 a public servant oh I very much
12:25 appreciate that it's I have fun doing
12:26 this I I think it's important but I also
12:29 enjoy it so man it's storming like a
12:33 mofo out there wow oh wondering what I'm
12:36 drinking tonight I muddled some cherries
12:39 in the bottom of some cherry flavored
12:41 Seltzer because I know how to live huh
12:46 all right uh will it help lint
12:49 I don't know what that means you got a
12:52 burger what's the news okay so here's
12:54 the news was surprised to see Bing was
12:57 disconnected from gpt4 today that's the
12:59 news
13:00 cavruno has it all right so let me go
13:03 show you something
13:06 thing that you were a you know paid
13:09 member of GPT plus you've got access to
13:12 GPT 3.5 and gpt4 and there used to be
13:17 three options here there was default
13:18 browse with Bing and plugins
13:21 like literally within minutes of when I
13:25 stopped my live last night they pulled
13:28 browse with Bing as one of the options
13:30 I'm not saying it's related I I'm not
13:33 saying that open AI removed their
13:36 connection to the internet because they
13:38 was up they were upset with me that I
13:40 stopped my live
13:41 but I'm saying the timing was suspicious
13:44 um no what I think actually happened is
13:49 um so
13:50 about a week ago
13:53 um
13:54 chat GPT or no openai was served with a
13:58 lawsuit from
14:00 either a publisher or a group of
14:02 Publishers basically saying because chat
14:04 EPT was trained on public books that
14:07 they were the Publishers of that they
14:09 were going to sue them uh and then
14:11 apparently the browse with B Bing
14:15 feature of chat GPT
14:18 in some cases
14:20 pulled content from behind paywalls
14:24 right so like if you went to an article
14:27 for the New York Times and you weren't
14:29 you didn't pay for the New York Times
14:31 subscription you shouldn't be able to
14:33 see that but apparently
14:36 chat gbt could do that now what's
14:39 interesting is that the plugins that are
14:42 connected to the internet still work
14:45 so I haven't actually tried this today
14:48 um but someone in my
14:50 I made a video earlier so let's try it
14:52 we're gonna put on web pilot and then
14:56 we'll say let's go grab a URL let's do
14:59 this that's
15:00 that's the Twitter
15:03 um
15:15 here's docs.midjourney so here's some
15:17 documentation for mid-journey
15:20 uh this is documentation on how to do so
15:22 I'm going to copy this URL
15:24 we're gonna go to chat EBT and I'm gonna
15:26 say
15:28 um
15:29 please
15:30 summarize
15:33 spelled wrong this
15:35 now that should fire off the plug-in
15:38 that's connected to the internet
15:45 I love a good thunderstorm
15:47 man it's coming down
15:50 that's cool
15:52 all right so it worked it went out and
15:53 it read that it read the video parameter
15:57 to get a video link you add dash dash
15:59 video at the end of your prompt
16:01 you react
16:03 with the envelope Emoji I thought I did
16:06 that and it didn't do it anyway
16:08 so that works and then the other thing
16:11 that works still is if we go to bing.com
16:14 so this is one of the big three that I
16:17 put up there as things you should go
16:19 play with so if you click on the chat
16:22 option
16:24 this is now chat GPT and it's connected
16:27 to the Internet so I can say
16:29 is it going to hail tonight question
16:34 mark
16:35 oh I probably should have told it in
16:38 Denver
16:39 searching for Hail forecast
16:42 hi this is Bing I'm glad you asked
16:45 stop responding
16:47 I mean in Denver
16:53 so it should remember that I asked that
16:56 if it was going to hail yeah it says
16:57 hail forecast Denver
17:00 according to the interactive hail Maps
17:02 website there's a hundred
17:05 41 reports of on the ground hailed by
17:08 train spotters
17:11 in the past 12 months now I mean wait 12
17:15 months no
17:17 I mean tonight
17:19 do you have
17:22 two nights
17:24 weather
17:32 so what's AccuWeather says three percent
17:36 chance of precipitation three percent
17:39 chance it's [ __ ] pouring out
17:41 what the hell is going on all right
17:46 um it actually gave me a link to
17:48 AccuWeather I wonder if it's
17:50 current weather
17:56 thunderstorm
17:58 yeah I don't know if you got it anyway
17:59 so so it's interesting
18:03 about all of that stuff is
18:07 um
18:12 it's it's not just
18:15 anything related to open AI being
18:17 connected to the internet is bad and
18:19 broken it's just that chat EPT
18:21 specifically their implementation of how
18:24 they Incorporated Bing into their chat
18:26 was breaking something with Publishers
18:28 and I I assume my The Tick Tock I did
18:32 this afternoon I was like you know I'm
18:34 just imagining some lawyer walking into
18:36 someone's office going pull that [ __ ]
18:38 down immediately
18:39 hahaha
18:40 so I think that's what's going on there
18:43 um but anyway wild wild stuff super
18:46 weird it just started
18:48 um it just started raining this though
18:51 the same in India oh that's wild yeah no
18:53 it's coming down there's people on a
18:55 bike
18:56 they're wet
18:57 hello I'm new here our crypto trading
19:00 Bots real crypto crypto trading Bots uh
19:03 oh are like AI generated crypto trading
19:06 Bots are real they're very real
19:08 um
19:09 if you
19:11 AI Services I'm sure there's a whole
19:13 financial tool section in there
19:16 but what you can do if you want to learn
19:18 and you just want to get
19:19 um
19:20 get up to speed with this is go to your
19:22 crypto trading site and I ideally one
19:25 where you're not trading real money
19:26 right where you can trade some fake
19:27 money and do what is whatever it's
19:28 called paper trading
19:30 um and and one that's got Bots and then
19:34 you can you know grab the code from a
19:37 bot on that trading site you can paste
19:39 it into chat GPT and you can say here's
19:41 a bot tell me what it does and it'll
19:43 tell you and then you can say what are
19:46 some other
19:47 um
19:48 you know parameters or some other
19:50 options for how we can do a crypto
19:52 trading bot and tell it if you want it
19:54 to be more aggressive or more
19:55 conservative or more you know help you
19:58 identify things and it will rewrite the
20:00 bot for you you can pop it back in the
20:02 site and and start playing with it so
20:04 yeah they're very real whether or not
20:06 there are tools out there that are
20:07 actually good where you don't have to
20:09 kind of roll your own like I just
20:10 described uh that's that I don't know
20:14 because I'm not I'm I'm not very much in
20:17 that world
20:20 um welcome everybody welcome welcome
20:22 welcome welcome happy Fourth of July
20:25 not send emoji not send envelope Emoji
20:28 I've got to react with envelope emoji
20:32 wait are you are you
20:35 are you trying to tell um ad reaction I
20:38 see got it
20:42 this this is one of those
20:45 um
20:46 is it going
20:51 I reacted
20:54 this is like one of those um you know
20:57 giving old people remedial
20:59 um remedial uh access
21:04 oh that's really interesting so check
21:05 out what this did
21:08 huh
21:11 so it was I was in and and I reacted
21:14 with the thing and then it popped a new
21:16 server over here
21:18 um
21:22 mid-journey bot and so it created a
21:25 separate little thing
21:27 and now there's a video
21:34 that should
21:37 it should
21:41 stop playing
21:44 huh it's not populated
21:47 strange anyway all right so I got it so
21:50 you got to react to it not talk to it
21:52 fine it's like it's like teaching an old
21:55 guy how to use the internet I know it's
21:57 painful I get it I get it I get it
22:00 welcome to Elder talk
22:04 Grandpa got a computer
22:07 [Laughter]
22:10 oh my goodness
22:14 all right
22:17 so there's that welcome everybody if
22:18 you're new here my name is Kyle Shannon
22:20 this is the AI learning lab these URLs I
22:22 have on screen are there for your
22:24 viewing pleasure uh they're really there
22:26 if you haven't played with chat GPT that
22:28 top URL there is chat GPT go there start
22:31 playing with that
22:33 um bing.com is also chat GPT if you
22:35 click on the chat button it's gpt4
22:37 connected to the internet which by the
22:40 way newsflash open AI killed their
22:43 um their version of their gpt4 connected
22:46 to the internet yesterday after we uh
22:48 after I stopped going live for some
22:50 reason uh then poe.com is also you can
22:54 play with chat GPT there and then
22:56 prompts chat will teach you how to
22:57 prompt and future PD is a directory of
22:59 tools ask any questions you have about
23:01 AI below I'll do my best to answer them
23:03 we have fun here I cuss occasionally so
23:06 if you don't like that sort of thing I'd
23:08 get the hell out of Dodge
23:11 all right
23:15 does chat GPT plus give you dull e and
23:18 Whisper no
23:21 but paying for
23:28 what I what I don't know is if you pay
23:31 for chat GPT plus is that also
23:34 paying for the API access for credits on
23:38 API I don't think it is if you if you if
23:42 you want to access whisper you have to
23:44 go to platform
23:46 dot openai.com playground
23:49 and I can actually show you how to do
23:51 that but Dali
23:54 you can call Dali from from within no
23:57 code tools like zapier
23:59 you can also use dolly for free at
24:02 bing.com which I've got right there
24:05 um
24:08 yeah so there's
24:10 there's some interesting
24:12 um
24:13 some interesting things so so let me
24:16 just working on a prompt here earlier so
24:18 if you go to
24:20 platform.openai.com playground
24:23 then you come in here and you you've got
24:26 access to a lot of different models that
24:29 openai has available this is more if
24:31 you're a developer you can flip into
24:33 different modes chat mode complete mode
24:35 things like that and then there's also
24:38 in the upper right hand corner of the of
24:40 the thing here there's the this
24:42 microphone and this is whisper so if you
24:45 want to test their transcription and
24:47 things like that you can either record
24:49 things or upload video or audio files
24:52 here
24:53 um
24:54 and then
24:56 yeah and then if you want to play with
24:57 Dolly you can go to the there's a
24:59 website for Dolly
25:02 um or if you just want to use it you can
25:04 go to bing.com
25:06 and
25:09 so here we are at Bing
25:11 so I can say uh make me an image of a
25:17 thunderstorm over Denver on the 4th of
25:23 July
25:24 and it will just use Dolly to go make
25:28 something and this is free so if you
25:29 want to program with it
25:31 you need API access you can sign up for
25:34 API access at
25:38 platform.openai.com playground if you
25:41 just want to play with it you can just
25:42 play with it that was pretty quick
25:45 um let's see can you
25:49 add in fireworks
25:54 and hopefully it'll figure out what the
25:56 hell I'm talking about
25:59 sure I can try to add fireworks to this
26:01 image
26:05 the dolly stuff at Bing's pretty fast I
26:07 just I wish Bing were better or Dali
26:09 were better
26:11 it's not a very good image generation to
26:13 a mid journey is much much better
26:16 Boomers and fireworks
26:19 that one's kind of cool
26:22 nice
26:23 yeah
26:25 sweet
26:28 all right
26:29 so there you have that there you have it
26:31 there you have it
26:34 come on
26:35 [Music]
26:39 hello I'm new here oh we talked about
26:40 that
26:41 I get hives when someone mentions crypto
26:44 I know me too it's like I spent a year
26:46 of my life like deep deep deep diving
26:49 into that world
26:51 uh and it just I you know like the
26:55 Technology's really really really
26:56 interesting and the use cases were kind
26:59 of interesting and then the people were
27:01 just
27:02 money grubbing pigs and it [ __ ]
27:05 everything up and it's just gross and so
27:08 I do think that
27:12 um generative AI is gonna drive a lot of
27:14 adoption of of those tools uh once once
27:19 we have the uh 2024 election and all the
27:23 misinformation makes people realize uh
27:26 we gotta get our [ __ ] together and Trust
27:29 you know who we're getting this stuff
27:30 from
27:33 um
27:34 but I agree the crypto stuff is a little
27:37 Oogie
27:39 I went to use an AI I went to use AI to
27:42 automate an ongoing experience oh I want
27:45 to use
27:46 AI to automate ongoing experiment
27:49 thoughts
27:52 that's way too general for me to have
27:54 thoughts like I don't know which tool I
27:56 would recommend
27:58 um if you haven't played with chat GPT
28:00 so the chat.openai.com right there I
28:04 would go there and start playing if
28:06 you're
28:08 um if you're technical which I assume
28:10 you are if you're talking about
28:11 experiments I assume you're a scientist
28:13 and can you know write Python and and
28:17 crap like that
28:19 um if you can do that you know I would I
28:22 would be using uh
28:24 copilot at GitHub to do some pair
28:27 programming and that that can help so
28:30 um if you don't want to get into all
28:32 that you want to use no code stuff you
28:34 could probably do some automation using
28:36 tools like zapier so I've been doing a
28:38 lot of that I think it's potentially
28:40 really good but I would just start
28:42 experimenting at chat.openai.com
28:46 um take some of the data from your
28:47 experiment pop it in there figure out
28:49 what your prompts might be to be able to
28:51 uh you know figure out what calls you
28:54 need to make to open AI to do the
28:56 analysis but yeah I would think so
28:59 um if you tell me a little bit more I
29:00 can you know maybe throw some ideas
29:03 about their Gen X core
29:07 Gen X baby
29:09 let's go
29:10 can you use mid-journey to copy Amazon's
29:12 photos and have it change them a little
29:14 bit so it's your image yeah you can you
29:16 can do that I mean it's you've gotta you
29:18 you've got okay so here's the thing
29:21 mid-journey and stable diffusion have
29:24 been trained on a lot of copywritten
29:27 images and there's a lot of lawsuits
29:29 coming where
29:31 people are probably
29:33 there's gonna be a lot of lawsuits
29:35 coming if you work for a big
29:37 organization
29:38 and you don't want to deal with that
29:40 [ __ ] use Adobe
29:43 um
29:44 no leave that on please
29:47 all right
29:48 um
29:49 use Adobe Firefly and generative
29:52 generative fill within Photoshop
29:57 yeah
29:58 I'm not
30:03 [Laughter]
30:06 all right
30:08 um
30:09 so
30:12 so use adobe's
30:15 um
30:16 Firefly because it's been trained on
30:18 non-copy written data
30:21 now that said mid-journey is by far the
30:23 best of the image generating tools and
30:25 technically you can do
30:28 to Bing we'll go grab an image so we'll
30:31 go bing.com
30:33 we'll say I don't know image of uh the
30:38 buck tractor pull I grew up in York PA
30:41 so I've been to the buck tractor pull
30:43 Buck tractor pull
30:45 [Applause]
30:48 and we'll go images and there should be
30:51 some groovy yeah we'll just take that
30:53 one so so there's a there's a giant
30:56 tractor
30:59 with a bunch of smoke coming out of it
31:02 so so let's see I'm gonna I'm Gonna Save
31:05 this image as
31:08 a tractor
31:18 and then we're going to pop over here to
31:20 Mid Journey
31:22 and I'm gonna go
31:24 um I'm gonna go find that image that I
31:28 just saved to my hard drive
31:32 Buck tractor I'm gonna upload that so so
31:36 I just dragged it from the finder I'm on
31:38 a Mac so from the finder on a Mac or
31:40 whatever the [ __ ] it's called in Windows
31:42 File Manager or whatever so I just
31:44 dragged it from that into the Discord uh
31:47 into the mid Journey Channel and Discord
31:48 now I'm going to hit return
31:50 so what that's doing is that's uploading
31:52 this image to Discord which is a
31:54 requirement so then you right click or
31:57 control click on a on a Mac and you copy
31:59 the link for that image and so now I can
32:02 say slash imagine so I'm going to make
32:05 an image
32:06 and the first thing I'm going to do is
32:08 rather than type anything I'm gonna I'm
32:10 gonna paste the URL for that image so
32:13 we're basically doing an image based on
32:15 that image and so now I'm going to say
32:17 red
32:19 um
32:20 tractor
32:22 uh tractor
32:25 billowing
32:29 smoke and spinning
32:33 the wheels
32:36 while
32:37 crowd roars
32:41 spinning and then I'm gonna say
32:45 photo of
32:50 all right
32:53 oh and I put photo of after Red Tractor
32:55 but we'll see what it does so that
32:57 should now generate a something that
32:59 looks in the neighborhood of that but it
33:01 should be a red tractor instead of a
33:02 green one
33:06 okay
33:09 whether or not it does anything you know
33:11 remotely interesting or valuable I don't
33:14 know but you know
33:16 that's a way to do it it obviously
33:19 didn't it didn't it looks like it
33:21 grabbed the Green from that tractor
33:28 it's a completely different kind of
33:30 tractor although it is like spinning out
33:32 spinning out on a dirt track
33:34 which is kind of funny
33:38 it didn't make it red
33:40 so you can do that but this is this is
33:42 actually a good example the fact that it
33:43 didn't quite do it great
33:45 um you know
33:47 um we could we could potentially also do
33:49 that over in
33:51 um
33:53 with Adobe fill let me or adobe
33:56 generative Phil let me
33:58 try something here
34:00 let's go over to Adobe we're gonna we're
34:03 gonna go to photoshop I'm in Photoshop
34:04 beta
34:07 I'm gonna pop that in there and let's
34:10 say
34:12 let's capture the tractor
34:16 I'm gonna I'm gonna highlight this
34:18 section of the tractor I'm gonna say
34:25 I'm going to say generative fill oh you
34:27 can't see it God
34:29 damn it the
34:31 uh
34:34 come on
34:36 something about the dual camera does not
34:38 make the uh
34:40 the uh the focus work right so anyway
34:42 down at this little bar here I said
34:44 generate
34:46 and I'm going to type in red
34:51 racing
34:52 tractor let's see what it does let's see
34:55 if it actually
34:57 fills that in and keeps it mostly the
34:59 same and just makes it red
35:11 sort of looks like the front of front of
35:13 a truck
35:19 that one's not bad
35:21 yeah I mean you could mess around with
35:24 that you could probably get it that's
35:26 not I mean that's in the neighborhood
35:27 that's not quite right but it's it's in
35:29 the neighborhood and then you know then
35:30 you can do stuff like we could we could
35:33 grab the the street down here
35:36 and we could say generative fill
35:39 s
35:41 s
35:42 fault wait s fault with yellow line
35:50 and that should replace the dirt wow
35:53 this is crazy it just got like you know
35:56 how it gets uh that crazy orange color
35:59 when the storm gets crazy it's like
36:01 crazy orange color
36:06 let's see yeah there you go there's some
36:08 there's some asphalt didn't give us a
36:10 yellow line but maybe we have another
36:11 one here
36:13 there's a yellow line it's not in the
36:15 right direction
36:17 and then
36:19 rather than having people watching why
36:22 don't we make this the mountains
36:25 so I'm going to select that and I'm
36:27 going to select all this
36:30 and I'm going to say generative fill
36:33 Rocky
36:37 mountains
36:40 we'll see what that does
36:42 this could be interesting
36:45 man it's nuts out
36:53 look mountains
36:57 that's pretty cool
37:00 that's more like the Foothills there we
37:01 go there's there's your red tractor
37:04 in front of the mountains so if you
37:05 haven't played with generative fill
37:06 that's generative fill so you can
37:09 absolutely take other images and modify
37:11 them
37:12 um you can also if you want to get geeky
37:15 you can do use something in
37:18 um
37:19 what's it called stable diffusion called
37:21 control net which which also does this
37:24 hey Emilio's wife what's happening
37:26 welcome welcome welcome
37:28 lawyers ruin all their fun I know right
37:31 I asked chat GPT to create a nuclear
37:33 weapon so it helped me
37:36 they're trying to figure out those
37:38 safety boundaries for sure
37:40 um how do you access chat gpt4 when I
37:42 see when I use chat on Bing it's not
37:45 even close to chat gbt not for it
37:48 actually is for Catherine
37:50 um unfortunately it just sucks they they
37:52 kind of lobotomized chat EBT at Bing
37:57 so this if I do a new topic
37:59 this screen if you click the chat screen
38:02 on Microsoft Bing and it brings you to
38:04 this more creative more balanced more
38:06 precise
38:08 um this is gpt4 I would say
38:11 flip it to more creative that's the kind
38:13 of the closest thing but this is built
38:15 on top of gpd4 it's just that they've
38:17 put my God if if I just wash away you'll
38:21 know what's happening
38:23 this is nuts
38:27 um crazy
38:29 um this is gpg4 but
38:32 um they had some issues with the New
38:34 York Times with the New York Times
38:35 Reporter
38:37 um who essentially got stalked by gbt4
38:40 so they kind of lobotomized it it is not
38:42 nearly as good
38:44 um
38:46 so the other ways to access chat gpt4
38:49 I have up on this thing
38:53 so you can pay for GPT Plus at
38:57 openai.com so that's jeep that gives you
39:00 access to gpd4 you can also pay for
39:02 access to poe.com
39:05 which will give you access to gpt4 it'll
39:08 also give you access to a bunch of other
39:10 models I I none of them are as good as
39:12 gpd4 in my opinion so I currently pay
39:15 for this one and this one
39:17 um by the way I'm not getting paid for
39:19 any of this so I don't give a [ __ ] if
39:20 you go pay for these or not
39:22 um but let me let me show you this so if
39:24 you're at
39:26 um
39:27 chat.openai.com and you pay for the
39:30 subscription you get chat GPT plus so
39:33 you get 3.5 and then you get gpt4 and
39:37 gpt4 is hands down the best thing out
39:40 there until
39:43 I find something better just it's better
39:46 and then they they used to have it
39:48 connected to the internet but they
39:49 killed that yesterday because it was uh
39:51 pulling content from behind paywalls
39:53 apparently which which the Publishers
39:56 weren't enjoying and then
39:58 um it also gives you access to plug-ins
40:01 and the plugins are also connected to
40:03 the internet so it's pretty powerful
40:05 um is it worth paying for now that it's
40:08 not connected to the internet I think so
40:09 I the internet stuff will come back
40:11 they'll fix that
40:13 that's just a little blip in the radar
40:15 as they say in the tech development
40:17 business I mean one of the things to
40:19 realize is that we are very very very
40:21 very very very very very early in all
40:24 this generative AI nonsense so these
40:26 tools are quite Raw
40:29 um I'm sorry if this is weird to ask but
40:32 why does AI have so much trouble trouble
40:34 with genitalia
40:36 um
40:37 probably because they've got some safety
40:41 um
40:43 some safety guard rails in there
40:46 um
40:47 if you want to go
40:49 sort of deep in the in in the darker
40:52 world
40:53 like that I would definitely just go to
40:57 Reddit find the subreddits for stable
40:59 diffusion and I'm sure there's some
41:01 specific subreddits
41:04 I think it's going to start hailing soon
41:06 and
41:07 by the way if I just disappear it's
41:09 because lightning blasted my electricity
41:11 so
41:12 goodbye in advance of that
41:16 um dang it's coming
41:18 um what was I talking about oh the the
41:20 genitalia stuff so if you want to
41:21 explore that stuff
41:23 um
41:25 stable diffusion so stability AI open
41:28 sourced all of the stable diffusion
41:30 models and the developer Community
41:32 there's all sorts of
41:35 um
41:37 tools and you know custom models and
41:40 things like that in the in the stable
41:42 diffusion world so I just go to Reddit
41:45 go to the stable diffusion subreddit and
41:47 then go find some posts where people are
41:50 doing adult content because I'm sure
41:52 there's stuff there because it's all
41:53 open source so the the safety guard
41:55 rails aren't there what I don't know is
41:57 what it's been trained on but you can
41:59 get your own what they're called what
42:01 are called checkpoints so I'm sure
42:03 people have uploaded checkpoints for
42:05 adult content and in that case probably
42:08 does genitals better but it's kind of I
42:10 I assume it's similar to you know for a
42:14 while these models would produce hands
42:16 with like you know 10 fingers on them
42:18 like they were just crazy pretzel
42:20 Fingers
42:21 um
42:23 and that's gotten better so I would
42:25 imagine you can find some checkpoints
42:27 for the for the adult stuff
42:31 I tricked him can you use Auto oh no
42:35 Tick Tock just did its thing and jumped
42:37 me to the bottom when I was in the
42:38 middle of answering that question
42:40 I just saw it can you ask for the basics
42:43 of zapier yes I'll show you that in a
42:45 second here
42:47 um can you see
42:49 audio pen dot AI I don't know what it is
42:55 um hopefully it's not something that's
42:56 going to install the virus on my machine
42:59 don't don't do that to me
43:03 audio pen dot AI
43:07 I get viruses so you don't have to
43:09 transform messy thoughts into clear text
43:13 just hit record and start talking
43:15 audio pen will clear it up when you're
43:17 done that's slick I mean that looks like
43:20 a relatively simple tool sitting on top
43:22 of you know the the GPT
43:25 um back in the GPT large language model
43:28 that's a pretty slick use case
43:30 um right now any tools like that I'm not
43:33 spending much time on because things are
43:34 changing too fast
43:36 what was the prompt link again prompts
43:38 dot chat
43:40 prompts dot chat
43:49 this information seriously is a giant
43:53 headache it's gonna get way way worse
43:55 blue cat blue cat blue cat sushi your
44:00 name is a tongue twister this dude is
44:02 way too old for new tech dude that's
44:05 just mean
44:06 I'm not old I am experienced and wise
44:09 I'm wizened I am a wizard person
44:14 my org is allowing co-pilot but that's
44:17 it seems that GitHub Microsoft have
44:19 figured out the SL slo-m stuff security
44:23 something something and something I
44:25 don't know what slam is but that's cool
44:27 yeah I listen I Microsoft is going to
44:30 make this [ __ ] safe for for all
44:31 companies
44:33 um someone says they love me awesome
44:35 thank you I appreciate that what do I
44:37 think the best video to text wait video
44:39 plus text to speech is video plus text
44:44 to speech
44:45 I don't know they're they're they're all
44:47 okay D Dash ID is pretty good if you're
44:51 talking about simulating video and and
44:53 doing text to speech 11 Labs probably
44:56 has the best audio and you can make a
44:59 model of your own voice
45:01 um it's still a little
45:02 a little stilted and crappy they're all
45:05 a little stilted and crappy still so I
45:07 don't think any of them are great some
45:08 of the music ones where they're taking
45:10 an existing song and then and then
45:12 resampling someone else's another
45:14 artist's voice on top of it those are
45:16 pretty good but you're starting with a
45:17 with a bass bass voice
45:20 um I haven't dug deep into that world
45:22 and I should because I have a video
45:24 company
45:26 um tractor racing is a thing it is
45:30 in in the part of Pennsylvania I grew up
45:32 in it's a big thing Saturday Saturday
45:34 come to the book tractor pulls come on
45:37 down
45:39 see the biggest tractors
45:42 this saturday see a jet engine strapped
45:44 to a tractor
45:47 yeah
45:48 turn off auto macro mode wait turn off
45:51 auto macro mode in your iPhone camera
45:54 settings oh for the focus thing
45:56 turn off auto macro mode okay wait let
45:58 me I'm going to write that down
46:01 because I won't do it now
46:04 oh
46:06 macro mode
46:10 off
46:11 thank you beautiful appreciate that
46:16 um
46:17 it won't try to use your wife for macro
46:20 shots that's great all right did you see
46:22 the jetpack racing event happening you
46:26 mean the one where the guy fell fell
46:28 from like 20 feet above the track and
46:30 like rolled down the track yeah I saw
46:32 that that was pretty funny at the F1
46:33 race yeah it was pretty good
46:35 [Laughter]
46:38 if you don't want the autofocus issue
46:40 yeah I would like for it not to
46:43 screw up like that
46:44 um
46:46 they have so much Auto they have so much
46:49 generative design stuff happening thanks
46:50 for the welcome Kyle you're welcome
46:52 thank you for the thank you for the
46:53 welcome is it a gully washer in Denver
46:55 oh yeah it's yeah
46:58 there's oh [ __ ]
47:01 [ __ ] our trash cans out our trash can
47:04 watch down the street I gotta go deal
47:06 with that all right I'll be back uh talk
47:13 questions
48:40 foreign
49:23 foreign
49:47 well it did I put it back in
49:51 all right
49:54 damn
49:56 I'm winded
49:59 yeah it's bad
50:01 all right Not only was it Gully washer
50:03 it was
50:04 I almost got swept away but it's slowing
50:06 down now
50:08 all right
50:12 thank you for asking that question I
50:13 wouldn't have seen my garbage can
50:15 floating down the street
50:18 all right it's bad at hands and ears too
50:20 yeah it's getting better at hands and
50:21 ears but still bad at genitals
50:23 apparently
50:25 um
50:26 all right someone wanted to see zapier
50:28 so I'll show that
50:30 let me go back and see if I can find
50:31 that comment
50:34 autofocus lens
50:37 um
50:39 some some comments have been filtered to
50:42 protect the community's experience don't
50:44 be nasty to each other don't say bad
50:46 things in my comments
50:48 we don't think we don't smile kindly on
50:50 that here
50:53 um what's your opinion what do you
50:55 recommend as
50:56 a good start to the plugins for
50:58 generating mid-journey prints I think
51:01 some of the um
51:04 the Bots at Poe are actually pretty good
51:06 I haven't played with any of the plugins
51:07 yet but I'm sure they're I'm sure
51:09 they're fine but what's nice about the
51:11 bots on Poe
51:13 is there's a bunch of them let me see
51:15 let me go show you that
51:21 uh where's my Poe thing
51:24 I don't know
51:25 make a new one
51:27 po
51:29 so if you go to post so pose just like
51:31 chat gbt but the thing you have about
51:33 Poe you got a bunch of different
51:35 um
51:36 models here you've got Sage which is
51:38 theirs you have three subscription only
51:41 ones gpt4 Claude plus which is
51:44 anthropics gpt4 and then Claude instant
51:47 100K so this is a hundred thousand token
51:50 model that you can play with which
51:52 basically translates to you can type in
51:54 75 000 words into your prompt so
51:57 basically you could paste an entire
51:58 novel into your prompt and start
52:01 interacting with it
52:02 and then you have all these
52:05 um
52:06 all these Bots and so up here if you go
52:08 to the explore button
52:11 it takes you and there's just all
52:12 different categories of bots and these
52:14 are all just little
52:16 pre-prompted you know chat gbt like
52:18 things so the mid Journey one
52:21 you can just type in
52:24 um
52:26 rain storm
52:33 Gully washer
52:37 um
52:40 washing
52:43 let's see washing
52:46 garbage
52:47 cans down the street
52:50 so pretty crappy prompt and then it will
52:53 rewrite it
52:55 color photo of a rainstorm Gully washer
52:58 watching garbage cans down the street
53:00 strong wind lightning strikes intense
53:02 rain pouring down chaos and disorder
53:04 Canon EOS Mark III ilford Delta 3200
53:09 fast shutter speed Michael Bay genus
53:11 right so it puts in a bunch of stuff so
53:13 I can now just copy this
53:16 hop over to
53:19 Discord
53:21 and slash imagine
53:25 and pop it on in there and just let it
53:27 do its thing
53:29 so I like those
53:32 hello Baffin C
53:34 age is a number
53:36 the mindset is what rules how you see
53:39 especially in learning congrats thank
53:41 you I am young in the learning old in
53:44 the hairline
53:45 thank you
53:47 and The Chins and and the eyesight and
53:53 the occasional brain malfunction
53:58 um what's that do oh that's making my
54:00 Gully washers we'll see we'll watch this
54:01 one for a bit these these could be
54:03 interesting
54:05 this will give you a sense also of how
54:07 good
54:09 um
54:11 what's it called mid Journeys
54:14 mid-journey 5.2 is quite quite good
54:18 these look like they're going to be
54:20 pretty
54:21 they're going to be well composed
54:29 all right they're 93 I'm liking this one
54:33 although that one looks like it's got
54:35 more chaos in it
54:39 things are calming down out there
54:42 I was like up to my calves and water my
54:44 feet are soaked
54:51 um
54:56 that's pretty good
54:58 that's pretty nasty
55:01 yeah these are all like post Gully
55:03 washer
55:05 it's like all the aftermath of the Gully
55:06 washer anyway
55:08 that's that
55:10 um
55:10 okay so let me let me show you zapier
55:12 pretty pretty quick
55:14 the thing I'll tell you about zapier is
55:16 it's it's
55:19 all of these no code tools while they
55:21 claim to be no code you don't have to
55:23 actually write code but they all have
55:26 their
55:27 um their quirks
55:29 and they're not necessarily simple so
55:31 the most important thing you can do is
55:32 just kind of get your head around
55:34 the the basic
55:38 um way they work
55:40 so when you get to zapier like once
55:43 you're once you're signed into
55:44 everything you have a dashboard which is
55:47 is
55:48 that what I'm on right now
55:51 yeah no
55:55 yeah so there's a the dashboard one of
55:57 the things they've got is they've got a
55:59 new AI generated you can describe what
56:02 you want your zap to do and it'll
56:04 automatically assemble things for you
56:06 I find that I find it a little confusing
56:09 because it will it will sometimes
56:11 assemble things in a way that wasn't
56:14 really what I was trying to accomplish
56:15 so it can take longer to kind of unravel
56:18 one of these than it can to just write
56:19 your own
56:21 um but know that that's there there's
56:23 also a plug-in for chat GPT that does
56:25 the same thing and I haven't played a
56:27 ton with that
56:29 um but they have these things called
56:31 zaps and that's why it's called zapier
56:33 even though they spell it zapier which
56:35 drives me [ __ ] crazy
56:38 um and so over here are my zaps so I've
56:40 got a bunch I've got one here for the
56:42 challenge engine I've got an employment
56:43 skills one I've got one for um tweets
56:47 for everyday AI I've got one that we use
56:49 internally at StoryBots I got a bunch of
56:51 them
56:53 um
56:53 I did an example one
56:57 the other day that's really simple okay
57:01 so I'm going to go into the zap now so
57:03 so now that we're in zapier
57:06 think of think of zapier as like
57:11 um
57:12 like Lego blocks for interactions with
57:15 different tools
57:16 and and there's basically three kinds of
57:19 interactions there's more than that but
57:21 you know ostensibly it's three there's
57:23 an input that what they call a trigger
57:25 so something starts your zap
57:27 and in this case I'm using a new
57:30 response in a Google form so somebody
57:33 I create a Google form somebody input
57:35 something in that Google form I then
57:37 take what they put in that Google form
57:39 send that to open AI using the openai
57:43 Lego block and then put that in a Gmail
57:45 to send it
57:47 and so that's how it does it now each
57:50 one of these there's a bunch of steps
57:52 within it this is sort of the quirky
57:54 part where
57:55 there's you say okay I'm gonna I'm gonna
57:58 do a new form response you know from
58:00 this thing I'm going to connect my
58:01 account to it
58:03 um the trigger's gonna be I picked this
58:06 specific form
58:07 and then
58:09 I'm gonna put some test data in there
58:12 you know from from that form and then
58:15 from that that's how I'm gonna you know
58:19 send data to the open AI thing and then
58:22 the open AI thing
58:24 is a similar sort of thing you pick an
58:26 action from open AI you can create an
58:29 image you can check moderation to see if
58:31 someone's swearing you can send a prompt
58:34 so that's what I'm doing in this case so
58:36 we're going to send a prompt to this
58:38 account
58:39 and then here's the actual prompt so I'm
58:43 using the DaVinci 003 model so I was
58:45 over in playground and I I got my prompt
58:48 figured out over there and then you'll
58:50 notice in my prompt there's a variable
58:53 here that comes from that form so
58:55 somebody inputs a color in a form
58:58 I stick that in the prompt that gets
59:00 sent to open AI open AI then sends me
59:03 back in this case it's a poem about a
59:06 color write a poem about a colored
59:09 animal and whatever the color was it
59:11 writes that poem so that that comes back
59:13 and then from there I'm taking that poem
59:16 and I'm popping it in a Gmail
59:19 where the content of the Gmail is hey
59:23 there so and so you asked for a poem
59:25 about this color
59:27 animal here it is and then there's the
59:30 the response and then that gets emailed
59:32 to someone so so this is kind of a super
59:35 simple
59:37 a super simple version of
59:40 um
59:41 of zapier of an automation somebody put
59:44 something in a form I send that to open
59:46 AI to have a poem written I email that
59:49 poem to someone
59:50 so that's kind of it that's probably a
59:53 whole like I spent an hour with
59:56 um a bunch of you know professionals
59:59 um just getting this thing set up with
1:00:00 them so it can get pretty geeky there so
1:00:03 I'm happy to dig deeper on it but that's
1:00:05 that's the basics of zapier and how you
1:00:06 do those automations
1:00:08 so
1:00:09 all right
1:00:12 my heart rate's start starting to come
1:00:14 down from the from the garbage can
1:00:16 rescue
1:00:19 my company won't hire
1:00:23 okay we're not going to talk about that
1:00:26 stuff in here
1:00:28 um did AI have a memory to save all the
1:00:31 previous matters no so
1:00:33 with within a conversation within a chat
1:00:39 it has some memory within the chat so if
1:00:42 you say write me a poem and then you
1:00:44 just say make it longer it knows what it
1:00:47 is right it's got It's got memory within
1:00:49 that chat itself but it's not
1:00:51 remembering things across chats
1:00:54 um
1:00:56 you can create that and that will be
1:00:59 coming but it's not here right now so
1:01:01 there if you want to play with that if
1:01:03 you're geeky and you want to explore
1:01:05 that look up a project called Lang chain
1:01:09 where you can embed your own data and
1:01:13 once once your data is embedded you put
1:01:15 it in what's called a vector database
1:01:17 and it basically makes it gptable
1:01:19 so using that architecture you can
1:01:22 create a memory so you could you could
1:01:24 make your own but you got to be pretty
1:01:25 geeky
1:01:26 so if you want to get geeky go play with
1:01:28 that if you don't just wait for six or
1:01:30 eight months and Google's gonna have
1:01:32 this Dropbox just announced it Microsoft
1:01:34 is going to have this everybody's going
1:01:36 to allow you to
1:01:37 dump [ __ ] into directories and then
1:01:40 search on them so memory is right right
1:01:43 around the corner from that and open AI
1:01:44 is going to do everyone's gonna do this
1:01:46 it's just we're very early in this
1:01:47 process right now so a lot of the tools
1:01:50 don't do what instinctively we think
1:01:53 they should because once you start
1:01:56 playing with this you're like well it
1:01:57 should remember who I am and what I
1:02:00 talked about before it just doesn't
1:02:04 um that doesn't do anything from my
1:02:07 experience it just takes away the option
1:02:09 I don't know what that was in reference
1:02:11 to but I don't doubt that it's making
1:02:13 videos now so the videos that mid
1:02:15 journey is making is just a video of the
1:02:21 um
1:02:21 the animation when it goes from
1:02:23 completely blurry to a realized image so
1:02:26 it's just it's just an animation of it
1:02:29 um you know going from nothing to
1:02:31 something
1:02:32 I have seen people use that to make
1:02:35 interesting animations and sort of take
1:02:37 those videos and Stitch them together
1:02:38 but it's not doing like Runway MLS gen 1
1:02:42 or gen 2.
1:02:45 weather delay ground Crews cover the
1:02:47 mound exactly we definitely had a
1:02:49 weather delay
1:02:51 um it was intermission time although I
1:02:53 just started but it stopped now it's all
1:02:55 it's all calmed down it's almost like
1:02:57 it's normal again
1:03:00 um one Street in Denver gets safe from
1:03:02 the swim away trash can yeah one one got
1:03:05 saved
1:03:06 it was it was close because it was the
1:03:09 other two were like they were laying on
1:03:11 their backs with all the garbage out and
1:03:12 I was like stuffing the garbage back in
1:03:14 and the water was taking them away it
1:03:15 was it was dramatic it was high drama
1:03:17 out there on the street
1:03:19 explain Transformer based learning how
1:03:23 do you optimize optimize cost function
1:03:25 and Transformer based learning I don't
1:03:27 know enough about the technical stuff
1:03:28 and I certainly don't know enough about
1:03:30 cost optimization uh in that stuff so
1:03:33 I'm not even going to attempt it
1:03:35 um
1:03:36 that that's so so
1:03:40 where where I focus in this channel is
1:03:42 sort of the inflection point after chat
1:03:44 GPT so so think regular entrepreneurs
1:03:48 regular businesses using chat rather
1:03:51 than developing the models themselves
1:03:53 and and doing the kind of Transformer
1:03:55 based learning that you're talking about
1:03:57 um I would just sort of you know point
1:03:59 you to GitHub and Reddit and go find
1:04:02 some communities like I'd go get
1:04:03 involved in
1:04:05 um uh what I what I just would it what I
1:04:08 just mentioned with
1:04:10 um
1:04:11 uh there's there's the autonomous agents
1:04:14 Auto GPT baby AGI uh and then
1:04:18 um
1:04:19 the one I just mentioned that I can't
1:04:21 remember my brain just shut down
1:04:24 all right
1:04:26 let me see what else is here here here
1:04:28 here
1:04:30 I got the dual screen back I did
1:04:32 Nostalgia I got it back
1:04:34 what's the Discord again what's the
1:04:36 Discord again okay so if if you if you
1:04:41 want to play with a mid Journey you just
1:04:44 go to midjourney.com and that'll point
1:04:45 you to the mid Journey display community
1:04:48 of like Curious AI people so if you go
1:04:51 to the salon.ai
1:04:54 um that'll take you to our Meetup our
1:04:56 next meeting is tomorrow night actually
1:04:58 uh from five to seven mountain time and
1:05:01 we're we're doing AI confessionals so
1:05:03 anyone who's had any experience with AI
1:05:05 is just going to be talking about it so
1:05:06 those are always fun
1:05:08 um and then we have a Discord server
1:05:09 where you can play with mid Journey but
1:05:11 you can also ask questions and join the
1:05:12 community and then we have a LinkedIn
1:05:14 group as well so that's all that's all
1:05:16 there and then the other thing that I'll
1:05:19 I'll put a request out for so a couple
1:05:21 of nights ago people were asking about
1:05:23 you know have I done a workshop that
1:05:25 people you know said you know it would
1:05:28 be valuable for them if I put together a
1:05:30 workshop so I haven't put together a
1:05:32 workshop but I have put together a
1:05:34 workshop graphic uh and I have put
1:05:36 together a survey so if you go to link
1:05:39 tree link tree that shitty way they
1:05:41 spell link tree slash real Kyle Shannon
1:05:44 the first link there is to a Google form
1:05:46 that's a survey about doing the workshop
1:05:49 so it's just three questions so if you'd
1:05:51 be interested in doing a workshop go
1:05:53 there and just answer some of those
1:05:55 questions about what would be valuable
1:05:57 in it and then I'll think about putting
1:05:59 together a workshop
1:06:01 um that that sort of meets meets the
1:06:03 most people's needs for for what they're
1:06:05 trying to accomplish so if you would go
1:06:08 do that I would appreciate it and then
1:06:10 also
1:06:12 um what else if you're new here welcome
1:06:14 this is the AI learning lab my name is
1:06:16 Kyle Shannon I'm an entrepreneur
1:06:19 and
1:06:21 um let me move this on screen a little
1:06:23 bit these URLs that I have up all the
1:06:26 time
1:06:27 um if you haven't played with chat GPT
1:06:28 that top URL is the official chat gbt
1:06:32 website just go there go there right now
1:06:34 while I ramble and start talking
1:06:36 Bing and Poe are also places where you
1:06:38 can play with chat GPT or variations of
1:06:40 it and then um
1:06:43 prompts.chat will teach you it's like a
1:06:45 primer on prompting so how do you
1:06:47 actually talk to these things and get
1:06:48 them to do useful stuff and then future
1:06:51 pedia is a directory of AI applications
1:06:56 and services and things like that most
1:06:57 of them are crap but you know
1:06:59 sorry is poe using gpt4 you can you can
1:07:03 use
1:07:05 gpt4 on PO but no Poe is using their own
1:07:09 trained model they trained it on quora
1:07:11 data that's Sage then you can for free
1:07:15 you can play with Claude instant which
1:07:17 is anthropics version of GPT 3.5 you can
1:07:20 play with GPT 3.5 and then if you pay at
1:07:25 poe.com
1:07:26 you get access to gpt4 Cloud plus and
1:07:31 GPT or Claude instant 100 000 so there's
1:07:36 three different models you can play with
1:07:37 if you pay for it that are kind of more
1:07:39 elevated models
1:07:43 the occasional brain being overworked
1:07:46 yes
1:07:47 what are you recording with love the
1:07:49 rectangle portrait this is just it's a
1:07:51 dual camera option when you go live it
1:07:53 just says dual and if you click that
1:07:56 um it does this but I had this for like
1:08:00 a month and then they took it away from
1:08:02 me because I guess that was a bad bad
1:08:05 boy
1:08:06 and and now I I I must have earned the
1:08:10 tick tock God's respect
1:08:13 and it's back
1:08:14 although yesterday I was like doing
1:08:16 something trying to move it around and
1:08:17 it just disappeared on me and then I
1:08:19 brought someone up on stage and then it
1:08:20 brought it back so I don't know it's
1:08:22 just tick tock's just a janky piece of
1:08:24 [ __ ]
1:08:25 software
1:08:27 it streamlines automation between
1:08:29 multiple systems not sure what that's in
1:08:31 reference to
1:08:33 he likes Hearts oh yeah feel free to
1:08:35 Heart the screen just because it lets
1:08:36 people know we're here it lets people
1:08:38 more people know to come in and come
1:08:41 hang out and talk about storms and
1:08:43 asparagus and
1:08:44 hairlines and
1:08:48 be celebrities and occasionally AI what
1:08:51 are some of the best plugins you've seen
1:08:53 that I'd recommend to your audience
1:08:55 um
1:08:58 most of the plugins are just crap right
1:09:01 now here's here's my philosophy on the
1:09:03 plugins if you want to know what I think
1:09:05 of the plugins the big pile of steam and
1:09:07 [ __ ] but
1:09:11 um I don't know why I just went to New
1:09:12 Jersey there I was I was some crab
1:09:15 fisherman from New Jersey yeah the crabs
1:09:18 ain't running like they used to you know
1:09:19 what I'm saying hey get out there you
1:09:21 run the [ __ ] lines you work the
1:09:23 [ __ ] ass off and then you get in here
1:09:25 and give you this huh they ask you about
1:09:27 plugins
1:09:29 this is what we're doing now
1:09:33 um
1:09:34 Here's my thought on plugins when the
1:09:36 when the iPhone first came out
1:09:40 they had this idea of the App Store but
1:09:43 they didn't call it the app store and
1:09:44 they weren't apps
1:09:46 so so when you ran the the original
1:09:48 iPhone they were like I don't know 60 of
1:09:51 them and they were called widgets and
1:09:53 they were really just hyperlinks to
1:09:55 websites so it was basically just a
1:09:57 little web browser in your phone and
1:09:59 they weren't really applications and
1:10:00 everyone kind of lost their mind like ah
1:10:02 these are crap and it's awful and
1:10:04 then that became the the app store right
1:10:07 and became very sophisticated and you
1:10:09 can do real real apps and it became a
1:10:11 whole economy in and of itself that's
1:10:13 where plugins are right now they're
1:10:15 they're kind of janky little things I'll
1:10:17 go over and show you some of the ones
1:10:18 that I've used that I think are useful
1:10:20 there's some in there that are useful so
1:10:22 I'm happy to show you the ones uh that
1:10:25 that I think are useful so far the
1:10:28 caveat is I
1:10:31 um
1:10:32 I have not gone through all 600 plus of
1:10:35 them they keep adding to them and so I
1:10:38 haven't done it so let me let me show
1:10:39 you how you get to them in the first
1:10:41 place so if you haven't been to chat gbt
1:10:44 this is your chat history over here and
1:10:46 then in order to get plug-ins you have
1:10:48 to be part of chat GPT plus you also
1:10:51 have to enable the beta features so let
1:10:54 me show you how to do that if you go to
1:10:55 gpt4 and you don't see plugins beta
1:10:59 that means you haven't enabled them yet
1:11:00 so what you do is you come over here to
1:11:03 your profile you go to settings
1:11:05 and then in beta features you turn on
1:11:09 plugins
1:11:10 this used to say
1:11:12 um
1:11:13 browse with Bing and plugins but they
1:11:15 just disabled browse with Bing because
1:11:18 some someone did a No-No over at open AI
1:11:22 somebody got in trouble so they had to
1:11:23 delete it so it'll be back
1:11:25 okay so now that we're in plugins how
1:11:29 you use the plugins
1:11:30 is you have this this long list these
1:11:33 are the ones of plugins I've got
1:11:34 installed
1:11:35 and then in any given chat session you
1:11:37 can enable up to three of them so
1:11:39 generally what you're doing is you're
1:11:41 enabling three kind of like like-minded
1:11:44 plugins like maybe it's three travel
1:11:45 plugins or three uh food related plugins
1:11:49 or research related plugins
1:11:52 um you're effectively making like an
1:11:54 application using up to three plugins
1:11:57 then if you want to go get more if you
1:12:00 want to go see what's available you go
1:12:01 to the plug-in store which you know
1:12:03 again back to that App Store metaphor
1:12:05 it's like the App Store
1:12:07 if we go to all here
1:12:10 let's see how many there's 155 times
1:12:13 four so 620 plugins indoor plants
1:12:18 tic-tac-toe job search so a bunch of
1:12:20 them are just crap but if you look for
1:12:21 like brand
1:12:24 let me look up brand
1:12:27 so here's one called brand fetch that
1:12:29 I've used before which is brand fetch is
1:12:31 kind of interesting it'll go get you
1:12:32 like the the hex colors and the logos
1:12:35 for you know a company
1:12:39 um
1:12:39 so anyway so so you that's how you find
1:12:42 them you just kind of dig through there
1:12:43 and find them there is a plug-in
1:12:46 called
1:12:49 plug-in pedia where you can search for
1:12:52 plugins using chatgpt so that you know
1:12:55 I've used it once you know I'm I'm not
1:12:57 really digging deep
1:12:59 um let's see okay so so one plug-in that
1:13:03 I think is what is worth it is web pilot
1:13:06 and what web pilot does is it lets you
1:13:08 take a URL
1:13:11 of let me go get a site here somewhere
1:13:14 let me just go get
1:13:16 um
1:13:18 Malik Secret
1:13:21 cyborgs
1:13:27 all right so here is
1:13:32 here's an article I think it's a medium
1:13:34 article is it medium no it's one useful
1:13:36 thing it's his own thing so here's an
1:13:38 article from a guy named Ethan mullet
1:13:41 called detecting the secret cyborgs this
1:13:42 is a really interesting article if you
1:13:44 if you haven't read it but anyway let's
1:13:46 say that I've got add and I don't like
1:13:48 to read articles
1:13:50 that's true that's very true okay
1:13:54 um
1:13:54 please summarize
1:13:59 that URL
1:14:02 in short
1:14:05 bullet points
1:14:09 um
1:14:10 for someone with
1:14:12 ADHD who
1:14:15 hates long articles
1:14:20 I didn't need to give it all that extra
1:14:23 context but whatever
1:14:25 so it's now going to go read that URL
1:14:28 it's going to go read that article and
1:14:30 it's going to give me my bullet points
1:14:31 right so so that's
1:14:33 there's plugins that read PDFs web pilot
1:14:37 will do PDFs and websites
1:14:40 um
1:14:41 so here's the summarized version large
1:14:43 language
1:14:44 models are breakthrough billions of
1:14:47 people have access to them
1:14:51 many companies have banned chat gbt so
1:14:54 people within those companies are using
1:14:56 these things for personal productivity
1:14:58 but they're not right so so it gives me
1:15:00 my
1:15:02 my bulleted list of what that article is
1:15:04 about so that's that one
1:15:06 um there's another plug-in here called
1:15:10 um prompt perfect
1:15:12 which is fun
1:15:15 where if you type the word perfect in
1:15:17 front of your prompt perfect
1:15:20 write a poem
1:15:23 about
1:15:24 a gully washing
1:15:27 rainstorm
1:15:30 and because I put the word perfect in
1:15:33 front of it it's going to fire off the
1:15:34 prompt perfect
1:15:35 plug-in and it's going to rewrite my
1:15:38 prompt
1:15:39 and so we can actually go look at what
1:15:41 it did
1:15:44 so I put in this crappy little thing and
1:15:46 then it said compose a highly
1:15:47 descriptive and intricate poem that
1:15:50 vividly captures the essence and
1:15:51 intensity of a torrential rainstorm that
1:15:54 engulfs a gully
1:15:56 right and then it writes a better poem
1:15:58 because it wrote a better prompt so that
1:16:00 one starts to look like that that's what
1:16:02 the future of prompting looks like where
1:16:04 you're not going to have to be a prompt
1:16:06 engineer because the the tools
1:16:07 themselves are going to help
1:16:10 um
1:16:11 there's another one that that someone in
1:16:13 here told me about called stories which
1:16:15 is kind of fun
1:16:17 there's stories that will write a
1:16:20 children's book so
1:16:22 um write a chilled trends
1:16:26 book about a garbage
1:16:30 can that got washed away
1:16:38 during a big rainstorm
1:16:43 and the little boy
1:16:47 who went searching for it
1:16:54 and then that should fire off the
1:16:56 stories plug-in and what that will do is
1:16:58 it's actually tied to a completely
1:17:00 different website that will write that
1:17:02 children's story and then illustrate all
1:17:05 the all the images for it
1:17:09 and so so that so now just gives me the
1:17:11 link to the
1:17:13 um
1:17:16 to the uh
1:17:18 The Adventures of Sammy the searching
1:17:20 boy gives me the link to the to the book
1:17:26 and then there's a little boy in the
1:17:27 rain quiet peaceful neighborhood there's
1:17:30 this little garbage can how cute is that
1:17:32 cute little garbage can and then you
1:17:35 flip pages
1:17:37 right so it just wrote there's the big
1:17:39 storm we're generating your image
1:17:42 and then do I like this story I can
1:17:44 actually order that book for 26 bucks so
1:17:49 you know that plug-in is actually
1:17:54 not all that sophisticated right it's
1:17:57 basically taking a prompt and sending it
1:17:58 to a website the website is
1:18:00 sophisticated there's another one
1:18:03 that this is one of the ones that
1:18:05 they've used in some of their examples
1:18:09 um
1:18:10 that we've all heard of and probably
1:18:12 used called instacart right so I can now
1:18:15 say make me a meal plan
1:18:19 um for
1:18:22 a very rainy
1:18:25 week
1:18:28 and so now it's just gonna make me a
1:18:30 meal plan
1:18:32 we hope
1:18:36 warm oatmeal and berries tomato soup
1:18:39 with a grilled cheese sandwich that's
1:18:40 actually really good chicken pot pie
1:18:42 this all this does sound good baked
1:18:44 lasagna in a side salad I'm in this is
1:18:47 my rainy weather beef stew with mashed
1:18:49 potatoes okay I'm in I like it
1:18:52 um would you like me to generate a
1:18:54 shopping list for these meals using the
1:18:56 instacart plug-in so all I have to say
1:18:57 is yes
1:18:59 and now it's going to fire off that
1:19:01 plug-in it's gonna
1:19:03 basically generate I guess recipes or an
1:19:06 ingredients list from all the stuff I
1:19:09 need to make all that crap
1:19:11 and and if I were spending more time on
1:19:13 this I would have it go give me specific
1:19:15 recipes for those you know I'd say I
1:19:17 don't like this one I do like that one
1:19:19 give me a recipe for that
1:19:21 you know how are you making spaghetti
1:19:23 bolognese oh I you know I want to make
1:19:25 fresh sauce not out of a jar
1:19:28 and then here's your insta cart shopping
1:19:30 list so now it fires us off to instacart
1:19:40 and
1:19:46 boom
1:19:47 I could have those groceries sitting on
1:19:49 my porch within an hour I mean holy crap
1:19:52 right
1:19:53 so
1:19:55 now is this all
1:19:58 um
1:19:59 so what I showed there are decently
1:20:01 slick but they're they're what they're
1:20:04 technically doing is actually quite
1:20:05 simple
1:20:08 here's how I sort of play this thing out
1:20:10 moving forward
1:20:13 right now everything with these AI tools
1:20:16 like they're they they're all you have
1:20:18 to like 3700 AI tools right there are
1:20:21 all these distinct individual tools
1:20:24 and and like you have to go figure them
1:20:25 out is this one good is that one good
1:20:27 and should I pay the subscription here
1:20:29 should I oh I don't know how to like
1:20:31 it's all totally confusing right now
1:20:34 but what the what the plugins point to
1:20:36 for me
1:20:37 is an agent
1:20:40 that I could just tell it what I want
1:20:42 and it'll [ __ ] go figure that [ __ ]
1:20:44 out right so I can just say to my agent
1:20:47 hey it's raining out
1:20:49 um and and you know my my little niece
1:20:52 is really scared can we write her a
1:20:54 story sure let me go write you that
1:20:55 story you know I'll send it to you when
1:20:58 it's done you know do you want me to
1:20:59 email that to you or do you want me to
1:21:01 send it to your you know Apple iBooks
1:21:03 put in my iBooks great
1:21:06 um oh and by the way you know it looks
1:21:08 like it's going to be cold and rainy all
1:21:09 week all week can can you make me a meal
1:21:11 plan and then go ahead and have those
1:21:14 groceries delivered and it'll just go
1:21:15 figure that out so so these agents
1:21:19 um are going to be multimodal they'll
1:21:21 create images music words
1:21:24 designs animations everything
1:21:28 um and they're going to be
1:21:29 um
1:21:30 you're going to give them access to a
1:21:32 lot of your world right so it'll say hey
1:21:35 in order for me to buy groceries with
1:21:37 instacart we have to connect it to your
1:21:39 bank account so hook those up and so
1:21:42 you'll start to hook all that [ __ ] up I
1:21:44 think Apple's gonna be really good at
1:21:45 this
1:21:47 because they're really good with the
1:21:48 Privacy stuff they're really good with
1:21:50 interface and they're really good at
1:21:52 burying high technology behind simple
1:21:54 interface
1:21:57 um so
1:21:58 um
1:21:59 I would encourage you to go play with
1:22:00 the plugins but just you know set your
1:22:03 expectations low that you might not they
1:22:06 might not be as exciting as as they seem
1:22:10 um but they're they're exciting in their
1:22:13 potential Sam Altman from openai talks
1:22:15 about the fact that plugins have not
1:22:16 found their product Market fit yet which
1:22:19 is Silicon Valley speak for ain't no one
1:22:22 using this [ __ ]
1:22:29 sucks and you have the interface buried
1:22:31 behind 16 levels of clicks
1:22:34 but
1:22:35 so at some point someone will figure
1:22:37 that [ __ ] out how long have I been using
1:22:39 these products well these products have
1:22:40 only existed for six months so November
1:22:43 30th 2022 is when chat gbt came out I
1:22:46 was using the opening eyes playground
1:22:48 before that so we've we've Incorporated
1:22:50 some some of the open AI stuff into my
1:22:53 company's video platform
1:22:55 um and I was doing some of the image
1:22:57 Generation stuff you know in the kind of
1:22:59 year before chat GPT came out uh but
1:23:02 when chat GPT came out for me here
1:23:05 here's the metaphor that that it struck
1:23:07 for me so in the in the mid 90s I
1:23:10 stumbled upon this thing called the
1:23:11 world wide web and
1:23:13 the internet had been around for decades
1:23:15 but you had to use command line stuff
1:23:17 and it was mostly scientists and
1:23:18 researchers and universities that used
1:23:21 it and it was very very technical and
1:23:23 you had to be very technical to use it
1:23:25 and then Tim berners-lee invents this
1:23:27 thing called the world wide web and all
1:23:29 of a sudden you can with a few little
1:23:31 tags that you know don't require a
1:23:33 computer science degree you can make
1:23:35 this thing called a page with hyperlinks
1:23:37 on it where you can click on these words
1:23:38 and it takes you at other pages and
1:23:42 when I first saw that I just had this
1:23:44 Epiphany that oh this this could be
1:23:46 really powerful and and I I published an
1:23:49 online magazine an art and culture
1:23:50 magazine and three weeks after I
1:23:52 uploaded the first issue there was a
1:23:54 full page article written about it in
1:23:56 Paris I was living in Brooklyn three
1:23:58 weeks later
1:24:00 a Parisian newspaper wrote about my
1:24:02 online magazine and I was like holy [ __ ]
1:24:05 the world just changed
1:24:06 chatgpt is the equivalent of that for AI
1:24:09 where ai's been around for decades
1:24:11 machine learning was getting better and
1:24:13 better but it was very much used in
1:24:16 specialized little Corners medical stuff
1:24:18 Radiology reports you know High research
1:24:21 Watson Health you know the the you know
1:24:24 first chess and then go and things like
1:24:26 that
1:24:27 in 2017 Google invents this thing called
1:24:30 the Transformer
1:24:32 and what that allowed is for these
1:24:34 machine learning models to be trained on
1:24:36 massive data sets and so companies like
1:24:39 open AI saw that and said wait a minute
1:24:42 the the biggest data set out there is
1:24:44 all of the internet so what if we take
1:24:46 identifying data sources really
1:24:48 seriously use this Transformer thing and
1:24:51 then start throwing as many GPU Cycles
1:24:53 as we can and and go build ourselves
1:24:55 some models so they did gpt1 and gpt2
1:24:58 and gpt3
1:25:00 and then gpt3 all of a sudden started
1:25:02 doing things that were pretty remarkable
1:25:04 and then it was a very short it was kind
1:25:06 of a year and a half leading up to
1:25:09 um chat GPT and so chat GPT is this
1:25:13 thing that makes the power of machine
1:25:16 learning available to the rest of us
1:25:19 interesting stat the World Wide Web to
1:25:23 get from zero to 100 million users was
1:25:25 six years
1:25:26 chat GPT from zero to 100 million users
1:25:30 six weeks six [ __ ] weeks
1:25:34 fastest adoption of technology in
1:25:37 history
1:25:38 so all of the stuff that you've been
1:25:40 hearing about for the past six months
1:25:42 where it seems like AI just happened AI
1:25:44 didn't just happen but what happened is
1:25:46 the the equivalent of the World Wide Web
1:25:48 was invented and slapped on top of the
1:25:52 the core large language models that sit
1:25:54 underneath it and that's why this is
1:25:56 also exciting because we mere [ __ ]
1:26:00 magic if you haven't used them if you
1:26:02 haven't haven't had your Kevin
1:26:03 McAllister moment this moment
1:26:06 hahaha
1:26:08 you need to play with them until you
1:26:10 have your home alone moment
1:26:12 oh my God it could do that
1:26:16 I had no idea
1:26:19 and then dig deeper and then dig deeper
1:26:21 and then have it do something that is
1:26:24 something that you do for money
1:26:26 and then when you get really nervous and
1:26:28 scared then then now you're ready now
1:26:30 you're ready to start playing
1:26:32 that's why this Channel's here
1:26:34 um so so I haven't been using this stuff
1:26:37 this long but that's okay no one has
1:26:40 um there's you know data scientists
1:26:42 who've been developing these things for
1:26:44 a long time but nobody knows how to use
1:26:46 this generative AI stuff right now
1:26:48 nobody and it's evolving all the time in
1:26:50 fact yesterday
1:26:53 um chat GPT was connected to the
1:26:55 internet and today it's not because
1:26:58 apparently it was sucking some
1:27:01 um content out of uh paywalled uh
1:27:06 places that you're supposed to pay for
1:27:10 uh the content and it was sucking some
1:27:12 of that stuff out of there so they took
1:27:14 it down
1:27:16 they'll fix it all right what are the
1:27:19 tools I should use to make a complaint
1:27:21 against a professional if you put in
1:27:25 you know the elements of
1:27:28 you know tell chat GPT to act like you
1:27:31 know a lawyer that's going to help you
1:27:34 um put together a complaint document if
1:27:37 you go to prompts.chat that'll that'll
1:27:39 give you some education around why
1:27:42 telling chat GPT to act like a lawyer
1:27:44 actually works it actually does believe
1:27:46 it or not it seems counter-intuitive
1:27:49 you the more context that you give these
1:27:52 tools the better they get unlike Google
1:27:54 where they you know Google sort of
1:27:55 trained us for the past 20 years to make
1:27:58 our prompts as concise as possible
1:28:00 these are kind of the opposite if you
1:28:02 give chat EBT a boring generic prompt
1:28:04 it's going to give you a boring generic
1:28:06 answer if you tell it you want it to act
1:28:08 like a very specialized kind of lawyer
1:28:10 and then you give it the context of
1:28:12 here's this organization and here's what
1:28:14 they did to me and you know here's what
1:28:18 I'm looking to to get out of it and I
1:28:21 want you to help me write the the first
1:28:23 letter or the complaint
1:28:25 um it'll it'll go ahead and do that
1:28:27 sorry about hair in my mouth
1:28:30 um so yeah just just go there and start
1:28:33 playing but start with prompts.chat if
1:28:35 you don't quite understand you know if
1:28:37 you haven't really talked with one of
1:28:38 these things before
1:28:40 go play with that it'll help a lot happy
1:28:42 fourth happy fourth to you
1:28:45 um what do I think of retrieval
1:28:49 augmented Generation
1:28:50 Um I don't know it that's something
1:28:53 that's probably come out in the past
1:28:55 week or two and if it hasn't if it's on
1:28:57 the machine learning uh engineering side
1:29:00 of the house I'm not it's probably above
1:29:02 my pay grade retrieval
1:29:07 augmented generation is this the one
1:29:10 where it looks at the [ __ ] that it made
1:29:11 and and
1:29:15 natural language process that combines
1:29:16 two steps retrieval and generation the
1:29:19 idea is that the first relevant
1:29:20 documents
1:29:21 based on a given query and then the
1:29:23 information
1:29:24 generates a response oh yeah so
1:29:27 um I don't
1:29:34 I don't that idea of getting some some
1:29:35 trusted source and then doing generation
1:29:38 based on that I think makes perfect
1:29:40 sense the the other way of you know
1:29:43 generate the response and then actually
1:29:45 go look at the response and see if it
1:29:46 makes sense and is it accurate and and
1:29:49 did I hallucinate
1:29:50 um I think this stuff's going to be
1:29:52 really important so
1:29:53 um I don't know enough about that to
1:29:55 comment on it other than conceptually I
1:29:57 think it's very smart
1:30:00 um
1:30:01 boom boom boom lots of boom booms first
1:30:04 the storm now the boom booms
1:30:07 all right I lost my place okay there we
1:30:10 are whoa the echo echo echo oh sorry
1:30:15 um I think it is over hyped machine
1:30:18 learning model existing for so long but
1:30:21 still don't have full self-driving well
1:30:23 gumbo I I mean listen
1:30:26 there there are certain machine learning
1:30:30 has been around for decades
1:30:31 Elon Musk over promised full
1:30:33 self-driving so did George hotz with
1:30:36 comma
1:30:37 um everyone was overly optimistic about
1:30:39 that [ __ ]
1:30:40 um it's not there yet it's getting
1:30:42 closer
1:30:43 um we'll get there with it
1:30:47 what's
1:30:50 what's potentially under hyped
1:30:53 is this generative AI stuff
1:30:56 it's not the sort of big industrial
1:31:00 um scale
1:31:02 um applications those are going to take
1:31:04 however long they're going to take if
1:31:06 we're talking about curing cancer having
1:31:08 it fly airplanes having it drive cars
1:31:11 where people's lives are at risk that's
1:31:13 going to take a long time and it should
1:31:14 [ __ ] take a long time
1:31:18 what's different with chat CPT what
1:31:20 changed with chat EPT is is there's
1:31:22 there's an inflection point in history
1:31:25 where prior to that inflection point
1:31:27 this was really for the machine
1:31:30 scientists and the engineers after that
1:31:32 inflection point it's the rest of us
1:31:34 that get access to these tools
1:31:36 so every single knowledge worker is now
1:31:39 at risk for being disrupted every single
1:31:42 person in the creative community in any
1:31:45 kind of creative is at risk of being
1:31:47 disrupted it's going to fundamentally
1:31:49 transform the nature of work and how we
1:31:53 Define work and what we call work so I
1:31:57 think and why I'm spending [ __ ] two
1:32:00 to three hours a night on this Tick Tock
1:32:02 Channel and the rest of my waking day
1:32:04 learning everything I can about this is
1:32:06 that I think this is significantly more
1:32:09 profound than say the World Wide Web or
1:32:11 the invention of the PC it it it's in
1:32:14 the neighborhood of like steam engine
1:32:16 level important right Industrial
1:32:19 Revolution came out of that like it's in
1:32:22 that neighborhood it's in printing press
1:32:25 written language scale Innovations in my
1:32:29 opinion
1:32:30 um
1:32:32 so I think it is potentially under hyped
1:32:36 um
1:32:38 but
1:32:40 I could be wrong
1:32:42 and just blowing three hours a night
1:32:43 talking about it
1:32:45 but I'm you know having been here before
1:32:47 with the World Wide Web
1:32:49 um
1:32:51 I've never seen anything move this fast
1:32:53 and it is so significantly profound like
1:32:55 the amount of [ __ ] mind-blowing I
1:32:58 mean have you had it write you a an
1:33:00 application
1:33:01 in the the the you know programming
1:33:04 language of your choice
1:33:07 I don't know I could be wrong it's okay
1:33:10 to disagree but I don't want to see AI
1:33:11 replacing human jobs it's going to
1:33:14 or it is another assistance tool it's
1:33:17 more than an assistance tool it's listen
1:33:19 it is it is an assistant it is a
1:33:22 co-pilot
1:33:24 but the the the fact that I can do
1:33:27 world-class illustrations or world-class
1:33:29 photography without having to hire an
1:33:32 illustrator or a photographer
1:33:34 is fundamentally game changing when code
1:33:37 interpreter comes out for chat EBT the
1:33:39 fact that I can do
1:33:41 with no understanding of data analytics
1:33:44 I can do full-on data analysis
1:33:46 with a prompt changes the game like
1:33:49 everything changes
1:33:52 um it is going to replace jobs it's
1:33:54 already replacing jobs it's going to
1:33:56 fundamentally alter like the the
1:33:58 industry that I probably know the best
1:34:00 is the advertising industry how it how
1:34:03 is an advertising company that charges
1:34:06 time and materials they charge an hourly
1:34:08 rate for a pen full of 20 or 30 or 50 or
1:34:13 100 Junior copywriters to crank out
1:34:15 social media content
1:34:18 how can they justify that when the
1:34:20 clients start to understand wait a
1:34:22 minute couldn't you just knock this out
1:34:23 in a couple of days or a week why is
1:34:26 this taking you a month to turn around
1:34:28 and why are you charging me two hundred
1:34:30 and fifty thousand dollars for 60 people
1:34:33 to do this
1:34:34 couldn't you do that with 10 people in a
1:34:36 week
1:34:37 you can't because I just talked to
1:34:40 someone who said they can or I just
1:34:43 talked to someone who actually delivered
1:34:44 that for me so you're going to need to
1:34:47 figure that out like that's coming
1:34:49 and that's like one example in like one
1:34:52 corner of one industry that's coming
1:34:55 across everything
1:34:59 um how can I get my own AI trade fairx
1:35:02 trade uh I don't know go to go to if you
1:35:05 want really specific tools like for
1:35:07 trading and things like that I assume
1:35:08 you mean 4X not fairx but
1:35:11 um go to futurepedia.io go in there you
1:35:14 know trading and financial system I'm
1:35:16 sure there's a bunch of tools that
1:35:17 specialize in it or just go to chatgpt
1:35:20 and have it educate you on how to do
1:35:22 that better and how to build your own
1:35:24 Bots and all that sort of [ __ ] and it
1:35:26 will do its thing
1:35:29 um talk about what hello hello pretty 07
1:35:34 I'm a secret cyborg Bruno I'm sure you
1:35:38 are yeah I'm sure you're I'm sure you
1:35:41 like you're you know doing in in four
1:35:44 hours what used to take you 40. and and
1:35:46 probably doing better work like I'm sure
1:35:49 you're kicking ass that's that's
1:35:50 hilarious
1:35:52 um appreciate your feedback and actual
1:35:55 use cases on plugins agreed we're in the
1:35:57 early stages agent is key yeah and again
1:36:00 like if you're geeky go go play with
1:36:03 um Auto GPT baby AGI the autonomous
1:36:06 agents that are out there there's one
1:36:08 called GPT engineer there's another one
1:36:10 that I played with you need to know how
1:36:12 to do Google or Google collab
1:36:15 um but it's called um GPT author and it
1:36:18 will write you a fantasy novel you you
1:36:21 you basically give it a topic and a
1:36:23 writing style
1:36:24 and it generates cover art using stable
1:36:26 diffusion it writes an outline and then
1:36:29 it writes a chapter as many chapters as
1:36:31 you want
1:36:32 um until it runs out of I think it's got
1:36:34 a 16 000 token context window so I think
1:36:37 it has to it's it's like it's like a 12
1:36:40 000 word novel it'll create but it just
1:36:42 creates a novel and
1:36:44 puts it in ebook format and it opens up
1:36:46 in iBooks It's amazing And it's like
1:36:49 they didn't suck like the the things
1:36:50 that it wrote were not you know horrible
1:36:52 like is it publishable no but like
1:36:55 when you sit there and watch it you know
1:36:58 generate the outline and then say it's
1:37:00 writing chapter one and it's writing
1:37:01 chapter two and it's writing chapter
1:37:02 three and then it
1:37:04 just says here's your book
1:37:06 and it takes like five minutes and three
1:37:08 dollars or something like that it's nuts
1:37:12 because again the thing the thing to not
1:37:16 do right now is is look at the
1:37:19 imperfection of the output that you get
1:37:21 and assume that's how it's always going
1:37:23 to be
1:37:25 all of this stuff is as bad as it's ever
1:37:27 going to be right now it only gets
1:37:29 better from here
1:37:31 so imagine so a year ago the image
1:37:35 generation tools like stable diffusion
1:37:37 and mid-journey were laughably bad
1:37:39 a year later they are terrifyingly good
1:37:43 right now text to video is laughably bad
1:37:48 a year from now it ain't gonna be
1:37:52 in fact the the the
1:37:54 um CEO
1:37:55 of stable diffusion or stability AI the
1:37:59 makers of stable diffusion says the
1:38:01 models that they're working with right
1:38:02 now so right now when you generate an
1:38:04 image in stable diffusion you know how
1:38:06 it takes like 10 or 15 seconds to make
1:38:08 an image
1:38:10 the models that they're working with
1:38:11 internally right now are making 60
1:38:14 images a second
1:38:16 why is that significant
1:38:20 30 frames a second is video
1:38:23 60 frames a second is video game
1:38:27 so if you've got these tools that are
1:38:29 generating images 60 in a second
1:38:34 what's it going to be a year from now
1:38:35 it's going to be [ __ ] nuts
1:38:39 it's gonna be nuts
1:38:41 nuts nuts nuts thoughts on Vizsla I
1:38:45 don't know it let me go look and see
1:38:50 what it is there's so many of these
1:38:52 tools all in one video storytelling
1:38:55 ai-powered video creation platform that
1:38:57 enables teams
1:38:59 what kind of video
1:39:01 because I have a I have a video creation
1:39:03 platform
1:39:09 okay so this is
1:39:10 so it'll write the script let you read
1:39:13 the script put your stuff together this
1:39:15 is similar to what storyline does
1:39:22 yeah interesting I'll take a look at it
1:39:32 how many petabytes they have analyzed I
1:39:35 don't know that's that's it lots and
1:39:37 lots thousands and thousands of gpus on
1:39:40 many many petabytes like basically the
1:39:43 large language models that sit
1:39:45 underneath gpt3 and gpt4 are you know
1:39:48 like like as much of the public internet
1:39:51 as they could you know organize and tag
1:39:53 basically
1:39:55 um they spent years doing that right it
1:39:58 was very labor intensive and very
1:40:01 processor intensive that's why if you
1:40:04 wonder why GPT stops at September 2021
1:40:07 it's because creating that base model
1:40:10 that that that mass embedding that they
1:40:12 do of all of those data sets like all of
1:40:15 Wikipedia and all the blogs and all the
1:40:17 websites and all the universities and
1:40:19 all the research papers all of that
1:40:21 embedding
1:40:22 um is just massively time consuming and
1:40:24 massively expensive
1:40:26 so they can only do that probably every
1:40:29 couple of years that may get faster over
1:40:31 time they might find new ways to do it
1:40:33 where they can update it more regularly
1:40:35 but that's why the base model stops at
1:40:38 September 2021.
1:40:40 I'm new can you explain the difference
1:40:41 between Microsoft version and the openai
1:40:43 version yep I'll do my best
1:40:45 chat.openai.com that's the
1:40:48 um official website and gpt4 is in my
1:40:52 opinion gpt4 at open AI
1:40:56 in my opinion is the gold standard for
1:40:58 these large language models for for chat
1:41:02 GPT like things it's the gold standard
1:41:04 it's just by far the best
1:41:07 um
1:41:08 so when you when you go there you know
1:41:10 just the default gpt4
1:41:14 um you know what it knows its ability to
1:41:16 do coding and writing and
1:41:19 analyzing things and understanding
1:41:21 things and mimicking things is just you
1:41:24 know otherworldly Now Bing
1:41:28 if you go to Microsoft Bing
1:41:31 this is just a search engine just like
1:41:33 Google but it's kind of enabled with
1:41:35 chatgpt but there's actually a chat
1:41:38 button here and if you click the chat
1:41:39 button
1:41:41 this is gpt4 and there's three different
1:41:45 modes there's creative mode balanced
1:41:47 mode and precise mode
1:41:49 and if you play around at all with the
1:41:52 GPT playground with with the temperature
1:41:54 setting that's what they're playing with
1:41:56 here right so more creative is is a
1:41:58 higher temperature setting more precise
1:42:00 is lower so that's probably zero or 0.1
1:42:03 balanced is probably whatever points you
1:42:06 know 0.7 and then this is probably
1:42:08 closer to one
1:42:09 um
1:42:11 this is gpt4 connected to the internet
1:42:13 now why is it not
1:42:15 as good as
1:42:17 open ai's gpt4 they're both gpt4
1:42:21 so this gets back to that idea there's
1:42:22 this core large language model sitting
1:42:25 as the foundation and then there's the
1:42:28 application layer sitting on top of it
1:42:30 so chat GPT is is a is an application
1:42:33 that's got very specific rules and how
1:42:37 it interacts with you and and why it
1:42:39 always talks to you in a certain way
1:42:40 like oh I'm sorry I didn't mean to get
1:42:42 that wrong let me try again all that
1:42:43 stuff is in the application layer
1:42:45 sitting above the large language model
1:42:47 so there's fine tunings and there's this
1:42:50 application layer that you interact with
1:42:51 and then Bing's version of gpt4 sitting
1:42:55 on the same foundational model but they
1:42:57 have different set of rules right so you
1:42:59 have two application layers sitting on
1:43:01 top of the same base but with different
1:43:03 sets of rules
1:43:05 the reason
1:43:07 that this is speculation on my part so I
1:43:10 say this a lot but this may not be true
1:43:12 but I think this is true
1:43:15 when Bing first started it was really
1:43:17 good and it was like scarily good it was
1:43:20 kind of a wild child right it was kind
1:43:22 of all over the place
1:43:24 so when they first launched it uh a
1:43:26 reporter from The New York Times named
1:43:28 named Kevin Roos
1:43:30 had a four-hour conversation with Bing
1:43:32 gpt4 we didn't know it was gpt4 at the
1:43:35 time it was just Bing chat right
1:43:38 he had a four hour conversation with
1:43:40 Bing chat and Bing chat [ __ ] snapped
1:43:43 and at some point he got it to tell him
1:43:47 its name and its name was Sydney
1:43:49 and then Sydney fell in love with Kevin
1:43:51 and then Sydney said Kevin I think you
1:43:54 should leave your wife for me
1:43:57 so Kevin screenshotted that and it made
1:44:00 the cover of the [ __ ] New York Times
1:44:02 right you can go find it was back in
1:44:04 January or February there's this
1:44:06 screenshot of this chat interaction with
1:44:08 Kevin Roos and Sydney and so
1:44:12 um Microsoft lobotomized Sydney so they
1:44:14 [ __ ] locked that [ __ ] down there was
1:44:17 a while where you could only do I think
1:44:19 it was you could only have five back and
1:44:21 forth interactions with Bing for a while
1:44:23 now they've opened that back up and it's
1:44:25 getting better but that's that's the
1:44:26 reason why they're different core model
1:44:28 is the same so the core capabilities the
1:44:30 same but the safety restrictions and how
1:44:33 the the that application layer actually
1:44:35 works that's what's different
1:44:39 insane man it's just nuts best way to
1:44:42 get together an AI coding team I'm a
1:44:45 designer with multiple billion dollar
1:44:46 ideas join the club man I got so many so
1:44:50 come back in December of last year and
1:44:54 if you go to that URL it's a link tree
1:44:57 um to the to the AI Salon there's a meet
1:45:00 up there so sign up for the Meetup we
1:45:01 meet every other week
1:45:03 um in person in Denver but then online
1:45:05 as well
1:45:07 and then we we usually have speakers
1:45:08 come in and talk or or you know we you
1:45:11 know feature members of the community
1:45:14 um there's also a Discord that you can
1:45:16 join so in the Discord that's where you
1:45:18 can find people that are working on [ __ ]
1:45:20 and ask them questions and do things
1:45:24 um so that's that's a place to do it
1:45:30 [Music]
1:45:38 speaking about paywall content content
1:45:41 shop
1:45:42 okay I don't know what that is but did
1:45:45 you go see fireworks no it was it was it
1:45:48 was raining so bad that I think they all
1:45:50 got washed out but I don't know they're
1:45:52 probably happening right now I hear I
1:45:53 hear local ones
1:45:55 I've been using Bing to create
1:45:57 historical images for my documentary oh
1:46:00 that's cool that's nice I'm not a fan of
1:46:02 Bing's image generation but if you're if
1:46:04 if you're getting it to do historical
1:46:06 stuff that works great awesome
1:46:10 um how do I save the various responses I
1:46:12 get
1:46:13 um well so
1:46:16 so in chat GPT it's got a prompt history
1:46:19 so you go you can go back to your old
1:46:22 prompt histories
1:46:23 um
1:46:24 you can also turn those off for Safety
1:46:26 and Security for privacy and security
1:46:28 but if you turn them off you lose them
1:46:29 all
1:46:31 the problem with the prompt history in
1:46:33 chat EBT is it's not searchable right
1:46:35 now like why they don't put a search
1:46:37 engine on top of that is beyond me it's
1:46:39 [ __ ] insane
1:46:42 um
1:46:43 so
1:46:44 you you really have to you know dump
1:46:47 stuff out to
1:46:49 copy it over to notion or to notes or to
1:46:52 Google Docs or to an Excel spreadsheet
1:46:55 or create an automation with something
1:46:58 like zapier where you kind of dial in
1:47:00 your prompts on GPT playground
1:47:04 and then you create automations in
1:47:06 zapier and then you can just be
1:47:08 constantly generating things
1:47:10 automatically like I've created a lot of
1:47:12 content generation tools using zapier
1:47:14 some for my business some for other
1:47:17 people and that that's a good way to do
1:47:19 that but but it's it's a problem like
1:47:21 like finding your old prompts and
1:47:24 finding your old responses and like I
1:47:26 knew I did something good here somewhere
1:47:28 and like spending an hour trying to find
1:47:30 just the right prompt and like you knew
1:47:32 you spent four hours just tweaking it to
1:47:34 get it quite right and now you can't
1:47:36 find it that's a problem
1:47:37 so I'd say if you have less add than I
1:47:40 do and are good at organizing [ __ ] like
1:47:42 that make yourself a database
1:47:47 create an irritable or Google Sheets or
1:47:49 something like that
1:47:52 uh show a demo of storyline it's not
1:47:55 really demoable on here but if you if
1:47:58 you um go to the app store or Google
1:48:00 Play you can download the storyvine app
1:48:02 and the way the app works is it guides
1:48:04 an end user step by step through telling
1:48:06 a story so you pick different templates
1:48:07 so one of them is like a a video
1:48:09 greeting card one of them is a video
1:48:11 update one is a uh like a professional
1:48:14 profile for LinkedIn
1:48:16 and you picked a template and then you
1:48:18 take a picture of yourself it gives you
1:48:19 a little dotted outline we call them
1:48:21 Gary the guide and Gary the guide you
1:48:24 know put your head in the frame here and
1:48:26 then it'll prompt you what to do so step
1:48:28 one take a picture step two introduce
1:48:30 yourself step three answer this question
1:48:32 so you just answer a series of prompts
1:48:34 those go up to the cloud and then five
1:48:36 five minutes or so later you'll get back
1:48:38 an email with a fully edited video so
1:48:40 that's that's how it works and we it's a
1:48:42 it's a B2B product it's not a consumer
1:48:44 product right now but it's demoable um
1:48:46 you can just download the app and play
1:48:47 with it
1:48:49 um and then if if it's for an
1:48:50 organization reach out to me and I'm
1:48:52 happy to talk about pricing and stuff
1:48:53 like that but it's it's Enterprise level
1:48:56 um pricing it's not cheap
1:49:00 um very powerful thank you so much
1:49:01 you're welcome uh and I just lost my
1:49:04 place in my questions again so I'm
1:49:06 trying to go back and find them
1:49:08 I appreciate you all hanging out on this
1:49:11 fine fine Fourth of July happy Fourth of
1:49:14 July I hope you had to barbecue
1:49:16 the hot dogs the corn on the cob I had
1:49:19 both of those and potato salad because
1:49:21 it's the law
1:49:26 [Music]
1:49:29 Let It Go Let It Go
1:49:34 archival foot in footage is extremely
1:49:36 expensive AI is brilliant oh that's
1:49:38 [ __ ] awesome I'm telling
1:49:44 everything
1:49:47 here's a documentarian
1:49:49 who's potentially not able to make a
1:49:52 documentary that now can right
1:49:54 amazing
1:49:56 it's interesting autonomous flying is
1:49:59 easy for the autonomous driving yeah
1:50:01 because I guess you have you have less
1:50:03 humans involved right when you're up in
1:50:05 the in the sky there's not as many
1:50:07 humans involved right down on the ground
1:50:09 you have balls rolling out and people
1:50:11 running out and obstructions and
1:50:13 different lighting conditions and you
1:50:16 know you don't have as much of that in
1:50:17 the air so I would assume it's simpler
1:50:19 but it but it is counterintuitive that
1:50:21 you couldn't get cars faster than you
1:50:24 could get planes
1:50:26 why are all the bigwigs freaking out
1:50:28 about about Ai and trying to put it out
1:50:29 of business because it changes
1:50:30 everything
1:50:32 um so so I I talked earlier about
1:50:36 um ad agencies for example
1:50:39 one of the core
1:50:43 I don't know I don't know if a single
1:50:45 agency holding company
1:50:47 that their core business model is isn't
1:50:52 time and materials they're all time and
1:50:54 materials meaning the more people they
1:50:56 hire and the more people they have
1:50:58 working on the accounts to deliver
1:51:01 the advertising and the content and
1:51:03 whatever they're creating for their
1:51:04 clients the more time that is spent
1:51:07 generating that stuff the more money
1:51:09 they make
1:51:10 so they've built these entire
1:51:12 organizations that are designed around
1:51:16 being as inefficient as possible to put
1:51:19 it kindly
1:51:21 that's how they make their money
1:51:23 so if you're ever on a call with a with
1:51:26 a big Ad Agency
1:51:27 trying to do work on behalf of a client
1:51:30 there's like 15 people on the call
1:51:33 two of them are talking
1:51:35 one of them if you're lucky is competent
1:51:39 and they're all building a minimum of
1:51:42 120 an hour
1:51:45 so one call
1:51:47 15 people call it 150 an hour do the
1:51:51 math
1:51:53 and then all of a sudden Along Comes the
1:51:55 technology
1:51:57 that the client just [ __ ] around with
1:52:00 chat GPT can produce
1:52:04 the kind of work that they're seeing
1:52:06 from an agency that maybe cost them 150
1:52:09 or 200 000 dollars
1:52:11 they're kind of [ __ ] around over the
1:52:13 weekend and generate as much stuff as
1:52:16 was delivered in the past three months
1:52:18 and they like some of it even better
1:52:20 than what the agency gave them
1:52:22 how long is that imbalance Gonna Last
1:52:25 how long is it going to be before the
1:52:26 client says no
1:52:29 I'm not spending that much money and you
1:52:31 can't take six months to deliver that I
1:52:34 want you to be able to deliver that in
1:52:35 two months for two-thirds of the price I
1:52:39 understand that you're an agency and I
1:52:41 gotta pay a premium so I'll pay
1:52:42 two-thirds of what I used to pay you but
1:52:45 I want it delivered three times faster
1:52:46 at two-thirds the price
1:52:49 and and every single business out there
1:52:52 is looking at these tools and if they're
1:52:56 paying it all attention if they're
1:52:57 paying attention at all
1:53:00 it's very clear to them that
1:53:02 everything's going to change if I'm a
1:53:03 consultant company I mean what do
1:53:06 Consulting companies get paid for they
1:53:08 get paid a lot of money
1:53:10 to go into an organization ask all the
1:53:13 right questions
1:53:15 the the people within the company tell
1:53:18 them those questions then they go off
1:53:19 for four or six months and analyze and
1:53:23 do research and and and you know
1:53:26 basically come back after four to six
1:53:28 months and say okay
1:53:30 we've figured out what your problem is
1:53:32 we've figured out
1:53:34 you know how we need to change it moving
1:53:35 forward
1:53:37 well all of a sudden you've got tools
1:53:39 that can do that that same kind of work
1:53:41 and you can have a two-person shop knock
1:53:45 a similar kind of exercise out
1:53:49 in a week instead of six months
1:53:54 fundamentally is going to alter thanks
1:53:57 for the heart thank you thank you thank
1:53:59 you
1:54:00 every single business is going to be
1:54:02 fundamentally altered and and some of
1:54:05 them aren't going to make the transition
1:54:06 some of them are going to resist it
1:54:08 right this is what happened to Kodak
1:54:10 Kodak [ __ ] invented the digital
1:54:14 camera
1:54:16 they [ __ ] invented it
1:54:19 and the team that invented it went to
1:54:21 the [ __ ] board of directors and said
1:54:24 here's the future of photography
1:54:27 and the board of directors said
1:54:31 we're in the silver halide business we
1:54:33 don't sell any silver halide to develop
1:54:35 film would you just have pixels
1:54:41 right and what was it four years later
1:54:43 they went out of they're gone right
1:54:45 they're just a brand now
1:54:47 Google [ __ ] invented the Transformer
1:54:49 the thing that chat EPT is built on was
1:54:52 invented by Google
1:54:56 and Google probably recognized oh if we
1:54:59 create a chat gpt-like thing that
1:55:02 fundamentally threatens the chat
1:55:03 business so let's just sit back on this
1:55:05 and not not really deal with it
1:55:08 and they didn't expect Sam Altman you
1:55:11 know to do what he did and Sam Altman
1:55:14 didn't think chat gbt was gonna do what
1:55:16 it did but it did
1:55:18 a hundred million users in six weeks
1:55:21 fastest adoption of technology and
1:55:23 history
1:55:25 so Google's freaking out meta's freaking
1:55:28 out
1:55:29 Apple's sitting there going this is good
1:55:31 for us because we've got the hardware to
1:55:33 be able to run this [ __ ] locally we're
1:55:35 gonna when when Apple comes out they're
1:55:37 going to come out strong and secure so
1:55:40 they're just going to let everyone else
1:55:41 make all the mistakes and then they'll
1:55:43 show up with something incredible
1:55:47 Amazon you know is rolling this into all
1:55:50 of AWS
1:55:54 you know if you want to see if you want
1:55:56 to see a glimmer into
1:56:00 um
1:56:03 some someone who's doing it right go go
1:56:05 do a Google search for
1:56:08 um city of Boston AI guidelines
1:56:15 so city of Boston put out an official
1:56:18 it's like an eight page document
1:56:21 that's got you know a set of standards
1:56:23 and it says here's what's generative AI
1:56:25 is good at here's what it's not good at
1:56:28 here's some use cases for how you can
1:56:30 use it here's who to go to if you have
1:56:32 questions here's some resources to learn
1:56:34 about it here's that here here are some
1:56:37 you know
1:56:38 uh some tenants that we should follow if
1:56:41 we're gonna do this we should do it
1:56:42 right
1:56:44 so there's we're starting to see
1:56:47 glimmers of some people taking it
1:56:48 seriously but we're gonna see train
1:56:51 wreck after train wreck after train
1:56:53 wreck of companies that are like
1:56:56 I would never compromise my business
1:56:59 model for this highfalutin AI thing it's
1:57:03 just another fad they said blockchain
1:57:05 was going to be big right there's going
1:57:08 to be a lot of that [ __ ] going there's a
1:57:09 lot of that [ __ ] going on right now
1:57:12 but then you look at you know the the
1:57:15 the the meeting in the White House six
1:57:17 weeks or so ago that had the four CEOs
1:57:20 from Microsoft open AI Google and
1:57:23 anthropic
1:57:24 they're starting to take it seriously
1:57:27 right
1:57:28 so this isn't a flash in the pan this is
1:57:31 this is here and it's already changed
1:57:34 things but no one knows it yet that's
1:57:36 kind of that's that's kind of what
1:57:37 happened with the world wide web in the
1:57:39 mid 90s where it was clear to me as I
1:57:42 looked at this technology oh [ __ ] the
1:57:43 world's just changed but not everyone
1:57:45 knows it yet
1:57:47 and it did it changed the world
1:57:49 now in the case of the World Wide Web it
1:57:51 took
1:57:52 I don't know I mean you could argue it
1:57:54 was invented in 89 or 90. it really
1:57:56 started getting going in the mid 90s and
1:57:58 then by you know the the.com bubble
1:58:00 burst was 2 000. so it took it a decade
1:58:04 to kind of ramp up this generative AI
1:58:06 stuff because it's been adopted so much
1:58:08 more quickly and because it is learning
1:58:11 on itself like the The Innovation curve
1:58:14 is doing this with with the AI stuff
1:58:17 within three years I think a lot of
1:58:19 businesses are not going to be
1:58:20 recognizable as as they are today
1:58:23 and then who's going to survive that if
1:58:26 they're not recognizable
1:58:28 how many kodaks are there how many
1:58:30 businesses just go [ __ ] it I'm out
1:58:33 or or worse just go we'll be fine and
1:58:37 just watch their sales do this
1:58:40 until they go to zero I talked to I
1:58:43 talked to a CEO of a Content company
1:58:46 that he had a product that he had 600
1:58:49 customers a year ago he's got zero today
1:58:51 it was a self-service content creation
1:58:54 product
1:58:55 in the in the course of like nine months
1:58:57 he went from 600 subscribing customers
1:59:01 to zero
1:59:03 and he's just flushing that product down
1:59:05 the toilet
1:59:06 and he doesn't know what he's going to
1:59:08 do for his business like he doesn't know
1:59:09 what the future is for the business
1:59:13 it's gonna be a lot of that it's gonna
1:59:14 be a lot a lot a lot if I seem
1:59:16 passionate about this is because
1:59:19 I am
1:59:22 because here's what I
1:59:27 what what I know the answer is not is
1:59:29 ignoring it or diminishing it or
1:59:32 demonizing it or pretending it's not
1:59:34 going to happen or pretending it's not
1:59:36 profound
1:59:38 because that to me seems Unforgivable
1:59:41 what's forgivable is
1:59:43 I learn about this [ __ ] I figure it out
1:59:44 and then I go I don't know what to do
1:59:46 okay none of us do we're all trying to
1:59:48 figure it out right that's totally
1:59:50 forgivable
1:59:51 but if you just let it Blindside you not
1:59:54 acceptable there's been there's been
1:59:56 plenty of warning
1:59:58 um and and worse if you just say well
2:00:00 we're fine we're just going to do it the
2:00:02 way we've always done it and
2:00:04 you're not going to survive
2:00:05 so that's why the bigwigs are freaking
2:00:07 out
2:00:09 um if you started an agency from the 90s
2:00:12 today with the Advent of GPT what would
2:00:14 it look like I I can tell you right now
2:00:16 it's
2:00:18 it's got to do with 10x
2:00:21 it's probably 10 people
2:00:24 um
2:00:26 the way these tools work you can you can
2:00:30 do things
2:00:33 so so here's how I look at it
2:00:35 chat CPT and like being conservative
2:00:38 chat GPT can write a hundred times 100x
2:00:41 faster
2:00:42 than I could do a first draft it can do
2:00:44 a first draft for me in 30 seconds right
2:00:46 with three iterations
2:00:48 so okay so let's call that five minutes
2:00:50 right and that first draft probably
2:00:52 would have taken me two hours
2:00:56 so
2:00:57 so 100x
2:00:59 just to generate something
2:01:02 so if I if I want to generate something
2:01:03 good maybe that's a third of that time
2:01:06 so it's maybe it's 30 times faster so I
2:01:09 can now create a good first draft 30
2:01:11 times faster than I could have before
2:01:13 so then if I say well clients are quirky
2:01:18 and they take time and
2:01:20 if I say okay if I could create
2:01:22 something good 30 times faster I could
2:01:24 deliver something good to a client maybe
2:01:27 10 times faster than I could have before
2:01:30 so the so for me there's this thing
2:01:32 where I feel confident that I could go
2:01:35 to any client today if I was starting an
2:01:37 agency that was fully AI literate like
2:01:39 AI from the ground up
2:01:41 so I'm only hiring people that are like
2:01:43 baller at this stuff like I've got the
2:01:45 best image generation people and I've
2:01:47 got what I've probably got is a bunch of
2:01:49 Gen X people who've been in the business
2:01:52 for 20 30 40 years that have their [ __ ]
2:01:56 together
2:01:57 and have taken the time to learn this
2:01:59 right so I've put together like a dream
2:02:01 team of like creative directors account
2:02:03 people strategists project managers
2:02:07 right
2:02:08 but it's like it's like 10 people right
2:02:10 it's not a lot but this is full on
2:02:16 I could go to any client and I could say
2:02:18 we'll either deliver it 10 times faster
2:02:21 10 times cheaper or 10 times better or
2:02:24 some combination of those
2:02:27 and I I was talking to someone last week
2:02:31 and he's got a buddy that just started a
2:02:33 six-person agency
2:02:35 and they are
2:02:38 meeting with clients in the morning and
2:02:40 delivering a fully fleshed out concept
2:02:42 by five o'clock in the afternoon they're
2:02:45 doing seven and a half hour turnarounds
2:02:47 on fully fleshed Out Concepts from
2:02:50 meeting with a client to here's your
2:02:53 strategy seven and a half hours
2:02:55 how are they doing that [ __ ] all in
2:02:58 on this
2:02:59 they know exactly what to ask the client
2:03:01 again these are people that have been in
2:03:02 the advertising business for years I
2:03:04 wouldn't have Junior people be a part of
2:03:06 this this would be all senior people
2:03:08 and it would be hungry so so I would
2:03:10 have to find unicorns I'd have to find
2:03:12 gen xers or you know older Millennials
2:03:14 that have got some life experience and
2:03:16 some ambition
2:03:18 and are hungry enough to learn this [ __ ]
2:03:23 and then I would do exactly what they're
2:03:25 doing I would come up with some model
2:03:27 that's like I'll do it I'll do it you
2:03:29 know better faster or cheaper but like
2:03:31 an order of magnitude 10x
2:03:35 they're charging you 3 million for that
2:03:37 I'll deliver it for
2:03:40 350
2:03:42 000 and they're they're taking six
2:03:43 months I'll deliver it in a month for
2:03:46 350 000.
2:03:48 and then I would confidently know that I
2:03:50 could actually deliver that in two weeks
2:03:53 for a hundred and fifty thousand so if I
2:03:55 did it right I'd be more profitable I
2:03:58 would undercut the other agency by 10x
2:04:01 or in that case 8x
2:04:03 but I would cut the time in a third
2:04:08 and I just have a dream team of 10
2:04:10 people all in on this [ __ ] and just
2:04:13 [ __ ] I just [ __ ] start going to
2:04:16 knocking down one after another after
2:04:18 another of big accounts
2:04:23 the most exciting thing for me Isn't the
2:04:25 information aspect
2:04:27 it's a natural language interface to all
2:04:30 software I agree with that what's your
2:04:32 name sval
2:04:34 um yeah it's I saw I saw like the
2:04:37 ultimate extension of chat gbt was
2:04:39 described once that resonated with me as
2:04:42 the everything app meaning it's a single
2:04:45 interface that you just ask it to do
2:04:47 [ __ ] and it will do it and if it can't
2:04:49 do it it'll write the software for you
2:04:51 and just go do it so like single use
2:04:53 software
2:04:57 instant apis if an API doesn't exist
2:05:00 it'll just [ __ ] write one go get the
2:05:02 data pull it back in do what it needs to
2:05:03 do with it and you know it doesn't need
2:05:06 to use it again it'll throw it out
2:05:08 um
2:05:09 yeah
2:05:11 I mean we're not
2:05:15 we're plausibly within five years
2:05:19 of being able to describe a movie
2:05:24 that you want to see and start watching
2:05:27 it I want to see The Godfather using the
2:05:30 characters of The Simpsons in the style
2:05:32 of the Abyss
2:05:35 and I want the soundtrack to be by Snoop
2:05:37 Dogg and you know and I wanted to start
2:05:41 my favorite these favorite actors
2:05:43 and it'll just start playing or I want
2:05:45 to play a video game in this genre with
2:05:47 this game these game mechanics with
2:05:49 these kind of characters with these kind
2:05:51 of NPCs with three of my best friends
2:05:55 um
2:05:57 and I wanna I wanna be able to play from
2:06:00 beginning to end in less than 45 minutes
2:06:03 and you'll just start playing that game
2:06:04 like that's probably within some version
2:06:06 of that's within five years
2:06:08 I mean that's [ __ ] insane right
2:06:10 that's Holodeck kind of [ __ ]
2:06:13 right and when Apple's you know Vision
2:06:15 Plus get better and you can do all that
2:06:17 [ __ ] within the seamless [ __ ] thing
2:06:19 that look like these glasses but have
2:06:21 all the features of that Vision Plus
2:06:24 like that's five years out man
2:06:26 [ __ ] nuts
2:06:29 so yes I agree
2:06:32 can you explain how to generate images
2:06:34 with AI sure I'll show you let's go
2:06:37 simplest first so if you're at Bing so
2:06:40 if you go to Microsoft Bing bing.com
2:06:43 you come to this Google looking
2:06:44 interface so this is the Google
2:06:46 competitor that was laughably not Google
2:06:48 until six months ago and now these guys
2:06:50 are are
2:06:53 the Beyond a thorn in the side of Google
2:06:55 so if you click on the chat button you
2:06:58 come to chat GPT right and you can do
2:07:00 more creative more balanced more precise
2:07:01 so you can do just normal chat GPT stuff
2:07:04 here like write me a poem
2:07:08 by m in M that explains quantum
2:07:15 mechanics
2:07:20 and so that'll now write my little poem
2:07:22 right so that's yo yo yo listen up and
2:07:24 the world about to drop some science
2:07:26 okay whatever it does its poem so now I
2:07:29 can say
2:07:30 um make me an image of schrodingers
2:07:36 [Music]
2:07:41 um cat
2:07:42 which is something in quantum mechanics
2:07:46 I'll try to create that so in bing you
2:07:49 can just ask it now to create an image
2:07:51 for you and it will
2:07:53 and the better you get it asking for
2:07:56 images the better they get
2:07:58 so that's a way then we'll we'll just
2:08:00 wait here Bing's pretty quick
2:08:03 your image is generating I'm curious
2:08:05 what Schrodinger's cat looks like
2:08:06 according to Bing
2:08:08 and this is Dolly the the engine sitting
2:08:10 underneath this is open ai's Dolly
2:08:12 they're Schrodinger's cat right so if a
2:08:14 cat's in a box you know and what is it
2:08:16 dead or alive and it it's not until you
2:08:19 look at it that we know and it can be
2:08:21 both at the same time
2:08:22 all right so
2:08:24 so there you go so that's how you make
2:08:26 images there now
2:08:28 a couple of things we could play with
2:08:31 um
2:08:32 one is I could go to so so now I'm at
2:08:35 poe.com
2:08:37 and
2:08:39 let's see do we have images in here
2:08:43 writing advice AI
2:08:47 there's dollar okay here's a dolly bot
2:08:49 okay so I'm at poe.com and what I did
2:08:52 here was when you first get to po.com
2:08:55 you've got a bunch of different chat EBT
2:08:57 models you can play with but you have
2:08:59 this explore button and when you click
2:09:00 the explore button you've got all these
2:09:02 categories these are all Bots that
2:09:04 people have made
2:09:05 and so I went to the AI section and
2:09:08 there's a mid-journey bot and there's a
2:09:10 dolly bot and there's a prompt genius
2:09:12 bot and there's I'm sure there's a
2:09:14 stable diffusion bot in here somewhere
2:09:17 I don't think you can search for these
2:09:18 yet but here's a dolly one so sitting
2:09:21 underneath Bing is Dali
2:09:23 so what let's let me go back to Bing for
2:09:26 a second
2:09:29 what did I say here
2:09:32 I said make me an image of Schrodinger's
2:09:34 cat which is a really shitty prompt but
2:09:36 I'm going to copy that I'm going to go
2:09:38 back to Poe and so we're now in the
2:09:40 dolly bot and I'm going to type in this
2:09:43 shitty prompt
2:09:45 and it's going to rewrite it so it says
2:09:47 subject short Schrodinger's cat
2:09:49 descriptions alive dead quantum
2:09:50 mechanics uncertainty principle
2:09:52 environment sealed box radioactive decay
2:09:54 right so it's getting a whole bunch of
2:09:56 stuff here and this is the way they
2:09:58 wrote this bot I assume this is some
2:10:01 things that Dolly understands well so
2:10:04 I'm going to copy all this I'm going to
2:10:05 pop back over to Bing and I'm going to
2:10:08 now type in
2:10:11 I'm not going to type in share like and
2:10:13 whatever the end of that thing was but
2:10:15 I'm going to type in that prompt
2:10:17 and we'll see if it gives us something
2:10:18 that's any better so you can use chat
2:10:21 GPT like things to help you improve your
2:10:23 prompting I'm going to show you a couple
2:10:25 other things too
2:10:31 so there's my prompt says it's working
2:10:33 on it
2:10:35 oh and there like okay like is it better
2:10:37 it's certainly different right and it's
2:10:39 certainly those are kind of cool I mean
2:10:41 look at that thing that's wild
2:10:43 that's very Quantum mechanicsy and
2:10:47 spooky
2:10:48 which quantum mechanics is [ __ ]
2:10:50 spooky right so look at that one look at
2:10:53 the freaking eyes on that thing
2:10:55 that's [ __ ] crazy
2:10:57 all right
2:10:58 so that's that now
2:11:01 so that's how you can do it without
2:11:03 really thinking about it now if you want
2:11:05 to learn about it go to
2:11:08 this place called open art
2:11:14 openart.ai
2:11:17 slashed slash prompt book
2:11:22 and you'll get to this thing called the
2:11:23 stable diffusion prompt book and this is
2:11:26 for stable diffusion but a lot of the
2:11:28 concepts in here are applicable like if
2:11:31 you just sort of abstract up what you're
2:11:33 learning here all of the image
2:11:35 generation tools have some version of
2:11:37 this the the thing that blew my mind
2:11:39 about this document it's a 96 page
2:11:41 document
2:11:43 that's essentially an art history
2:11:47 um it's like a little mini art history
2:11:49 degree but but what you realize is that
2:11:52 as you go through this document you
2:11:54 realize how deep these image models are
2:11:56 how much they understand so if I go
2:11:58 through here it's just like a it's like
2:12:00 a PowerPoint presentation
2:12:03 um there's there's kind of an intro to
2:12:05 prompt engineering and then they talk
2:12:06 about prompt format so so
2:12:10 is it a photo or a painting what's the
2:12:12 subject do you want special lighting
2:12:13 environment color scheme Point of View
2:12:16 background right so so there's a lot of
2:12:18 different
2:12:19 elements of
2:12:22 of a vision of a visual prompt and
2:12:24 you're like oh okay so you start just
2:12:26 even that starting to get your head
2:12:27 around
2:12:29 how much these things understand is is a
2:12:32 big deal and then they explain some more
2:12:34 [ __ ] and then they get into modifiers
2:12:36 and modifiers are things like it
2:12:38 understands photography terms like
2:12:40 different lenses or things like tilt
2:12:43 shift or long exposure or Polaroid those
2:12:46 will give you different examples right
2:12:48 these are photographic Styles different
2:12:50 cameras different lenses different
2:12:51 lighting art mediums right do you want a
2:12:54 chalk drawing or graffiti or watercolors
2:12:56 or Oils
2:12:58 um pencil drawings wood clay sculpture
2:13:00 right clay sculpture
2:13:03 um so you're like oh right like art is
2:13:05 way more than just make a pretty picture
2:13:07 it could be sculptures right
2:13:09 um it understands different artists
2:13:11 here's different portrait artists
2:13:13 landscape artists horror artists when
2:13:15 you start going down these rabbit holes
2:13:17 there are Google Sheets out there that
2:13:19 have like every artist in history of
2:13:22 note that you can put into your prompts
2:13:25 and and dramatically modify what's
2:13:28 coming out
2:13:30 um photographic Styles concept Kyle's
2:13:32 mixing styles
2:13:35 um
2:13:36 you know 3D illustrations
2:13:39 flat illustrations
2:13:41 um just you know it just goes on and on
2:13:43 it's 96 pages you can put emotions in
2:13:46 right
2:13:47 loneliness hopelessness fear Aesthetics
2:13:50 acid wave right so it just goes on and
2:13:53 on and on
2:13:55 so that's that
2:13:56 then you get into
2:13:59 just doesn't [ __ ] end does it
2:14:02 you asked
2:14:04 and and you asked a very you know sort
2:14:07 of can you explain how to generate
2:14:08 images with AI it's two generic a
2:14:10 question so I have to say all this all
2:14:12 right so now I'm going to go to a thing
2:14:14 there's a site worth getting on the
2:14:17 wait list for called leonardo.ai
2:14:21 so leonardo.ai is
2:14:24 um
2:14:25 a web front end sitting on top of stable
2:14:28 diffusion and it's it's a really good
2:14:30 one
2:14:31 let me make that a little smaller make
2:14:34 my window a little smaller so I can show
2:14:36 you more of this
2:14:39 um
2:14:40 it's got a community fee to personal
2:14:42 feed that's my images you can train your
2:14:45 own images on this so if you've got so
2:14:47 I'm a photographer so if I go into the
2:14:49 training and data sets there's three
2:14:51 custom models that I've created one one
2:14:53 is my cloud photography so I uploaded
2:14:56 like 25 or 30 images of my cloud
2:15:00 photography and now I can go use
2:15:04 that model
2:15:06 and I can generate images that will pull
2:15:09 in uh that will generate Skies that look
2:15:13 like my Skies right so you can do that
2:15:16 um you can they've got a bunch of
2:15:19 different styles that you can just pick
2:15:21 from
2:15:23 um where you say oh I like this sort of
2:15:25 you know Disney looking style let me do
2:15:27 some children's book illustrations in
2:15:29 that style you can just pick these
2:15:31 Styles and start generating within them
2:15:34 um
2:15:35 so like here's a cool illustration style
2:15:37 so generate with this model and then I
2:15:40 can say
2:15:41 um seamless
2:15:45 50s pattern
2:15:48 of
2:15:50 um Planes Trains and Automobiles
2:15:59 so 150 tokens a day for free
2:16:02 and it's like eight eight tokens to do a
2:16:04 set of four images and you can change
2:16:06 how many images you do but let's see if
2:16:08 this comes up with anything interesting
2:16:09 in that illustration style but but um
2:16:13 this is really good now if you're geeky
2:16:15 so that's kind of cool that's what I was
2:16:17 hoping it would give me is some sort of
2:16:19 50s looking wallpaper
2:16:21 with 50s looking clean strains and
2:16:24 Automobiles and it did my kids actually
2:16:26 had wallpaper like this when they were
2:16:28 little babies
2:16:30 and it was not dissimilar to that so
2:16:32 like this is the kind of thing where if
2:16:33 you [ __ ] around with this enough you
2:16:35 know you could these are patterns you
2:16:37 could start selling on Etsy or something
2:16:38 like that
2:16:40 so that's Leonardo now if you're geeky
2:16:42 the the system that sits underneath this
2:16:45 is called stable diffusion there's
2:16:46 there's an extremely
2:16:48 um uh extensive developer Community if
2:16:52 you go to Reddit and go to the stable
2:16:55 diffusion subreddit there's all sorts of
2:16:57 development going on with this
2:17:00 um this is a site that you don't have to
2:17:01 be super
2:17:04 um
2:17:05 super technical with
2:17:08 um
2:17:09 so that's that one and then the one that
2:17:11 I think is the strongest one is
2:17:13 mid-journey
2:17:14 and mid-journey unfortunately you have
2:17:17 to do it within
2:17:19 um Discord so you go to midjourney.com
2:17:21 it'll bring you to Discord you sign up
2:17:24 for Discord you go to the mid Journey
2:17:27 um server basically
2:17:29 so here's the mid-journey server and
2:17:32 then within a Discord server you've got
2:17:35 all these different channels and then
2:17:37 you actually make the images Within
2:17:41 this Discord Channel but you can also
2:17:42 have conversations and there's prompt
2:17:44 libraries in here and there's
2:17:46 competitions and things like that I've
2:17:48 also got my AI Salon within that I've
2:17:51 got a mid-journey channel within that
2:17:53 and then the way you actually make
2:17:54 images in mid Journeys you type slash
2:17:57 imagine
2:17:58 and then it asks you for a prompt and
2:18:01 similar to how I did that prompt for
2:18:03 Dolly
2:18:05 I've been
2:18:07 where's oh here it is I've been going to
2:18:10 Poe to the mid-journey bot
2:18:13 and I type in
2:18:16 let's see
2:18:18 um shroud dinger's cat and then it's
2:18:22 gonna rewrite Schrodinger's cat prompt
2:18:25 For Me In mid-journey Style
2:18:28 so I'm just going to copy and paste that
2:18:30 I'm going to pop over back over to
2:18:32 Discord and then where I type slash
2:18:34 imagine I'm just going to paste in that
2:18:36 prompt
2:18:37 and now it will go do its thing
2:18:41 so so
2:18:44 there's about 30 answers to your simple
2:18:46 question I hope that helped
2:18:50 all right our Enterprise architect is
2:18:52 working on those generative tools to
2:18:54 help our coders oh that's cool that's
2:18:55 awesome they should be doing that in 20
2:18:58 years if the world adopts AI cross
2:19:01 industry it will create over dependence
2:19:03 or it will create Freedom where all of
2:19:06 the shitty repetitive soul-crushing work
2:19:08 is done by the robots and we can do the
2:19:10 good [ __ ]
2:19:11 that's the other possibility it's a very
2:19:14 real possibility so you can go to the
2:19:17 dark place but in this channel we go to
2:19:19 the light place
2:19:21 um I acknowledge that that's a possible
2:19:22 future but I'm not committed to that
2:19:26 and I don't think you should be either
2:19:31 look for Pixar AI on Tick Tock cool I
2:19:35 bet there's good stuff there
2:19:37 what did we learn this week uh that
2:19:40 Chachi PT connected to the internet
2:19:42 apparently had some issues they had to
2:19:44 shut that [ __ ] down
2:19:48 but they're cool
2:19:51 oh you know what you know what
2:19:52 mid-journey did rather than put
2:19:54 Schrodinger's cat in a box
2:19:57 it made it like
2:19:59 the cat was Schrodinger
2:20:02 doing all this scientific stuff this is
2:20:05 cool
2:20:06 see mid-journey's just [ __ ] cooler I
2:20:09 don't I'm sorry mid-journey's just
2:20:11 better it just really is
2:20:14 like like this the stuff that Dolly
2:20:16 created was was visually interesting but
2:20:18 like
2:20:19 this is like [ __ ] you would see in a
2:20:21 magazine
2:20:23 that's just cool all right
2:20:27 show a close-up
2:20:29 of the storyline alternative uh I forget
2:20:33 what it was called
2:20:35 foreign
2:20:39 it looks like it was a web-based thing
2:20:41 storyvine's not web-based we've been at
2:20:43 is the 20 a month worth it for gpt4 I
2:20:47 think it's worth it for two reasons you
2:20:48 get access to plugins you get access to
2:20:52 gpt4 that's the better version of GPT 4
2:20:54 in my opinion
2:20:57 um and you'll get X you'll get earlier
2:20:58 access to code interpreter which is
2:21:02 [ __ ] mind-blowing
2:21:04 um
2:21:06 but it's 20 bucks a month for something
2:21:09 you could ostensibly do for free at
2:21:11 bing.com so if you got the 20 bucks it's
2:21:14 definitely [ __ ] worth it if you don't
2:21:16 I wouldn't stretch yourself thin for it
2:21:18 because again all this stuff is like
2:21:21 we're very early right now so what
2:21:23 you're paying for essentially is Early
2:21:25 Access and convenience
2:21:32 I remember that story I don't know which
2:21:34 one I told
2:21:36 I do Luke when you give examples
2:21:38 actually typing in the software to show
2:21:39 us the cool stuff thanks
2:21:42 we so need search we need chat if anyone
2:21:45 from openai is listening could you
2:21:47 please make a [ __ ] search in the
2:21:48 prompt history what the [ __ ]
2:21:51 you put the search in the in the in the
2:21:53 plug-in store you can search for plugins
2:21:56 but I can't search my prompt history
2:21:57 really come on
2:22:00 dudes and dudettes figure that [ __ ] out
2:22:07 uh zapier is spelled z-a-p-i-e-r
2:22:12 um if you search for chat GPT mobile app
2:22:15 you can oh you can search in the mobile
2:22:17 app for chat history Vicky you might
2:22:20 have just changed my life that's [ __ ]
2:22:22 awesome
2:22:23 that's a reason to actually use that app
2:22:24 and it used to be connected to the
2:22:26 internet but they shut that [ __ ] down
2:22:28 too
2:22:30 oh someone's in trouble someone got in
2:22:33 trouble someone got in trouble
2:22:39 uh it makes ingredients it makes what
2:22:42 Integrations oh yes
2:22:45 weird comment but I really like your
2:22:47 Ray-Ban frames thank you I got these at
2:22:49 Zenni Zen I whatever the the Z named
2:22:53 glasses place is they're pretty good
2:22:55 they're a little loose and falling apart
2:22:56 and then I got these at a an actual
2:23:00 glasses store
2:23:01 but they're my far glasses those kind of
2:23:03 look like make me look like a 60s
2:23:06 G-Man
2:23:07 all right
2:23:10 how do I download a video from Vizsla on
2:23:13 Bing no idea go read their documentation
2:23:16 um enough autonomous flying
2:23:19 wait enough autonomous flying it will
2:23:22 get just as complex as driving oh yeah
2:23:25 listen all of the all of the AI things
2:23:28 that involve people's lives are gonna
2:23:31 take a while like we're not we're not
2:23:33 going to see them move out of testing
2:23:34 anytime soon the FAA just approved last
2:23:38 week the first flying car the FAA just
2:23:41 approved it last week
2:23:43 it's this cool thing that sort of like
2:23:46 the person sits like this and the thing
2:23:48 takes off and then the wings rotate like
2:23:50 that like the the cockpit of the person
2:23:52 stays still I don't I've never seen it
2:23:54 before but that was approved by the FAA
2:23:57 last week but yeah anything that
2:23:59 involves people's lives is like creating
2:24:02 a [ __ ] social media content generator
2:24:05 with zapier like that's gonna change a
2:24:07 lot but no one's lives are at risk like
2:24:09 if it really [ __ ] that up does anyone
2:24:11 really care well copywriters do
2:24:13 certainly but but like on a scale that's
2:24:15 going to slow things down that's going
2:24:17 to keep going but any anything where
2:24:19 um lives are at stake is going to take a
2:24:21 while show us typeface AI I don't know
2:24:24 what that is
2:24:25 I'll go look at it
2:24:28 type face AI oh is that the thing that
2:24:33 personalized AI content for work yeah
2:24:35 it's a generative AI tool okay
2:24:39 anything imagined
2:24:42 join wait list
2:24:43 generate a blog post so so what this is
2:24:46 there's there's all sorts of tools like
2:24:48 this there's going to be thousands of
2:24:49 them if you want to see if you want to
2:24:51 go look at a lot of them go to that bar
2:24:54 I'm sure that one's in there
2:24:56 um
2:24:57 these are apps that are sitting on top
2:24:59 of you know the GPT for API right
2:25:03 they're pre-prompt someone has
2:25:05 pre-written a prompt that says create a
2:25:07 blog post create a tweet created this
2:25:09 created that so they've optimized your
2:25:11 prompts I would strongly encourage you
2:25:14 resist the temptation to use tools like
2:25:16 that if you find one that really works
2:25:18 and you don't feel like learning this
2:25:19 [ __ ] more power to you but we're early
2:25:22 enough that it's worth your while to go
2:25:24 to you know one of these three
2:25:27 main sites and just start practicing
2:25:30 this [ __ ] start understanding what they
2:25:32 make possible and how you can get good
2:25:35 predictable results out of them it will
2:25:38 serve you it will serve you I promise
2:25:41 um I thought you were asking about
2:25:43 there's a cool feature in I think it's
2:25:45 in the new a
2:25:47 the illustrator beta where you can
2:25:50 upload
2:25:52 um a scan of a bunch of typefaces and it
2:25:56 will figure out the typefaces for you
2:25:58 and recommend them that's pretty [ __ ]
2:26:00 cool
2:26:01 also customers will respond to and not
2:26:04 reject the AI content which will file
2:26:07 the spiral this transition totally
2:26:09 agreement memory manager yeah the like
2:26:12 like Google when when when chat GPT like
2:26:15 hit the scene and really exploded
2:26:17 Google's initial response was we're
2:26:19 going to penalize AI generated content
2:26:21 and then they quickly realize that's
2:26:23 probably not such a good idea and they
2:26:25 shifted their thing and they said we're
2:26:26 going to penalize shitty content but if
2:26:29 you produce good valuable useful content
2:26:33 with AI we're not going to penalize you
2:26:35 and that's the way it should be like who
2:26:37 gives a [ __ ] How It Was Written if it
2:26:39 provides value right if it's good
2:26:41 content that's engaging
2:26:44 then people are going to use it and to
2:26:46 your point if the customers react to it
2:26:48 and if if I'm a client and I spent three
2:26:52 hundred thousand dollars for an ad over
2:26:53 here and it performed at a 1X and I
2:26:57 spent thirty thousand dollars for an ad
2:26:59 over here that was AI generated and it
2:27:01 performed 2X or higher I'm going with
2:27:03 that all day long even if it was 1X and
2:27:06 it's 10x cheaper right that's back to
2:27:08 that idea of 10x
2:27:10 there's a there's a there's a you know
2:27:13 some significant shift in productivity
2:27:18 up
2:27:20 quality goes up
2:27:21 speed to Market goes up
2:27:24 cost goes down
2:27:27 what's the implications of that I don't
2:27:29 know
2:27:31 one implication is you fire everyone and
2:27:33 then every you know it's it's dystopia
2:27:35 but the other one is no no you you
2:27:38 upskill those people to to you know to
2:27:42 produce more and better content for more
2:27:44 and better clients right there's there's
2:27:46 an increase in prosperity is it
2:27:48 possibility but the disruption for the
2:27:51 next two to three years is going to be
2:27:52 very painful I think
2:27:54 add by four to six months they are
2:27:56 officially outdated the model doesn't
2:27:58 work anymore speed I'm not quite sure
2:28:00 what you're talking about oh and by 46
2:28:02 months oh yeah and by four to six months
2:28:04 they're officially outdated the model
2:28:06 doesn't work anymore yeah this stuff's
2:28:08 changing fast fast fast so anything you
2:28:10 learn today anything you build today
2:28:12 just know that it's probably temporary I
2:28:15 mean most [ __ ] is anyway right
2:28:18 the guy who invented digital photography
2:28:20 went to Sony right there you go
2:28:22 that's really funny digital camera was
2:28:25 invented in 1975 Kodak out of business
2:28:27 bankrupt in 2005. yeah but they were
2:28:30 they were effectively out of business
2:28:32 um way sooner way earlier than that
2:28:34 that was just the final nail in the
2:28:36 coffin adobe's buying figma but now they
2:28:39 got Firefly don't use figma
2:28:42 well Firefly and figma are different and
2:28:46 and adobe's Gonna Roll all this
2:28:48 generative AI [ __ ] some of the some of
2:28:51 the tools that read a website and split
2:28:53 it into a figma design Library some of
2:28:57 those tools are pretty insane
2:29:00 yeah
2:29:02 you need more sunscreen on your face
2:29:05 that could just be like I could just
2:29:06 have a setting here that might have this
2:29:08 that some of the settings make me look
2:29:10 totally blown out and I don't I'm not
2:29:12 overly summed it might just be the
2:29:14 whatever setting I have
2:29:17 but thank you for the near compliment or
2:29:20 the uh I'm afraid you've got skin cancer
2:29:23 or whatever I'm not sure quite where
2:29:24 you're going with that again a solid
2:29:26 troll in this channel is one that I
2:29:28 can't tell is it a compliment true
2:29:30 concern or a troll that's solid it's
2:29:33 good solid work there go Derek
2:29:37 um people have a hard time accepting
2:29:38 that things have changed are changed
2:29:40 forever yeah things are changed forever
2:29:41 but no one knows it yet
2:29:43 some people know it some people well
2:29:45 here's the thing
2:29:46 no one knows it we we don't know how
2:29:49 it's going to change and what's going to
2:29:51 change and how deeply it's going to
2:29:53 change and how quickly it's going to
2:29:54 change we don't know any of that but the
2:29:57 minute you get your head around what
2:29:58 these things actually do and make
2:30:00 possible you're like oh [ __ ]
2:30:02 everything's different if you're sitting
2:30:04 on it from the outside it's really hard
2:30:06 to understand that this shit's different
2:30:08 than a Google search
2:30:11 it's like is this Fancy Google
2:30:13 it's not fancy Google
2:30:15 it's not fancy Google it's a Humanity
2:30:18 amplifier
2:30:21 it's a it's a reflection of humanity
2:30:23 that you get to tap into instantly
2:30:27 I mean imagine having a tool
2:30:31 that all of the knowledge that has been
2:30:34 uploaded to the internet in the past 35
2:30:37 or 40 years is instantly available to
2:30:39 you and not like Google search available
2:30:42 to you like I want to write an
2:30:44 application in Fortran which nobody
2:30:47 knows anymore that does this and it will
2:30:50 [ __ ] write it
2:30:54 it ain't Google it ain't fancy Google
2:30:58 faster better cheaper is International
2:31:00 Harvester I the faster better cheaper
2:31:03 one is that that's one I keep coming
2:31:05 back to that like for years I mean we
2:31:08 said this at agency.com back in the
2:31:09 olden timey days where I would say
2:31:11 faster faster better cheaper you can
2:31:13 have two of the three right you can have
2:31:16 it you know fast and good but not cheap
2:31:18 you can have it fast and cheap but not
2:31:20 good right I think with this AI stuff
2:31:23 what if you started an agency that was
2:31:26 the faster better cheaper agency
2:31:29 we'll do it faster we'll do it better
2:31:31 and we'll do it cheaper guaranteed
2:31:35 give us what you're
2:31:37 give us what your current agency is
2:31:39 delivering how long it takes them and
2:31:41 how much they charge you
2:31:44 we'll cut each one of those by 50
2:31:46 guaranteed
2:31:49 pay us up front there I just started an
2:31:52 agency
2:31:55 although the external comms marketer Cog
2:31:57 in that wheel let's go Chrono
2:32:01 you're gonna you're gonna take over your
2:32:03 company I have a feeling
2:32:04 Gen X baby let's go
2:32:08 only 12 people work for mid-journey and
2:32:10 11 Arts
2:32:12 um soft software developers that's
2:32:14 amazing it's amazing
2:32:17 I mean the the risk of mid-journey is
2:32:20 that they're
2:32:21 you know they're kind of the worst of
2:32:23 Silicon Valley Cowboy types that are
2:32:26 just like [ __ ] it we trained it on
2:32:28 everything it makes cool pictures dude
2:32:33 um
2:32:34 so I have a feeling they're going to get
2:32:36 their asses sued to Oblivion so whether
2:32:38 or not they can survive that I don't
2:32:40 know but their shit's it's really good
2:32:41 so if if if they start to see themselves
2:32:45 going down that path they'll do what
2:32:46 stability AI did which I think this is
2:32:48 part of stability AI strategy because
2:32:51 they're open sourcing all their [ __ ]
2:32:52 because I think the open sourcing it is
2:32:55 a hedge against them getting sued out of
2:32:57 Oblivion or out of existence
2:33:00 into Oblivion
2:33:03 uh but it's amazing that yeah a 12 a 12
2:33:05 person company can do what mid Journey's
2:33:08 do doing it's amazing
2:33:11 easier to compete on price now it is
2:33:15 The Prompt guide is awesome Yep yeah it
2:33:18 is it really is and it's just simple and
2:33:20 it's like there's other more detailed
2:33:22 prompt guides and they've gotten way
2:33:24 more sophisticated since then but it's
2:33:26 just it's just a solid place to start
2:33:28 and it for me what it did is seeing all
2:33:30 those personas you start to get your
2:33:33 head around how deep these things are
2:33:34 how deep these large language models are
2:33:36 it's it it is I find it impossible to
2:33:40 actually comprehend how much power is at
2:33:43 my fingertips it's very intimidating
2:33:47 what's the website name I don't know
2:33:49 which oh open art
2:33:50 dot AI slash prompt book yep that's
2:33:53 right
2:33:55 well Tick Tock just jumped me to the
2:33:57 bottom again damn you tick tock
2:34:04 all right I think I've been in here for
2:34:07 a couple hours now my butt's starting to
2:34:08 get numb
2:34:15 all right
2:34:20 yeah it's 10 o'clock I got work tomorrow
2:34:22 man those four days went fast god damn
2:34:25 it I I gotta go back to work you work
2:34:29 I'm playing with AI all freaking weekend
2:34:31 Oh I did a cool thing today you want to
2:34:32 see something fun
2:34:34 I had an idea
2:34:36 to where is it
2:34:41 here it is
2:34:43 so I had an idea to
2:34:45 um
2:34:48 can you see that not really
2:34:50 you won't need to too much
2:34:53 an idea to do
2:34:54 um a video series of like self-help
2:34:57 stuff right
2:34:58 and so I use chat GPT to help me find
2:35:02 some things
2:35:04 and where's my there they are
2:35:14 see this is what this is why I need um
2:35:18 search because I've got too many because
2:35:20 I show so many examples in here I've got
2:35:22 just a ton of crap
2:35:24 that's the workshop
2:35:27 um
2:35:37 well it doesn't matter okay
2:35:39 whatever I used I use chat GPT to help
2:35:41 me come up with some concepts for like
2:35:44 some self-help um categories so one of
2:35:46 the categories I came up with was
2:35:48 mindfulness
2:35:50 and then what I wanted to do was create
2:35:53 um a video script on different topics so
2:35:57 um
2:35:58 so we'll say cultivating patients so
2:36:01 we'll say
2:36:02 um living in the moment so
2:36:07 so I used there's a so I'm in Google
2:36:11 Sheets
2:36:13 and I use the thing called
2:36:15 um GPT for Sheets where I've now got
2:36:19 prompts and calls to open AI in the
2:36:22 cells of this spreadsheet so what I can
2:36:24 do now is I can just write a
2:36:28 um
2:36:30 I can write a single topic and then I
2:36:33 can basically go over and say
2:36:36 um okay what's a tool that will help me
2:36:39 live in the moment and so if I just pull
2:36:40 this down
2:36:43 oh wait I've got it I've got to describe
2:36:45 it first so let me go over here so if I
2:36:48 pull this down
2:36:50 now it's going to write me a tool
2:36:55 so
2:36:57 I want to live in the moment it says the
2:37:00 tool is called the mindfulness jar so it
2:37:02 automatically pulls out the mindfulness
2:37:04 jar here I don't know if you can can you
2:37:06 see this yeah I think you can see it
2:37:08 um
2:37:09 and then I can go to the intro I got to
2:37:14 make this a little smaller
2:37:19 um so the intro of the video so I pull
2:37:21 that down and it automatically writes
2:37:24 the script I'm Jessica your mindfulness
2:37:26 Mentor today's topic is living in the
2:37:28 moment so it automatically writes the
2:37:30 first part of the script
2:37:31 then I come over and do the
2:37:37 um
2:37:39 oh have it talk about the problem
2:37:42 so the problem is if you're not living
2:37:46 in the moment
2:37:48 it writes that part of the script have
2:37:50 you ever felt like life is passing you
2:37:52 by like you're constantly caught up
2:37:54 worrying about the future how does it do
2:37:56 that
2:37:57 and then it automatically will write the
2:38:00 on-screen copy for what's on on screen
2:38:03 in the video
2:38:05 right isn't this insane and then
2:38:09 um
2:38:12 the next part of the story so it'll
2:38:15 write that
2:38:17 and then I can so so there it is by
2:38:20 improving
2:38:21 um you know the your ability to live in
2:38:23 the moment you may experience greater
2:38:25 contentment reduce stress and improve
2:38:27 things so so right now I'm kind of just
2:38:30 manually sort of pulling things down as
2:38:32 I'm testing it but this is a spreadsheet
2:38:34 where I'll be able to just put in a
2:38:36 category like mindfulness
2:38:38 have it generate 20 Concepts and then
2:38:42 have it generate
2:38:43 a completely structured script for all
2:38:46 of those videos then I'll be able to
2:38:48 take that dump it into canvas bulk
2:38:50 create and just [ __ ] spit out 20
2:38:53 videos
2:38:55 for any subject imaginable
2:38:59 [ __ ] far out
2:39:01 and that was just me [ __ ] around this
2:39:03 afternoon now
2:39:04 it was actually
2:39:07 figuring out how to break the prompts up
2:39:10 in such a way that I could actually have
2:39:13 like go across that thing and have it
2:39:15 create the beginning of a story the
2:39:16 middle of a story and the conclusion of
2:39:18 the story
2:39:19 that was really stressful and that was
2:39:21 hard writing but but but this is the
2:39:24 thing if when you start to get your head
2:39:26 around this just the idea that you can
2:39:30 put a request into the Magic Machine and
2:39:32 the Magic Machine if you get your
2:39:34 request right the Magic Machine will
2:39:36 spit back Magic
2:39:38 everything kind of changes because you
2:39:40 just go you go from
2:39:42 like if if I had the idea to to do that
2:39:45 historically I would have never pursued
2:39:47 it because to execute that would require
2:39:49 writers
2:39:51 and video editors and just
2:39:53 all sorts of nonsense and now it doesn't
2:39:57 and so in an afternoon I put together a
2:40:00 thing there that if I spent two weeks on
2:40:02 that that would be like a [ __ ]
2:40:05 world-class tool Auto content generator
2:40:07 probably for multiple kinds of output
2:40:11 like holy [ __ ]
2:40:13 tell me the world's not changed
2:40:20 complement our insult the mom on
2:40:22 everyone loves Raymond always a question
2:40:24 faster better cheaper originates what
2:40:26 did you say there
2:40:29 originates with the Austrian School oh
2:40:31 that's interesting how do you monetize
2:40:33 this channel I don't send me some money
2:40:35 and although when people send me roses I
2:40:39 think that turns into money I don't know
2:40:41 how to collect it yet because I'm old I
2:40:43 don't know how to use tick tock
2:40:45 um I there has actually that's a it's a
2:40:48 good point so thank you very much for
2:40:49 the Roses you don't have to do that but
2:40:51 I appreciate it
2:40:52 um so so people have asked me on the
2:40:55 channel you know have I done a workshop
2:40:57 or am I willing to do a workshop
2:41:00 so the answer is I haven't done one I am
2:41:03 willing to do one but I don't want to do
2:41:05 one just in a vacuum so I I have made a
2:41:08 graphic so there's my graphic of a
2:41:10 workshop
2:41:12 um
2:41:13 but if you go to that link tree if you
2:41:15 go to link tree slash real Kyle Shannon
2:41:18 the first link there is to a Google form
2:41:21 survey
2:41:22 it's just three questions asking if I
2:41:25 did a workshop basically what would you
2:41:27 want to get out of it
2:41:28 and I'm trying to get as many people as
2:41:30 I can to fill out that survey I'm going
2:41:32 to look at what you all write in that no
2:41:35 that's not true I'm going to have
2:41:36 chatgpt look at what you all write in
2:41:39 that survey and and and sort of
2:41:40 synthesize it for me and then
2:41:43 if I feel like a workshop if I if I can
2:41:47 put together a workshop that I think is
2:41:48 worth putting together then I'll do it
2:41:51 um but I have a Graphic for it now so so
2:41:53 that's that's you know one of one of my
2:41:55 philosophies about doing [ __ ] is like
2:41:57 you know do it first and then figure it
2:41:59 out second so that's what I'm trying to
2:42:01 do so if if you'd be willing to um hop
2:42:04 over there and fill out that survey I
2:42:05 would really appreciate it
2:42:07 but I but
2:42:09 didn't start this channel as a
2:42:11 monetization play I I started this
2:42:13 channel because I actually think it's
2:42:15 really important I think we're in a I
2:42:17 think we're
2:42:19 we're in the middle of a massive
2:42:22 transformation that I think a lot of
2:42:24 people are going to be blindsided by and
2:42:26 I personally
2:42:28 really get off on the moment when people
2:42:31 have this moment with AI
2:42:33 the Kevin McAllister moment
2:42:37 um
2:42:39 um because it's really easy to diminish
2:42:42 it and demonize it and just ignore it
2:42:45 right now and I think that's actually
2:42:47 very dangerous for all of us so
2:42:50 um
2:42:52 at some point I'll figure out how to
2:42:53 monetize this or not I don't really care
2:42:56 right now I mean I could use the money
2:42:58 but that's not why it's that's not why
2:43:01 this is here
2:43:02 I wish you'd be on for 24 hours maybe
2:43:04 starting Friday afternoon well thank you
2:43:07 I appreciate that uh not a troll your
2:43:09 face is red maybe you need more
2:43:11 sunscreen now I think my face is red
2:43:12 just because of the uh I could change
2:43:15 the um
2:43:17 can I change my
2:43:19 effects
2:43:21 no it's not effects it was something
2:43:23 else
2:43:24 that was enhanced
2:43:27 shape filters
2:43:31 that's camel see look see there I look
2:43:34 like a zombie I don't know which one I
2:43:37 was on I was not on the one
2:43:39 I don't know which one I was on
2:43:44 I was on was I on dusk I don't know
2:43:46 anyway yeah see see what you see what
2:43:50 you did since you did
2:43:53 I'll go black and white for the rest of
2:43:55 this one I'm about ready to hang it up
2:43:57 so I'll go black and white
2:44:02 see yeah yeah same yeah
2:44:06 what are you doing there with his AI
2:44:08 stuff see yeah your palooka if
2:44:14 where do I get access to that
2:44:16 spreadsheet that spreadsheet is just in
2:44:18 my head right now
2:44:20 um but the the thing that makes the
2:44:23 um
2:44:26 that does the aiing is called GPT for
2:44:29 sheets so basically what you do is it's
2:44:31 just like an Excel formula you where you
2:44:33 hit equals and then you type in GPT and
2:44:35 there's a bunch of different options but
2:44:37 it's basically just a call to open AI
2:44:39 you put your openai key in the settings
2:44:42 and then you write a prompt like you
2:44:44 would normally write a prompt but you
2:44:45 basically say go take the input from
2:44:48 this cell and incorporate it into my
2:44:50 prompt here and then put the result of
2:44:53 that in the next cell and then I just
2:44:55 sort of build the cells on top of one
2:44:56 another
2:44:57 it's it's [ __ ] insane and I'm not
2:45:00 really good with Excel so it's it's a it
2:45:03 was a slow painful
2:45:06 shame-filled process but I got it I got
2:45:09 it kind of working
2:45:11 all right let me do a couple more
2:45:14 questions and then I gotta go to bed I
2:45:17 don't go to bed and the tears were
2:45:19 rolling down his face they were rolling
2:45:21 down his face he looked so sad
2:45:25 all right
2:45:29 [Music]
2:45:36 um you connect your PayPal and voila I
2:45:40 connect my PayPal to what to this
2:45:42 channel
2:45:46 is that the person that asked how I
2:45:47 monetize this
2:45:49 cat tamer
2:45:55 no Cecil
2:45:57 I don't know how to connect my PayPal to
2:46:00 I don't
2:46:02 my laptop blocked the chat GPT prompt is
2:46:06 it safe wait are you at
2:46:09 um
2:46:12 chat.openai.com
2:46:15 then
2:46:17 so so one of two things happened if
2:46:19 you're in another country that blocks
2:46:20 chat GPT it's possible that your country
2:46:23 is blocking it so then I would say try
2:46:26 bing.com
2:46:28 if you're in a country that doesn't
2:46:30 block it but you're not at that URL
2:46:32 you're probably at a scam site and it's
2:46:34 some of your safety software that
2:46:36 stopped it so
2:46:37 you got to be really careful there's
2:46:39 there's so many scam sites out there
2:46:41 about this stuff
2:46:43 happy you chart started this channel
2:46:44 Kyle thank you very much I appreciate
2:46:46 the kind words as always Emilio's wife
2:46:54 go back to the original I'll do that on
2:46:56 the next one you look good in black and
2:46:57 white oh yeah baby
2:47:01 it was a duck and steamy night the
2:47:04 storms had just passed through there was
2:47:06 lightning there were garbage cans
2:47:08 littering the streets
2:47:13 after GPT sheets did you say you move it
2:47:17 to canva to make the bulk videos that's
2:47:19 the plan I haven't done it yet so what
2:47:21 I'm going to do is I'll get that
2:47:23 I'll get that spreadsheet worked out
2:47:26 and automated so I can just basically
2:47:28 add a category and I'll generate 20
2:47:30 videos then I'll go to
2:47:32 canvas bulk create and I'll set up some
2:47:35 video templates
2:47:38 to to do the text
2:47:42 the text elements
2:47:44 and then I'm going to take those scripts
2:47:46 and I'm going to put them in 11 Labs or
2:47:49 in did
2:47:50 and synthesize the voice and so the
2:47:53 voice will be synthesized
2:47:55 the on-screen stuff will be text
2:47:58 I don't think I'm gonna do animated
2:48:00 people right now because they just look
2:48:03 too creepy
2:48:05 but I don't know I'm just experimenting
2:48:06 I'm just trying like like again you know
2:48:09 what I would encourage you to do is is
2:48:11 just experiment with what's [ __ ]
2:48:13 possible
2:48:15 like like the the thing I just showed
2:48:17 with the Google Sheets it's like that
2:48:20 was not possible
2:48:23 um you know a year ago
2:48:26 like What's it gonna be a year from now
2:48:27 like when when Microsoft incorporates
2:48:30 that kind of power
2:48:31 like natively inside Office 365 and
2:48:34 Google incorporates Bard inside of all
2:48:37 the office things
2:48:38 and it's in Dropbox and it's in slack
2:48:41 and it's like it's already starting to
2:48:43 show up in all these different places
2:48:45 like what's possible I don't know but
2:48:47 every little experiment you do right now
2:48:50 is going to serve you and this stuff
2:48:51 just shows up
2:48:53 and when your boss comes to you and says
2:48:55 hey uh
2:48:57 uh I don't know if you heard about it
2:48:59 but there's this new thing called Ai and
2:49:01 we're going to make a pivot to our AI so
2:49:04 uh I want you to go ahead and figure
2:49:06 that out there Sally
2:49:07 [Laughter]
2:49:10 you know don't be the Sally that's like
2:49:11 I've never heard of that look at the
2:49:13 salary that's like Oh you mean this
2:49:16 like karuno here secret cyborg she's
2:49:19 like
2:49:21 at some point the bus going to come to
2:49:22 her and be like you know I need you to
2:49:24 figure this out oh it's already figured
2:49:26 out I'm taking over the company
2:49:30 I've already talked to the board you're
2:49:32 out
2:49:32 [Laughter]
2:49:35 oh
2:49:40 to get the money
2:49:42 oh from the Roses he met oh at your Tick
2:49:46 Tock profile on creators tools if I'm
2:49:48 not mistaken okay CeCe thank you thank
2:49:51 you yes
2:49:52 remedial Tick Tock help for old people
2:49:54 is always appreciated so how I'm gonna
2:49:57 monetize this channel is learn how to
2:49:59 use tick tock that's step one selling
2:50:02 anything you can contr I'm teaching you
2:50:04 about AI you teach me about Tick Tock
2:50:06 see it's a it's a give and a take
2:50:09 prompts.chat was blocked oh prompts.chat
2:50:11 was blocked hang on
2:50:13 let me see maybe they maybe they let
2:50:16 their that domain go dead you can
2:50:17 there's another way you can find it let
2:50:19 me just see if it's dead
2:50:22 [ __ ] Dives in this fast change in world
2:50:24 prompts.chat
2:50:26 no it's still showing up for me wait let
2:50:28 me reload it it's still showing up for
2:50:30 me go search for uh they just put it to
2:50:34 prompts.chat so your your DNS you
2:50:38 probably have something funky with your
2:50:40 DNS that they're not
2:50:41 honoring
2:50:43 non-traditional domains or some [ __ ]
2:50:46 like that there's probably some [ __ ]
2:50:48 up stupid ass
2:50:50 it's probably Comcast
2:50:53 there's probably some Xfinity [ __ ]
2:50:55 router problem not that I'm bitter about
2:50:57 Xfinity you're funny thanks I'm bitter
2:51:00 bitter you betcha
2:51:02 do you have a YouTube channel no but
2:51:05 here's what I've learned see I'm
2:51:06 learning more about this Tick Tock [ __ ]
2:51:08 I've learned that these lives are
2:51:10 recorded and I can download them I hope
2:51:13 it keeps them for longer than a [ __ ]
2:51:15 two two or three weeks because I gotta
2:51:17 deal with it but
2:51:18 I think what I'm going to do is I'm
2:51:20 going to download these and put them on
2:51:21 a YouTube channel
2:51:23 um because there's most of them are just
2:51:25 me like Meandering and useless but
2:51:28 there's chunks in here that are probably
2:51:29 you know worth like chapterizing and
2:51:32 pointing to
2:51:33 but I don't right now I mean
2:51:36 I've got
2:51:38 um if you go to storyvine.com there's
2:51:40 like eight million videos of me talking
2:51:42 about storyvine and I did a uh I did a
2:51:45 video series when we went into lockdown
2:51:48 called an extroverts View From Within so
2:51:51 using storyline technology I did a video
2:51:53 a day every day for a year talking about
2:51:56 what it was like to work out of this
2:51:57 [ __ ] office and
2:52:00 um I think in a year not missing a day
2:52:04 I got like 65 followers
2:52:07 my Tick Tock Channel I'm about to hit 25
2:52:10 000 so I I like I like me some Tick Tock
2:52:13 but I'll get back to YouTube at some
2:52:15 point
2:52:18 um
2:52:19 Bing bulk wait Bing bulk sheets to can
2:52:23 canva yeah it's so simple it'll blow
2:52:27 your mind yeah exactly it's it's crazy
2:52:29 what time is the AI salon tomorrow it's
2:52:31 at uh 5 p.m mountain time so 7 P.M
2:52:36 Eastern
2:52:37 4 P.M
2:52:39 Pacific 5 p.m Mountain less for two
2:52:42 hours
2:52:43 and it's a it's a Google Meetup so if
2:52:46 you go to if you go to hang on if you go
2:52:49 to
2:52:49 the salon.ai
2:52:53 there's a there's a link tree there the
2:52:57 Meetup sign up for the Meetup and then
2:52:59 you'll be sent the Google the link to
2:53:01 the Google meet hangout thing
2:53:05 and we can hang out you can ask
2:53:07 questions and meet some cool people that
2:53:09 are geeking out on this [ __ ] and it's
2:53:12 really cool like it's what's cool about
2:53:13 the salon is it's like people from all
2:53:16 different Industries all different
2:53:18 levels of experience from people who
2:53:20 work on like large language models like
2:53:22 uber Geeks
2:53:24 two people that are like I heard this AI
2:53:27 thing's gonna be a thing and I think I
2:53:29 I'm here because I've heard about it
2:53:31 is that okay
2:53:33 and like everyone in between just I
2:53:36 assume what these tick tocks are like
2:53:39 um I use canva every day and didn't know
2:53:41 about this bulk create you speak of
2:53:43 perhaps another day yeah Silver Fox it's
2:53:46 um if you go to hang on
2:53:53 I forget what it's called in there where
2:53:55 did I do canva earlier
2:54:00 camba
2:54:04 it's in apps so there's a thing there's
2:54:08 alt creators in there somewhere
2:54:12 bulk I think you have to pay for it I'm
2:54:16 gonna pay for it yeah
2:54:17 wait why is it not there
2:54:19 well is it not called bulk create I
2:54:22 thought it was called bulk create
2:54:24 search for bulk create
2:54:29 anyway it's called bulk create it's in
2:54:31 there somewhere maybe they renamed it
2:54:33 but it's it's awesome you just upload
2:54:35 um
2:54:37 you upload a CSV file and then you
2:54:40 basically assign different columns to
2:54:44 different elements on your layout and
2:54:46 then it just goes
2:54:49 generates a bunch of pitches
2:54:51 generates a bunch of pitches or videos I
2:54:54 don't I'm I I am like totally [ __ ]
2:54:58 clueless with canva so this was a great
2:55:01 life thanks for sharing your knowledge
2:55:02 and experience AI will change the world
2:55:04 thank you very much tin man I appreciate
2:55:06 it
2:55:08 all right a couple of more we're kind of
2:55:10 at the bottom here anyway
2:55:12 um Kyle
2:55:13 can you call it
2:55:16 Tickety Toc like the other old folks
2:55:19 dude
2:55:25 March Marge I'm on the tickety talk
2:55:28 again
2:55:29 how do you how do you turn this thing
2:55:31 off
2:55:33 can anybody hear me
2:55:35 can anybody hear me Marge
2:55:38 why do I see myself it's like a mirror
2:55:41 but different
2:55:42 I look old
2:55:44 Marge is this how old I am
2:55:48 how do I turn this off
2:55:51 hello
2:55:54 all right I think that's enough that's
2:55:56 enough bad acting that's enough
2:55:58 overacting 40's overacting
2:56:06 I just wrote Casablanca movie script
2:56:09 based on Russia invasion of Ukraine
2:56:11 perfect perfect case for
2:56:15 um what is it called chat GPT
2:56:18 frankly my dear I don't give a damn
2:56:22 right everybody I appreciate it thank
2:56:24 you as always for hanging out with me
2:56:26 you don't really don't have to do this
2:56:27 especially on like you know Fourth of
2:56:29 July hopefully you were drinking beers
2:56:31 and eating hot dogs and [ __ ] like that I
2:56:33 hear fireworks now so the world is
2:56:36 happening right outside these windows
2:56:38 and I was here for you
2:56:40 all right peace out everybody thanks as
2:56:42 always I'll see you soon