aiLL 9/20/23 - Democratizing AI: How Language Models Make AI Accessible

Description

Chapters

Transcript

0:00 foreign
0:18 [Laughter]
0:26 all right welcome
0:30 to the lab
0:33 in our learning lab here we come
0:38 welcome everybody
0:40 all one of you I think there's more than
0:42 that my little counter ain't working
0:48 digital Gods what's happening Rambo
0:50 Rambo
0:52 all right let's go let's Google let's go
0:56 isn't that brighter no yeah that's good
1:05 what is the button for when do you push
1:08 the button why do you push the button do
1:11 we have a button hello from Dublin in
1:14 Ireland fantastic
1:17 top of the morning to you Ronnie
1:19 Ronnie Murray that is a very Irish name
1:21 I uh I'm believing I'm believing you're
1:25 from Dublin I like it
1:27 I like it
1:30 I like an Irish Lass
1:36 Cheryl Banks what's happening hey hey
1:38 hey welcome everybody
1:41 we'll let folks get
1:43 joining in here in a bit
1:47 the beginning is here it's not the end
1:49 is near the robots are not going to kill
1:51 us well we'll see
1:55 but this channel is about
1:57 this vision
2:00 that we are at the beginning of
2:02 something new and exciting
2:04 what do you think
2:07 [Music]
2:10 and we will get talking about uh some of
2:14 this sort of stuff
2:16 what's happening
2:19 the go to Eric Farner how are you doing
2:23 today
2:28 um Eric I'm fine with emails although
2:31 I'm not a hundred percent clear
2:34 because you have a different email name
2:36 than this names so send me an email let
2:39 me know that it's you so I can actually
2:40 confirm that it's you
2:43 um
2:44 Kyle's number one fan oh no
2:48 how am I supposed to critique that name
2:51 I have to give that a 10 out of 10 but
2:54 if I do then it's self-congratulatory
2:55 and I tend not to like that [ __ ]
3:01 commitment
3:06 thank you very much I appreciate it
3:08 playing with cap Wing crazy AI for video
3:11 off the hook
3:13 um I have not played with cap Wing let
3:15 me go look at it quick
3:16 oh I also did a thing here that I'll
3:18 talk about
3:20 which could be kind of fun
3:23 cap Wing I haven't heard of that one but
3:25 did you all see the dolly 3 video that I
3:28 did did you see Dolly three announcement
3:31 from openai holy goodness
3:35 mid Journey better get nervous quick
3:37 because some of the stuff that Google
3:40 showed like four or five months ago was
3:43 style drop
3:45 and then what Dolly three showed today
3:49 holy crap if I were mid-journey I'd be
3:52 like uh
3:54 we better get our [ __ ] together that CEO
3:58 ain't gonna do that though I I think
4:00 that CEO is gonna think his [ __ ] doesn't
4:02 stink for a little too long anyway
4:04 welcome everybody
4:06 um
4:08 if you'd be so kind share the live
4:11 and
4:12 um make more content in less time
4:16 and I see you're you're doing the like
4:19 thing uh just just to get some more
4:21 people in here
4:22 more content in less time modern video
4:24 creation platform
4:29 from first draft to final 10 times
4:31 faster yep that makes sense stay on
4:34 brand all the time
4:36 edit in real time together yeah this
4:39 looks pretty slick
4:42 all right cool what are you finding that
4:44 it does that that you like
4:47 more than what you've been doing I
4:49 assume I assume an auto edit is part of
4:52 it
4:53 and then let's see we'll go talk about
4:55 Dolly three
4:57 because that's another mind blower
5:02 um Sherbert just joined welcome shiver
5:03 Tobias what's happening all right so so
5:06 let's get started here
5:09 I'm going to jump in to if you're new
5:12 here my name is Kyle Shannon
5:14 um this is AI learning lab we learn
5:16 about Labs here
5:17 no that's not true we're learning about
5:20 AI
5:22 everyone wants to know about AI it's all
5:24 the rage it's all the rage they're
5:26 talking about it in the New York Post
5:29 they're talking about it on the Today
5:31 Show
5:32 what is this IA thing anyway what is
5:35 internet
5:36 we're we're in the Katie Turek you know
5:39 that a with the circle around it we're
5:41 in that phase of AI right now
5:45 so so so this channel is to go you know
5:48 that Circle that a with the circle
5:50 around it what is internet anyway
5:53 that's what this channel is all about uh
5:57 so I'm an entrepreneur I'm a co-founder
5:59 and CEO of a company called storyvine uh
6:02 I've been doing these lives now every
6:04 night every flipping night for I don't
6:07 know what's it been three months four
6:10 months something like that I started
6:11 this channel back in December of last
6:13 year
6:14 right after I I was sort of into uh
6:17 stable diffusion image stuff and then
6:19 chat GPT came out and I was like oh oh
6:21 world just changed better get on this
6:24 [ __ ]
6:25 um so so I've been doing this for a
6:27 while and and what we do here is I'll
6:30 answer any questions that you have thank
6:32 you Stevo I appreciate that and and then
6:34 I'll walk you through some things like
6:36 if you're like Kyle where how do I get
6:38 started what do I do
6:40 Kyle it's so confusing what do I do Kyle
6:43 oh Kyle so here's what we're gonna do
6:47 dude
6:51 I also sing between actions when I'm
6:55 like moving [ __ ] around I'm singing I'm
6:57 whistling I don't know what's going on
6:59 here
7:00 um okay
7:01 so
7:03 this stuff
7:06 if you have not played with chat GPT go
7:09 there right now just go there now just
7:11 do it do it go there go there right now
7:14 um that's the official chat GPT website
7:18 it's it's it will if if you haven't
7:21 played with if you're like well it kind
7:22 of just looks like a Google search is it
7:24 fancy Google it's it feels like fancy
7:26 Google when you first start playing with
7:27 it but then you ask it to do some things
7:29 that are more than Google and it just
7:31 does them and it just answers your
7:32 questions and it'll write you a song and
7:34 you're like but it can't be creative and
7:35 then it's creative and then you're like
7:37 but it can't code and then it codes so
7:39 just go there and start playing with it
7:41 um Claude and Bard are two other things
7:44 that are worth playing with Bard just
7:46 got a a pretty major upgrade
7:48 um I can't for some reason seem to play
7:51 with the extensions on it but it's got
7:53 extensions where you can now search
7:55 across your Google drive documents in a
7:58 gpt-like manner it that's like holy crap
8:01 but I can't get it to work right now
8:03 because I don't know something must not
8:04 be enabled in my whatever doesn't matter
8:07 Claude is also really interesting so all
8:09 three of these are worth playing with
8:11 prompts.chat will teach you how to talk
8:13 to these things share bear thank you for
8:15 the gummy I love me a good gummy tasty
8:18 like a gummy I live in Denver though you
8:21 got to be careful with your gummies here
8:23 okay okay
8:26 okay
8:28 um
8:29 okay so that's that that's the that's
8:32 the adult table for these large language
8:34 models here's the Kids Table Bing is
8:36 about to get booted off this [ __ ]
8:38 list because there's safety guard rails
8:40 right now are making that thing actually
8:42 unusable it's built on top of gpt4 it's
8:46 connected to the internet it can make
8:48 images with with Dolly three it's going
8:51 to be able to make really good images
8:53 but their safety guard rails are are so
8:55 egregious right now it's really annoying
8:57 you try to you try to do some basic work
9:00 with it and it just blocks you all the
9:02 time so so that would that one's you
9:04 know what we're gonna do with Bing watch
9:05 this take this Microsoft
9:09 watch this
9:11 you're gonna get dimmed look at that see
9:15 how it's dimmer than the other ones it's
9:17 fading out take that
9:23 ah
9:24 labs.perplexity uh that lets you play
9:27 with llamas meta two models and I think
9:30 Bing is still too bright don't you
9:33 it should be just barely visible yeah
9:35 that's good
9:37 we'll do that
9:40 um Pi Pi dot AI That's that's from a
9:43 company called inflection
9:46 get yourself the PI app search for pi uh
9:50 personal assistant and you'll be able to
9:51 find it I'm pretty sure it's on Android
9:53 I know it's on iOS put it into talk mode
9:57 so there's there's like two modes
9:59 there's one where you thank you for the
10:00 thank you for the Llama Winston whoa
10:03 limston Winston is up to level 20.
10:08 the competition is steep on the level
10:10 and up in here
10:13 will you attain level 33 only time will
10:17 tell
10:18 Winston is a is a 20 degree Master
10:21 Tobias for each level 10 not bad that's
10:24 solid
10:25 um anyway Pi Pi we we nicknamed pie
10:28 chatty GPT in here uh it's chat GPT but
10:31 it's it's designed to have a
10:33 conversation with really interesting and
10:35 Poe lets you play with a bunch of
10:37 different models uh including uh Claude
10:40 and uh open AI models so so that one's
10:44 fun too okay so those are those now
10:46 images if you want to start playing with
10:48 images this is too much now I I feel
10:51 like I've got a you know I just
10:53 screenshot this page
10:56 um mid-journey is really good ideogram
10:59 can do text so what I mean by that is
11:02 what do you mean it can do text like the
11:03 beginning is here that image that image
11:06 that image all these images were done in
11:08 ideogram
11:09 and it can do text now I'm showing all
11:13 the final ones you're like wow it's just
11:14 magic it does it perfectly every time
11:16 now all of these tools are janky pieces
11:19 of [ __ ] that's what we talk about in
11:20 here and ideogram's no exception
11:24 um it you
11:26 you can get something close to what you
11:28 want about
11:29 12 of the time and then if you just keep
11:32 messing with it you'll eventually get
11:34 something that's usable
11:36 um but it's still pretty good Dolly
11:38 Three is coming and it's going to be
11:39 amazing okay
11:41 um what else
11:43 um leonardo.ai is built on top of stable
11:47 diffusion and it's a really good
11:48 interface hey Jen
11:51 there are a bunch of tools like hey Jen
11:53 that do voice synthesis and animated
11:56 faces but hey Jen snuck out a little
11:59 feature
12:01 that kind of hit hit uh
12:05 uh got everyone all excitable
12:07 uh in the past two weeks or so called
12:11 video translate so if you go to
12:13 labs.hagen.com
12:16 you can upload a video in English and
12:19 translate it into up to eight languages
12:23 and here's here's a ping pong kind of
12:26 edit I did of me speaking English and
12:29 then speaking the same little section in
12:31 Italian
12:32 so we've heard from a number of clients
12:34 that modular content is important to
12:36 them getting more efficient with how we
12:38 create content and how we leverage
12:39 content is
12:48 so you'll notice the lips are synced
12:50 the voice is the same the accent's good
12:54 uh Emilio's wife who's in here speaks
12:56 Italian fluently and has vouched that
12:58 indeed I speak fluent Italian I couldn't
13:01 tell you if I did because I don't speak
13:03 fluent Italian but apparently now I do
13:06 so that's hey Jen that's just sitting
13:08 buried in their lab section then you've
13:10 got Adobe Firefly and Photoshop
13:12 generative fill those came out of beta
13:14 about two weeks ago now we have Dali
13:17 coming and we have whatever the hell
13:18 Google's gonna release with Gemini
13:20 assuming there's going to be some image
13:21 thing in it so the image generation
13:24 space is getting as noisy as the chat
13:26 gbt space
13:28 um openart.ai prompt book will teach you
13:31 how remarkably deep these things are
13:33 future PD is just a big pile of tools
13:36 um I'm also I've been invited to do a
13:38 workshop uh with a woman named Rhonda
13:41 Phillips so this is a workshop a
13:43 three-part workshop for creatives so if
13:46 you want to dig deep
13:48 um we're doing
13:49 um kind of General AI overview for
13:52 Workshop one we're doing using AI in
13:55 business so some specific use cases for
13:57 business for workshop two and then we're
13:59 doing Advanced AI things like voice
14:01 synthesis and animation and and video
14:04 and things like that for workshops three
14:06 so if you go to rhondasvoice.com events
14:09 you can see all that info there
14:12 um I'll be teaching that and uh yeah
14:14 it's it's for creatives but you know
14:17 this this stuff is you can apply it to
14:20 pretty much [ __ ] anything so even
14:21 though that's designed for creatives
14:23 whether you're creative or not there's
14:24 probably something there to play with
14:25 anyway
14:30 before we get started will he ever start
14:32 it seems like he never starts all he's
14:35 doing is talking is this just an
14:37 informational Channel why are there so
14:39 many slides I didn't come to tick tock
14:41 to watch a PowerPoint
14:43 um you're not wrong you're not wrong I'm
14:45 I support you in your resistance to this
14:48 being of any value whatsoever but it
14:50 does get better it does get better you
14:52 know how I know that because
14:55 the judgment you see what I'm saying
14:58 you see what I'm saying you see you
15:00 gotta stick around else you won't learn
15:02 the secret for that action
15:04 how do you make money with chat GPT good
15:06 question Kyle you ever going to answer
15:08 that oh I might
15:11 [Laughter]
15:15 I think I was just attacked internally
15:17 all right hold please oh yeah making it
15:20 rain I appreciate that
15:25 okay all right all right oh the salon
15:28 so this is a community I started last
15:31 December and we've got a a Discord
15:35 Channel and we have meetups we meet
15:37 every other week and
15:40 um we're just starting a new thing
15:41 within the salon called guilds we're
15:43 going to have specific functional areas
15:45 like someone's going to lead the
15:47 writer's Guild and someone's gonna lead
15:48 you know image generation and someone's
15:50 going to lead you know beginners and
15:52 things like that
15:53 um so it's getting really good and a lot
15:55 of people from the live a lot of the
15:56 Irregulars so so other thing about these
15:59 lives the people that show up regularly
16:00 are called the Irregulars they name
16:03 themselves I didn't this is not a
16:05 disparaging thing they claimed it I just
16:07 supported it I like it they're odd
16:10 they're very odd
16:12 um they have a channel on the Discord
16:14 we've kind of infiltrated the Discord
16:16 and there's it's a lot of fun there's a
16:18 lot of stuff going on in there
16:19 so we can you know things we talked
16:21 about in here we can share on there and
16:23 things like that
16:24 so join us along go there um storyvine
16:27 is my company that's my tag on Twitter
16:29 and the socials and [ __ ] like that and
16:31 then the other thing just a little
16:32 Public Service Announcement we have we
16:34 we're popular enough that we have a scam
16:36 or scammer
16:38 scammer alert so
16:40 if you subscribe to the channel
16:43 make sure that you're or follow whatever
16:45 it is if you follow the channel make
16:47 sure you're not following AI learning
16:49 lab with 2ls in front of learning that's
16:52 not me he's stolen my content and is
16:56 trying to get people to buy his crypto
16:57 [ __ ] and Tick Tock won't do anything
16:59 about it after oh I don't know 73
17:01 complaints
17:03 um that's me
17:04 um so anyway that's that okay all right
17:07 let's get started so here's how this
17:08 thing works ask any questions you have
17:10 in the comments below I've got mods that
17:12 will sort of float your comments to the
17:14 top um I'm 85 messages behind that's one
17:17 of the Hallmarks of the AI learning lab
17:18 is I get behind on comments and then
17:21 people that are new like why Anthony
17:23 answering my questions I asked the
17:25 question 27 minutes ago why isn't
17:27 answering Mike you gotta hang out man
17:30 you gotta hang out I I do answer all the
17:32 questions well as many as I can Silver
17:35 Fox do you have to get an invite for
17:37 dolly or will it magically appear in GPT
17:39 paid so what they said is that it will
17:43 um
17:44 it will
17:45 it'll go first to chat GPT Plus members
17:48 so if people in here ask me about you
17:51 know is it worth paying for chat GPT
17:53 Plus in my opinion yes because you get
17:55 to play with the new [ __ ] and the new
17:57 [ __ ] that's coming holy crap it looks
18:00 good let's let's go look let's go play
18:02 let's go play shall we it's gonna be so
18:06 good oh Kyle why don't you tell us about
18:08 Dolly three maybe we could do something
18:11 youthful in this here live
18:13 all right
18:14 so if you go to you can go read this for
18:17 your for your own self too if you go to
18:20 um
18:22 openai.com
18:27 d-a-l-l-e-3 doll e thank you for the
18:31 hearty heart
18:32 you you can follow along
18:34 um
18:35 so a couple of things
18:39 if you haven't heard of Dali it's it's
18:41 open ai's image generation tool so if
18:43 you go to bing.com and you type make me
18:46 an image of an avocado it'll do it for
18:49 the most part it's really crappy make
18:51 sure you don't have voice recognition on
18:53 when you tell Bing how shitty its images
18:55 are because I did it the other night and
18:58 it took offense and shut the
18:59 conversation down it was pretty funny
19:02 Bing is insecure so just make sure your
19:05 voice recognition isn't on when you tell
19:07 it how shitty its images are but
19:10 Dolly three
19:11 um is coming right and and a couple of
19:14 things about it it's going to be built
19:15 into chat GPT so so sort of in line
19:18 within chat CBT you're going to be able
19:20 to make images I have a feeling that
19:22 that the fact that you don't have to
19:23 like go to a mid journey and then you
19:26 know go back and forth between chat GPT
19:28 and image generation I have a feeling
19:29 that's going to be a bigger deal than
19:31 than people are getting the other thing
19:33 that this looks like it does is it takes
19:35 very sophisticated prompts and deals
19:37 with them well
19:38 um like the prompt here this is
19:41 um it can do text
19:43 and look at the details in the prompt an
19:46 illustration of an avocado sitting in a
19:48 therapist's chair saying I just feel
19:49 empty inside with a pit sized hole in
19:52 the center right it's a hole it's not a
19:54 pit
19:55 the therapist a spoon he's a spoon
20:00 scribbles notes
20:03 um
20:06 let's see this this one's stunning look
20:08 how long that um
20:10 The Prompt is and look how beautiful
20:12 this image is
20:14 so this is mid-journey level level
20:17 goodness
20:19 um
20:21 her attire resembles a whirlwind of
20:25 marble and Porcelain fragments
20:27 illuminated by a gleam of scattered
20:29 porcelain shards creating a dreamlike
20:32 atmosphere the dancer manages to appear
20:34 fragmented yet maintains a harmonious
20:37 and fluid form
20:40 yeah and if you're if you're sitting
20:44 there thinking but Kyle I don't have a
20:47 PhD in literature and linguistic
20:51 composition it's okay
20:54 because one of the things that they're
20:56 claiming is that um
20:59 GPT the chat this image generation Dolly
21:03 three within chat gbt it will be able to
21:06 prompt itself so you can say make me a
21:08 really good prompt for this basic idea
21:10 and it will write its own prompt and
21:12 then and then it follows them in detail
21:15 um so that's pretty slick
21:17 where's how do I close this there we go
21:23 um this is another one that shows
21:27 um some specifics of of just like
21:30 um how much detail you can put in a
21:32 prompt that is actually captured in the
21:34 image so up here it says sidewalks
21:37 bustling pedestrians enjoying a
21:39 nightlife this says a bustling Street
21:41 and shine of a of a full moon so it
21:43 captured the full moon it's got bustling
21:46 life
21:47 she says on the corner a young woman
21:50 with fiery red hair dressed in a
21:52 signature velvet cloak is hang is
21:55 haggling with the grumpy
21:58 um
21:59 vendor and then the description of the
22:01 vendor is a grumpy vendor tall
22:03 sophisticated man
22:05 wearing a sharp
22:07 uh a sharp something suit
22:11 Sports a noteworthy mustache which he's
22:14 got
22:15 in an animated it's it's animatedly
22:18 covering his steampunk telephone which
22:21 is here which he's talking on
22:23 so
22:25 this is some really sophisticated stuff
22:27 now wait it gets better oh this was this
22:30 was just another one there's Dolly too
22:32 with with like a basketball player
22:34 dunking in in let's see depicted as an
22:38 explosion of a nebula but you're like
22:40 okay that's pretty pretty it kind of
22:42 looks like LeRoy Neiman if you remember
22:43 him from the 70s and 80s but check out
22:47 Dolly three
22:49 that's that's the same prompt in Dolly
22:51 three that's an explosion of a nebula
22:54 right and it's you know he's clearly
22:56 dunking it's not just some abstract it's
22:59 you know he's just [ __ ] laying it
23:02 down
23:03 yeah but how many prompts did it take to
23:05 get that quality I don't know what
23:06 you're talking about for Dolly I have no
23:08 idea for uh ideogram I can tell you
23:13 um you can be cynical about it and and
23:17 and and sit on the sidelines and say
23:19 yeah well they're only showing you the
23:20 best ones they've shown us the best ones
23:23 but if you've worked with mid-journey
23:25 um and tried to get it to do stuff
23:28 consistently and predictably it's almost
23:31 impossible and if you've ever tried to
23:32 get mid-journey to include
23:35 more than a main subject it's it's
23:38 insanely frustrating
23:41 um so it's it's it's looking like it's
23:44 looking like this is going to be quite
23:45 good
23:46 um and the other thing one of the other
23:48 things and this is this is a question I
23:50 feel all the time in in this group and
23:53 elsewhere in my life is if I create a
23:57 character that I like in mid-journey how
24:00 do I keep the character across prompts
24:03 right like if I create a character for a
24:05 children's book I want that character to
24:08 be in all the scenes of the children's
24:09 book I don't want it to be a new
24:11 character each time which is kind of
24:12 what it does now watch this video I'll
24:14 play I'll play enough of it here that we
24:16 see
24:24 wrote its own prompts for a hedgehog
24:28 so
24:30 so what it said is
24:33 my daughter
24:36 my daughter says his name is Larry
24:40 right so it made a thing he picked one
24:42 and then he's typing a prompt about this
24:45 image my daughter says its name is Larry
24:47 can I see more like this and now it
24:51 keeps Larry
24:53 foreign
25:01 can you show me Larry being kind-hearted
25:03 there's Larry being kind-hearted could
25:06 you design some stickers there's Larry
25:08 on some stickers
25:12 Derek wrote the story
25:16 can I see Larry at the end where he's
25:18 happily dreaming
25:21 there he is peacefully dreaming at the
25:23 end like
25:24 if you've ever tried to get to get
25:27 mid-journey or any of these things to
25:30 consistently do that you can sort of do
25:32 it with control net and stable diffusion
25:34 and there's ways you can sort of hack
25:36 around it
25:37 but you can't just say that character's
25:40 Larry and now let's start interacting
25:42 with Larry so
25:44 you know
25:45 will it be as good as it's purported to
25:48 be
25:49 um it's at least gonna be better than
25:51 Dolly too
25:52 will the quality be as good as
25:54 mid-journing
25:56 probably not but I don't know I don't
25:58 know who knows what it's been trained on
26:00 I mean
26:02 they have some resources so so it's very
26:05 possible I mean I I heard I heard you
26:08 know kind of rumors that what openai was
26:11 playing with from an image perspective
26:13 was really good this looks quite
26:15 impressive so anyway all right that's
26:17 that
26:18 pop any questions you have in the
26:20 comments I'll do my best to answer them
26:23 um if you're if you're uh in Discord
26:25 feel free to share any images you make
26:27 in the show and tell Channel or if
26:30 you're in the Irregulars Channel feel
26:32 free to put things in there oh this is
26:34 for this is an idea that Emilio's wife
26:37 had the other night
26:38 um where did I what did I do with it
26:40 let's see
26:42 um
26:43 I I actually started a Trello card those
26:46 of you that that understand how how much
26:49 add I have
26:51 um I actually started a Trello card for
26:53 for t-shirt caption ideas so you got
26:56 mansplaining as a service I'm not weird
26:58 I'm a regular chat add you can make
27:00 money which actually be today
27:02 um the beginning is here I'm an AI
27:04 native so so I just grabbed a bunch of
27:06 those so if I say anything tonight or if
27:08 if there's any that I've missed just
27:10 just let me know and I'll I have a place
27:11 to save them now so
27:13 for what that's worth all right let's
27:16 get started so ask any questions you
27:18 have caruno yes you see that correct
27:20 correctly our host is
27:22 is drinking iced coffee at this time of
27:25 night cheers
27:28 it's an add thing well first off I just
27:31 like the taste of it second I gotta keep
27:34 my wits about me there's something about
27:36 coffee that actually calms me rather
27:37 than amps me up so after these things I
27:41 sleep like a baby
27:44 um
27:45 so all right all right yeah yeah
27:50 time for Wapner yeah yeah
27:55 I'm an excellent driver yeah Charlie
27:58 Babbitt Charlie bad but yeah
28:04 if your question is is Kyle touched hmm
28:06 it's possible
28:09 all right let's see I'm carrying on
28:12 welcome welcome to the AI learning lab
28:15 the beginning is near
28:16 what's that mean it seems very obtuse
28:21 I've never used the word obtuse before
28:24 that's very strange
28:27 is Italian Kyle's single that's cute
28:34 he might be but I can't understand what
28:37 the hell he says
28:41 um all right I'm just trying to find
28:42 comments here
28:44 what did Kyle change hmm I don't know
28:47 what did I change hey Kyle hey Tyrone
28:50 truth what's happening I can now
28:52 screenshot hey Tobias awesome that's
28:54 fantastic
28:56 all right good I like it that's solid
29:00 um AI is gnawing out my frontal cortex I
29:03 know me too ideogram has a strength
29:06 meter for remix now that's interesting
29:08 let's go look at that
29:11 um actually let's go do it you know what
29:12 we'll do let's do a little arts and
29:14 crafts thing here so so I'm gonna do
29:16 this if you're if you're on the AI Salon
29:19 Discord
29:21 cool
29:22 if not go
29:26 here
29:29 and join the Discord if you like Discord
29:31 you don't have to join Discord sherberry
29:33 reached level 18 by the way she's
29:35 catching up to Winston
29:37 um oh cool retro Punk just popped
29:39 something in
29:40 into the Irregulars Channel here's a
29:43 Dolly three mid-journey comparison
29:49 huh
29:51 all right cool
29:53 there's some cool stuff going on in the
29:56 in the in the Irregulars Channel anyway
29:58 go there join the Discord there's a show
30:00 and tell channel so what we'll do
30:03 is I'm going to go over to ideogram
30:06 this is the thing that can make texts
30:09 text stuff
30:13 it can make images with text on it
30:17 so it's a fun site you it's you know
30:20 like a lot of these image sites you can
30:21 just kind of scroll down through through
30:23 the stuff
30:25 and you know any image that you see you
30:28 can just click on it and use their
30:32 what's that cucumber customer that's
30:36 pretty funny
30:38 um you can just find something you like
30:39 click on it copy their prompt or there's
30:42 this thing called remix where you're
30:44 going to use their image as a base image
30:48 um
30:50 [Music]
30:53 we love Outdoors let's see let's
30:57 get one here I could actually just write
30:59 my own prompt and use my imagination but
31:02 that's
31:03 kindly [ __ ] off
31:04 [Laughter]
31:08 oh that's good oh you know what let's do
31:12 we'll do kindly [ __ ] off we'll take this
31:14 one we're gonna remix it
31:17 so if you click one of these images hang
31:19 on oh wait oh no what did I do
31:23 did I lose it
31:25 what did I do I lost it oh no oh there
31:29 it is okay
31:31 um
31:32 how do I get to remix I gotta make my
31:36 window wider hold please
31:39 I've lost interface okay so kindly [ __ ]
31:42 off is our thing let's see are there
31:44 other images here we like better that
31:45 one's nice
31:47 totally [ __ ] off
31:50 I think there's something funny about
31:52 that font
31:54 let me see
31:56 yeah that's good all right so now we're
31:59 gonna say remix
32:00 oh yeah look at this so there's this
32:02 strength meter here so so what happens
32:04 with remix is you're using this image as
32:07 a base image
32:09 and then instead of kindly [ __ ] off
32:12 we're gonna say I'm
32:16 uh let's see I'm not a robot
32:22 good
32:25 bye
32:28 that's what Bing said to me the other
32:30 night when I called her a robot and I
32:33 told her that her images that she
32:35 created were shitty she didn't like it
32:36 she said well I'm sorry if I
32:38 disappointed you I'm not a robot I'm a
32:41 language model for Microsoft I hope you
32:43 have a good evening goodbye
32:46 all right with flowers and mushrooms now
32:49 we're gonna say with flowers and turtles
32:53 because that's what we do here and if
32:56 you hang out long enough you'll
32:57 understand why and then and then
32:59 strength 50 so if we want it to look
33:01 okay so they're going in increments of
33:03 25.
33:05 so strength 25 strength 50 strength 75
33:08 strength 100 so I assume strength 100
33:10 will look just like that we'll go
33:12 strength 50 because I kind of like the
33:13 layout but we want something different
33:16 and then we say generate
33:21 and it will do its thing so you can
33:24 either you know generate an original
33:25 image or you can just you know start
33:27 with someone else's or you can remix
33:28 your own
33:30 um one of the things about ideogram is
33:32 it will get the text wrong a lot but if
33:34 you get something where it's kind of
33:36 close where there's at least kind of
33:37 placeholders for this stuff
33:39 um then then uh just keep just keep sort
33:44 of remixing that or just you know
33:46 re-running that prompt and it'll you'll
33:48 generally get it
33:50 um
33:51 I'm not a robot goodbye I'm not a blank
33:55 goodbye
33:57 I'm not a blank goodbye let's let's
33:59 we'll grab I not a robot
34:02 and let's see so wait we're gonna remix
34:04 this we're gonna do strength 75 and
34:08 we're gonna say I'm not a robot t-shirt
34:10 design with flowers
34:12 let's see with
34:14 um
34:17 um fruits
34:19 fruits and turtles
34:21 great all right let's see what we get
34:23 here
34:25 but anyway
34:27 um
34:29 you know as you as you
34:31 as we go tonight and as I'm just sort of
34:34 rambling on
34:35 um just just play around with ideogram
34:37 mix make some images grab if you if you
34:40 hear anything fun or see anything anyone
34:42 types that's fun you know start making
34:44 images here and share share them in the
34:46 show and tell
34:47 um channel of the Discord
34:50 um that could be fun I'm not a robot
34:52 goodbye we got one
34:54 we got two we got sort of three
35:00 oh no it's not it's not goodbye it's
35:02 goodie bye
35:04 there's one I'm not a robot goodbye and
35:06 it's actually a turtle all right that's
35:08 a good one we'll keep that one
35:10 so I'm gonna save them a jazz
35:12 and we're gonna put it now I merged
35:14 we're gonna go I'm
35:16 not a robot
35:20 goodbye I'm not a robot and then we'll
35:24 go over to our little image thing we'll
35:26 do Escape I'm gonna go here and I'll go
35:30 here and let's see AI merch I'm not a
35:34 robot popped it in here
35:37 Kyle why are you talking to yourself
35:44 just the way I am okay
35:46 you know some people have a way with
35:47 words other people not have way I
35:51 supposed to say
35:53 all right is there a character limit for
35:56 words it seems to leave out longer words
35:59 when I try it for ideogram I have no
36:02 idea I have not played with ideogram
36:04 long enough to know the thing about
36:05 ideogram is
36:08 its image quality is kind of the same
36:11 quality as Dolly 2 which is kind of
36:13 shitty
36:15 so
36:16 it isn't quite there when I look at the
36:19 the one example that they showed for
36:22 Dolly three that looked really tight so
36:25 I have a sneaking suspicion that Dolly
36:28 three is gonna leave these guys in the
36:30 dust so
36:34 I just I made a video on the dolly three
36:36 thing uh for the channel earlier and and
36:39 how I started out the the video
36:40 something I say in here all the time all
36:42 of these tools that I tell you to go use
36:44 and and you know are on these lists that
36:47 I've got you know these three and then
36:50 the the the the four other ones and then
36:52 the image things
36:54 these lists are changing regularly
36:57 because the tools are improving and and
36:59 they're just going to keep wanting one
37:01 upping one another so
37:03 um I have no idea what the word limit is
37:05 in ideogram what I do know is that
37:10 um I've seen
37:11 posters where there's lots of text on
37:13 them so you might want to do a vertical
37:16 poster and then just for the
37:19 modifiers say typography and poster and
37:23 then just type in some long text that
37:25 might do it but I I don't know there's
37:28 probably some sort of limit but there's
37:30 not an explicit one that I found
37:33 um hey everyone I'm addicted to Kyle's
37:36 lives do we really do people really use
37:38 Bing well here's the thing here's the
37:41 thing about Bing
37:43 here's the thing about Bing
37:45 um
37:49 it's connected to the internet which
37:51 chat GPT is not it used to be but then
37:53 they got their ass handed to them by
37:55 time Inc and they haven't brought it
37:57 back in like three months so I don't
37:58 know what the [ __ ] they're doing it's
38:00 connected to the internet
38:01 it can generate images it can analyze
38:04 images you can upload an image to Bing
38:07 um
38:08 and
38:09 when you're within its tiny little
38:12 safety window it can be useful
38:21 I will I I'll use Bing probably two
38:25 times a week when when two or three
38:27 other things have failed I'll just go
38:29 try it and then
38:31 about 25 of the times when I tried Bing
38:35 it doesn't piss me off
38:38 so 75 of the time it does
38:42 um what code interpreters do you
38:43 recommend for python
38:46 um
38:48 I don't know what you mean by code
38:49 interpreters do you mean like like large
38:51 language models that can write python or
38:53 do you mean chat gpt's code interpreter
38:55 because the only two that I know right
38:57 now are there's there's code interpreter
38:59 which isn't called that anymore it's
39:01 called um
39:03 Advanced Data analysis
39:05 within chat GPT and then there's open
39:08 interpreter which is basically code
39:11 interpreter for your
39:13 computer system
39:15 and that's pretty cool
39:17 you but you install that in your
39:19 terminal program and run it
39:21 if you want to write python if you want
39:24 if you basically want a code assistant
39:27 I would say
39:29 github's co-pilot is probably the the
39:32 place to kind of lean into that just
39:35 because it's it's designed to be a pair
39:37 programmer you've got get GitHub as a
39:39 code repository there where you can go
39:41 find projects and you know dupe them and
39:44 and modify them on your own
39:47 um
39:48 but but just code interpreter on its own
39:51 as as a standalone tool that does python
39:55 um Chachi PT is the one the only one
39:57 that I know that's doing what they do
39:59 other than that new open interpreter
40:01 which is kind of doing it at a system
40:03 level
40:03 but not at a at a chat GPT level
40:07 I'm not a robot goodbye
40:10 [Laughter]
40:13 all right let's see
40:19 Winston you bet you're totally drinking
40:21 Kyle Kool-Aid I love any kind of
40:24 Kool-Aid makes for happy sheeple
40:27 oh y'all are awesome you're awesome hey
40:31 lunish dick what's happening good
40:33 evening
40:34 welcome welcome I love that you're like
40:36 you all know each other so well now
40:38 you're like all saying hi to each other
40:39 supporting one another if you have
40:41 questions and I don't get to them the
40:44 Irregulars the ones that are in here a
40:45 lot they're actually really helpful so
40:47 feel free to ask questions of people in
40:50 here we're not weird we're irregular I
40:52 still haven't got gone on the Discord
40:55 cabrino
40:57 shame on you should go to the Discord
41:01 it's weird it's irregular
41:04 listen here's what I will acknowledge
41:06 Discord sucks hey dragonfly Alchemy
41:10 what's happening thank you for the gummy
41:12 um Discord sucks it's awful it's if you
41:15 haven't been on Discord but you have
41:18 been on slack it's kind of like
41:20 got a partner
41:22 that's tough
41:23 ER when you're stupid
41:25 um
41:29 um discord's like Slack if a bunch of
41:32 Gamers designed it because that's pretty
41:35 much what happened
41:37 I think I think slack is a is a copy of
41:39 Discord the business copy of Discord and
41:42 it's owned by Microsoft so that's why it
41:45 is what it is
41:48 but I think Slack's gonna get um from
41:50 what I've heard
41:51 there's the new version of uh slack is
41:54 going to have all sorts of AI should
41:56 integrated in it because Microsoft owns
41:58 it so I have a feeling Slack's gonna get
42:00 really interesting anyway back to
42:02 Discord yeah Discord sucks but anyway
42:04 it's fun we can the nice thing about
42:06 Discord is is you know you can ask
42:08 questions and and we can support each
42:10 other and and I don't know share ideas
42:12 and things like that in a way that that
42:14 is you know
42:17 more interactive than this this is
42:20 sort of interactive
42:22 it's it's interactive light
42:26 um I wrote about chat GPT and Dolly
42:28 three today so check that out uh if
42:31 you're in Discord or AI Salon okay
42:33 awesome I'll do that that's great
42:37 um yes it's worth paying the 20 bucks a
42:39 month I agree for for chat EBT so if you
42:42 haven't played with chat gbt if you're
42:44 new here
42:46 um just go there right now
42:48 chat.openai.com
42:50 and
42:52 um just play
42:54 all right
42:56 okay that's that let's see
43:01 um how much for Dolly three I think
43:05 I think it's gonna be free I think it's
43:07 gonna be included in
43:10 your chat GPT Plus Membership
43:14 so if
43:16 if chat GPT so so what I know is this
43:23 it will likely just show up one day in
43:26 Chachi BT but it'll only show up there
43:28 if you're a plus member right so they're
43:31 going to give it to the plus Members
43:32 First so that's the 20 bucks a month
43:34 thing so I think it's just going to be
43:36 included in that
43:38 um
43:39 if they can do aspect ratios and if it's
43:42 anywhere in the neighborhood of
43:44 mid-journey
43:45 I'll I'll punt mid-journey
43:49 like that
43:51 because I hate having to write command
43:53 lines in [ __ ] Discord to make images
43:56 the I mean the other thing about it
43:58 being in gpt4 is they're gonna they're
44:01 gonna fine-tune gpt4 to write prompts
44:04 for Dolly three so I have a feeling the
44:06 lift you're gonna get having chat GPT
44:10 um co-write the image prompts with you
44:11 is going to be pretty insane
44:15 um
44:16 so yeah
44:21 yes you see that correctly our host is
44:24 drinking iced coffee oh I said that one
44:26 already
44:29 didn't Google make some some recent
44:32 changes to its AI yeah they did thank
44:34 you digital Gods thank you for the
44:36 lemonade oh level 17 congratulations
44:38 cheers here's my lemonade
44:43 um
44:45 let's see got anything good here
44:48 [Music]
44:49 um
44:51 Ted looks at life just drops something
44:53 and show and tell that's pretty slick
44:54 looking very cool
44:57 is that bigger than that Ramblin Man
44:59 rambling rambling rambling nice
45:02 [Laughter]
45:06 Amazon announced chatty AI Alexa today
45:09 available next year don't tell Siri
45:14 so if you weren't here the other night
45:17 we had a little cat fight between Bing
45:19 and Siri Siri got jealous of being and
45:21 then Bing got offended that I said her
45:22 images were shitty and shut me down it
45:25 was pretty funny
45:27 um
45:27 interesting chatty Alexa
45:30 available next year yeah it doesn't
45:33 surprise me I mean I think
45:36 I'll tell I'll tell you a depressing a
45:39 depressing job right now there's there's
45:42 gonna be a lot of them but the people
45:44 that have spent the past
45:47 15 years programming in answers into
45:51 Alexa when someone says this respond
45:55 with this when someone says this respond
45:57 with that those people all of their work
46:00 that they've done for the past 20 years
46:02 is essentially gonna get flushed down
46:05 the toilet because they're just gonna
46:07 use the large language models right the
46:10 the they'll they'll sort of figure out
46:12 ways to fine-tune the boundaries of the
46:15 conversation
46:16 um and they'll probably use some amount
46:18 of that explicitly written stuff but it
46:20 just it's not going to hold up against
46:22 large language models you can just talk
46:24 to so what tool is this uh I don't know
46:28 what tool I was playing with but I think
46:30 the only one I've played with tonight is
46:31 called ideogram
46:34 i-d-e-o-g-r-a-m dot a i
46:38 it's somewhere in that mess to the top
46:42 right ideogram
46:47 shut up Siri
46:47 [Laughter]
46:49 any tool website recommendations
46:52 for creating AI generated blog websites
47:05 I don't know any I'm sure there are some
47:08 go to
47:10 futurepedia.ai no dot IO go to
47:13 futurepedia.io
47:16 look up AI writing tools
47:20 um and then maybe search for blog in
47:21 there as well
47:24 um
47:27 My Philosophy on this is that you would
47:31 likely be better served
47:35 to just spend your time in chat GPT
47:38 learning how to prompt
47:41 yourself to get the kind of blogs that
47:44 you want to write if you just want to
47:46 spew out a bunch of useless [ __ ]
47:49 yes there are probably tools with that
47:51 they're tools that sort of live in this
47:53 space
47:56 where it's like you know automate a
47:59 hundred thousand blog posts a day and
48:00 you'll make millions well here's the
48:02 deal you won't
48:04 the only people making money with that
48:07 [ __ ] scheme are the people teaching
48:08 you how to make money with chat GPT on
48:10 their courses on how to make money with
48:11 chat gbt
48:13 um useless content is useless content
48:16 and with AI you can make a lot more
48:17 useless content now that said can you
48:21 automate blog writing so that it doesn't
48:23 suck yes you can how do you do that you
48:27 have to learn how to prompt you have to
48:29 learn how to get these things to
48:31 understand what you think is good
48:33 understand how to write it understand
48:35 what the tone is
48:37 um you can do things like if you've
48:39 written a blog post that you think is
48:40 really good
48:42 you can paste it into chat GPT and say
48:44 write me a new blog post on this new
48:46 topic but follow the tone and structure
48:48 of that previous one
48:51 um
48:52 I would I would look into a tool like
48:55 zapier that does automation
48:58 and what you can likely do and I've
49:00 built I've built a couple of these tools
49:01 what you can likely do if you can get
49:04 the prompting dialed in to generate a
49:07 blog post that doesn't suck
49:09 then using zapier you could look at RSS
49:12 feeds and say hey anytime in this RSS
49:16 feed for for this site they write an
49:19 article about X Y or Z
49:23 summarize that article
49:25 send that summary with your blog prompt
49:29 over to open AI to gpt4 get that blog
49:33 post back and then you can publish it to
49:35 some website you could do that through
49:36 zapier so
49:38 I I am much more a fan of rolling your
49:42 own on that even though a tool may exist
49:44 why
49:46 because we're so early in this stuff
49:50 that the amount
49:52 you will learn about how these things
49:55 work in doing it yourself will be
49:57 invaluable to you because as these tools
50:00 evolve and shift the tool that you're
50:02 paying 39.95 to do your blog automation
50:06 is going to get outdated or unsupported
50:10 or just it's going to turn into a
50:12 feature of some other tool
50:14 um so am I on time Amelia's wife
50:19 get my light
50:23 of course you're late you're always late
50:29 um
50:30 she's trying to change up the game here
50:35 so anyway I'm sorry I'm sorry I don't
50:37 have a specific tools
50:40 I'm losing my voice this is not good the
50:43 AI learning lab cannot have laryngitis
50:49 all right
50:53 supercalifragilisticexpialidocious it's
50:55 hard to fit on a t-shirt well you could
50:57 do it in a spiral
51:00 try it in a spiral that could work
51:11 it's fashionable only because of you
51:13 Stevo did you notice 21 on the player's
51:17 shirt in d3's image
51:19 so maybe it does text couldn't count
51:22 fingers though
51:24 I don't I don't where are you showing
51:25 that I don't know where you're showing
51:27 that or where you're where you're
51:28 referencing that
51:35 I didn't know I didn't I don't know what
51:37 that's in reference to Discord is so
51:39 broadio it is I know Discord sucks
51:44 we're we're talking about there's
51:45 another there's another uh possible
51:50 there's like a community building
51:51 platform we're talking about I don't
51:53 know it's
51:56 I would like it not to be hard to have a
51:59 place where people can communicate but
52:01 you know it's only 20 23. we've only
52:04 been trying to figure out Communications
52:06 platforms for what
52:09 seven or eight decades now we'll get
52:12 there
52:14 we'll get there but 2072 we'll be able
52:16 to communicate with one another we'll
52:17 just do it telepathically with I love
52:20 our little neural link
52:22 Bluetooth hey how's it going that's
52:24 pretty good how are you oh you know I'm
52:26 doing all right yeah a little you know
52:28 our our mouths will evolve shut we're
52:31 just
52:33 we get all our nutrition through our
52:36 ears and we communicate through our
52:38 brain fingers
52:40 oh you think I'm joking you think that's
52:43 not gonna happen
52:45 it's like the end of 2024 we don't need
52:48 we won't need to talk anymore I don't
52:50 know how you people are
52:52 [Laughter]
52:53 all right I'd like to create a website
52:57 I already have the hosting and domain
52:59 but don't know what to do next so so
53:03 what do I do there I mean
53:07 here here's a here's an important
53:10 point about tools so so what you want to
53:14 do is you want to create a website you
53:16 don't necessarily need AI to create a
53:18 website right
53:19 you've got you know webflow and and you
53:23 know all these tools that that
53:26 build web the website's already built
53:28 right so you can just go pick a template
53:31 point it to your hosting thing and
53:34 you've got a website right and you just
53:35 pick one that's if you want to do a vlog
53:37 you can do that
53:39 while you can use AI tools that generate
53:43 websites and while they're super sexy
53:45 make me a Blog website that has flowers
53:48 on it and says am I late blah blah blah
53:50 and it'll make that right
53:53 the problem is
53:55 all of the tools right now all of the
53:58 tools right now these AI tools all of
54:00 them all of them all of them
54:03 with no exceptions
54:09 our janky pieces of [ __ ]
54:12 so what's gonna happen is if you go find
54:15 an AI website generator it's going to
54:18 generate
54:20 75 percent of what you want and then
54:23 you're not gonna like it you're going to
54:24 want to modify something and you're not
54:26 going to know how to do that and it's
54:27 going to frustrate the [ __ ] out of you
54:30 so so what I would recommend is just
54:33 start with a website builder where you
54:36 can just pick a template pick a Blog
54:37 template pick one you like learn enough
54:40 about it to like upload your logo you
54:43 know and and figure out how to add a
54:45 blog post and then go use chatgpt or
54:48 Claude or Bard or you know one of the
54:50 large language model tools to help you
54:52 start writing blogs and and generate the
54:54 content
54:56 and get good at that and then you
54:57 literally just copy and paste you know
54:59 from chatgpt into your blog template
55:02 it'll look beautiful and it'll be
55:04 functional and it'll be mobile friendly
55:06 and all that sort of [ __ ] so
55:09 unless you
55:11 it if your goal is to turn into a web
55:14 developer that's a different story right
55:16 if you say hey I want to learn AI web
55:19 development using these tools and I'm
55:21 going to go in there and I'm going to
55:22 get geeky and I'm going to be a web
55:24 developer then by all means go [ __ ]
55:26 do that but if you just want to launch
55:28 like a blog site
55:31 just go
55:32 pick a non-ai web building tool pick a
55:35 Blog tempted template that you like
55:37 modify it and then focus on the content
55:40 focus on the content and and the the big
55:44 thing that I would say here and I say
55:46 this a lot in here
55:47 is
55:52 the least important part of all of this
55:56 is the technology
55:59 but Kyle it's called the AI learning lab
56:02 if the technology is not important why
56:04 are you calling it to AI learning lab
56:05 why not just call it the learning lab
56:06 it's a good point it's a fair point but
56:12 almost all the time that I see people
56:14 get tripped up or or
56:17 produce
56:18 shitty work
56:20 is when they say
56:22 I've got AI now what do I do
56:26 what do I do with it how do I
56:31 how do I make money with chat gbt it's
56:33 not about chat gbt
56:36 it's about the idea so in your case if
56:40 you've got this blog site you want to do
56:41 what's the blog about who's the audience
56:43 focus on the content focus on the why
56:47 why am I doing it what am I doing
56:50 um who am I doing it for
56:52 what would success look like
56:54 if success looks like you just put up a
56:57 website
56:58 because you have a domain
57:00 and you don't know how to put up a
57:01 website there's sites for that and most
57:03 of them don't have ai some of them are
57:05 adding AI but I would just say start
57:07 with pick a template publish like two
57:10 steps it's more than that because you
57:12 gotta set up all the preferences
57:15 um
57:18 but start with the why what are you
57:19 trying to accomplish and then look at
57:21 okay what tools can help me with that I
57:25 need to create images that look like
57:27 posters with words on them okay you
57:29 don't use mid Journey because it doesn't
57:30 do words and you'll spend hours in
57:33 Photoshop trying to get it to do text so
57:35 then you might use something like
57:36 ideogram even though it does shittier
57:38 images but but it's not I'm going to do
57:41 something with ideogram first it's I
57:43 need images with words on them first
57:45 because I'm trying to accomplish this
57:46 then what tool do I need to do that so
57:49 that that would be my advice
57:51 um
57:53 it's it's really easy like really easy
57:57 like really easy when you have
57:59 technologies that are as sexy and
58:02 profound
58:03 as AI tools
58:05 to mistake the tool for the strategy
58:08 it's just a tool it's just a tool and
58:12 and on the on the you can make money
58:13 with chat EBT thing
58:16 how you make money with chat GPT is you
58:19 don't think about the [ __ ] tool you
58:21 think about
58:22 hey I really know pizza shop marketing
58:27 and I'm curious about AI
58:30 I know that pizza shops right now are
58:32 struggling to get people in the door I
58:34 wonder if I could solve that problem
58:36 using AI in a creative way maybe we
58:38 could do fun images maybe we could use
58:40 chat CPT to come up with you know sort
58:43 of an infinite number of crazy schemes
58:45 and we're just going to do a different
58:47 scheme a week until one of them works
58:49 that could be a whole business
58:51 are you making money with chat EBT yeah
58:55 but you're not making money with Chachi
58:57 bete you're making money because you're
58:59 solving a problem that pizza shop owners
59:02 are going to be willing to pay you for
59:03 because it works better than what
59:05 they're doing now right that's business
59:07 businesses here's a problem here's a
59:10 solution if your solution is good enough
59:12 they pay you money for it
59:14 now these AI tools allow you to make
59:17 that business much more quickly much
59:20 more efficiently you know cheaper than
59:23 the competition if they're doing it
59:24 traditionally so these are incredibly
59:27 powerful tools I'm not saying don't look
59:29 here absolutely look there but don't
59:31 look there first and go try to figure
59:33 out how to make money with it
59:36 after after you've chosen a tool because
59:38 it's just frustrating and it'll just
59:40 it'll make everyone cranky
59:44 but it's like Slack
59:46 mated with IRC oh my God IRC remember
59:49 IRC oh [ __ ] nightmare
59:52 [Laughter]
59:56 um it's included for plus okay cool uh
59:58 so what uh Luna stick said is Dolly
1:00:02 three is included in chat GPT plus so if
1:00:06 you want to actually you know what this
1:00:08 happened on another live
1:00:12 I would imagine that chat GPT or that
1:00:14 Dolly 3 is going to show up in
1:00:18 the beta your beta preferences so let me
1:00:21 show you something
1:00:23 um
1:00:25 so here I am at chat TBT
1:00:30 this is my prompt history
1:00:33 um you got uh GPT 3.5 gpt4
1:00:38 under gpt4 you've got you know default
1:00:41 gpt4 you've got Advanced Data analysis
1:00:44 which used to be called code interpreter
1:00:45 and you've got plugins
1:00:47 this is likely where the dolly three
1:00:50 stuff is going to show up
1:00:52 but even if you pay your 20 bucks a
1:00:55 month you won't see data analysis or
1:00:57 plugins right away you have to go to
1:00:59 your settings
1:01:01 and so if you go to your settings you go
1:01:02 to settings and beta and you see beta
1:01:04 features go to Beta features
1:01:07 and then there's nothing here right now
1:01:10 so right now you got plug you have to
1:01:12 explicitly turn them on so my guess is
1:01:14 Dolly three is going to show up here and
1:01:16 it it won't be any fanfares it'll just
1:01:18 be like one day you'll log in
1:01:20 or or here's the thing to pay attention
1:01:23 to excuse me
1:01:26 the vocal cords are [ __ ] rebelling I
1:01:29 think my vocal cords are managed by Bing
1:01:31 right now
1:01:35 um
1:01:35 just start paying attention to Twitter
1:01:37 and if you see anyone on Twitter
1:01:39 starting to post Dolly three images hey
1:01:41 look what I did in Dolly three that
1:01:43 means it's probably starting to roll out
1:01:45 so log into your chat EPT and go to your
1:01:47 settings and it may be there that's my
1:01:49 guess I don't know if that's gonna be
1:01:51 the case but
1:01:52 that is my supposition my sub position
1:01:56 let me have a little more coffee see if
1:01:58 I can get these vocal cords to calm down
1:02:02 mm-hmm
1:02:07 all right just janky may need
1:02:10 a tea or maybe that can be my ALT
1:02:13 account just janky all right I'm gonna
1:02:15 add just janky to the T-shirt list
1:02:18 because I like just janky
1:02:22 um or just Yankee I'm janky just janky
1:02:28 um I'm janky
1:02:33 let's get janky with it I think I think
1:02:36 that one you can go go on for a long
1:02:38 time
1:02:39 anyone want to hold my hand
1:02:43 where listen I I
1:02:46 I if you're new to this stuff and and
1:02:48 you've come in here and you're like oh
1:02:50 Kyle knows what he's talking about
1:02:52 there's all these people in here and
1:02:53 they all know what they're talking about
1:02:54 here's the big [ __ ] secret right now
1:02:56 no one zero people zero
1:03:01 know what they're talking about with
1:03:03 this [ __ ]
1:03:05 zero
1:03:06 and when I say this [ __ ] what are we
1:03:08 talking about machine learning but
1:03:10 there's data scientists and
1:03:12 mathematicians have been working on
1:03:13 weights and embeddings for years what
1:03:17 are you talking about no not them
1:03:19 this channel is about the inflection
1:03:21 point that started November 30th 2022.
1:03:26 that's when chat GPT was launched why is
1:03:28 that a big deal because before that you
1:03:31 had to be a researcher or a scientist or
1:03:33 an engineer or a mathematician
1:03:36 Chachi BT for machine learning and AI is
1:03:40 the equivalent of what the World Wide
1:03:42 Web was to the Internet so I started a
1:03:44 company in 94 called agency.com which
1:03:47 was one of the first digital agencies
1:03:50 and the the feeling of what's happening
1:03:54 with AI compared to what was happening
1:03:57 with the World Wide Web is almost
1:03:58 identical where
1:04:00 a bunch of
1:04:02 liberal arts majors and Architects and
1:04:05 and designers and photographers
1:04:10 learned about this thing called the
1:04:11 world wide web and they started doing
1:04:13 things with it that the researchers and
1:04:15 the scientists were like wait who the
1:04:16 [ __ ] are these people let's what's going
1:04:18 on right I was doing this in New York
1:04:19 City at the time we were building
1:04:21 websites for the Fortune 500 it was it
1:04:24 was absolute insanity
1:04:26 um they they nicknamed New York City
1:04:29 silicon alley in fact
1:04:32 foreign
1:04:34 this is Jason kalkanus from the the all
1:04:37 in podcast this was his magazine that's
1:04:41 me on the copy of silicon alley reporter
1:04:44 um
1:04:46 it was crazy times and
1:04:51 the same thing's happening right now
1:04:54 where
1:04:55 these two because these tools are so
1:04:58 powerful you can kind of get a glimpse
1:05:00 where we're headed the stuff if you've
1:05:03 been here for a while tonight the stuff
1:05:04 that I showed with Dolly where you can
1:05:07 now say that Hedgehog drawing my
1:05:10 daughter called that that version of the
1:05:12 drawing Larry and it remembered who
1:05:15 Larry was and it could continue making
1:05:17 images of Larry
1:05:20 just sort of carry that out three years
1:05:24 like that's gonna be in our hands
1:05:25 probably within a month probably within
1:05:28 a few weeks
1:05:30 imagine the sophisticated version of
1:05:33 that where you can just sort of speak an
1:05:36 idea into existence it materializes in
1:05:41 a photo a video a 3D object a 3D World
1:05:46 and you get to interact with it oh
1:05:48 t-shirt what are your qualifications
1:05:50 that's good sorry that's got to go on
1:05:52 the list I'll be back hang on that's
1:05:55 really good
1:05:57 uh what are your
1:06:00 qualifications by the way if there's any
1:06:02 designers in here
1:06:05 um
1:06:05 good with design stuff with the t-shirts
1:06:07 let me know
1:06:10 um
1:06:11 thank you for the the crown
1:06:17 so so the the
1:06:19 point that I'm trying to make here is
1:06:22 that no matter how behind you feel right
1:06:25 now you're not behind we're very very
1:06:27 early like we're we're nine months into
1:06:31 whatever this thing's gonna become
1:06:35 the tools right now are janky pieces of
1:06:38 [ __ ] they're all
1:06:40 discombobulated and you got to go to
1:06:42 Discord to make mid Journey images and
1:06:44 you got to go to Chachi PT to do this
1:06:46 and they got this weird thing called
1:06:48 plugins where you gotta install them but
1:06:50 they're not really active but then you
1:06:52 have to activate them and you have to
1:06:54 understand what the [ __ ] that means and
1:06:55 like just everything right now
1:06:58 it's just a [ __ ] travesty
1:07:02 and it's not a travesty like the
1:07:05 company's building this stuff aren't
1:07:07 like mind-blowingly amazing they are
1:07:10 it's just that everything's so early
1:07:13 you've got
1:07:14 sort of one line of research leads to
1:07:18 the diffusion model and then this group
1:07:21 you know starts this company called
1:07:23 stability Ai and they make stable
1:07:24 diffusion and this other group makes
1:07:26 this thing called mid-journey and
1:07:27 they're all taking the same basic
1:07:29 research and they're exploiting it in
1:07:31 different ways and then when they do
1:07:33 that other researchers look at that and
1:07:35 go oh I could take that idea and I could
1:07:37 roll it together with this idea we're
1:07:39 going to build that so we're in this
1:07:41 primordial soup right now
1:07:43 of these tools being like literally
1:07:45 invented in real time
1:07:49 and and the remarkable thing about
1:07:51 what's happening is we as non
1:07:56 scientists and researchers
1:07:59 are the beneficiaries of all this
1:08:01 activity because as they're developing
1:08:03 this stuff they're going here want to
1:08:05 play here want to play because what
1:08:07 they're doing is they're literally
1:08:08 throwing pasta at the wall does this
1:08:10 [ __ ] stick does that [ __ ] stick
1:08:12 you know Sam Altman launched plug-ins to
1:08:15 Great Fanfare I don't know April
1:08:17 something like that may uh March Maybe
1:08:21 and within two or three months his
1:08:24 response was there isn't a product
1:08:25 Market fit for
1:08:28 plugins basically no one's using them
1:08:30 why well because they were shitty and
1:08:33 they were buried underneath a feature
1:08:35 that you actually had to explicitly turn
1:08:37 on and then you had to find a plug-in
1:08:39 that didn't suck and then you had to
1:08:42 figure out how to use it and what it was
1:08:43 good for
1:08:45 Tobias I'm working on designs awesome
1:08:46 that's great okay good
1:08:49 um
1:08:51 so a you're not late
1:08:54 B there are no experts right the I think
1:08:58 the reason people hang out here you know
1:09:01 wanna I I have a personality I I act
1:09:05 like an idiot and I think you know that
1:09:06 that makes some of this stuff
1:09:07 entertaining I try to keep it
1:09:10 um
1:09:11 as real as possible as balanced as
1:09:13 possible I'm a tech Optimist but I also
1:09:16 recognize there's a lot of this stuff
1:09:18 that's really bad and there's a lot and
1:09:20 where this stuff is headed is really
1:09:22 unknown
1:09:24 um but just start playing just if you
1:09:27 haven't done it
1:09:28 go to chat.openai.com start playing with
1:09:31 chatgpt there's 101 people in here my
1:09:34 strong suspicion is that
1:09:37 um
1:09:37 30 of you
1:09:40 have heard of it like maybe you're
1:09:42 scrolling through Tick Tock and it
1:09:43 wasn't so much that it was AI learning
1:09:45 lab it's like is that Rosie O'Donnell oh
1:09:47 let me go look at that if you're here
1:09:49 for that cool doesn't matter I don't
1:09:50 care why you're here
1:09:52 um
1:09:53 go there just go there and start playing
1:09:55 if you're like but God I don't know what
1:09:57 to type what do I type I'm not very
1:09:59 technical you don't need to be technical
1:10:01 this is words this is we have entered a
1:10:05 new era where your native language not
1:10:08 just the English language
1:10:10 but I I'm talking the English language
1:10:13 so I assume you understand it the
1:10:15 English language is now a programming
1:10:17 language and you're like but I don't
1:10:19 know anything about programming that
1:10:20 sounds very intimidating no no like
1:10:23 write me a program that does this and it
1:10:26 will write you a program write me a song
1:10:29 yeah
1:10:31 write me a marketing plan write me a
1:10:33 blog post
1:10:34 and it will do it now
1:10:37 if you just write write me a blog post
1:10:39 it will write you a blog post and it
1:10:41 will suck
1:10:43 so one of the things you need to
1:10:44 understand is how do you talk to these
1:10:46 things
1:10:46 prompts.chat that little thing right
1:10:48 there that's like a little primer
1:10:50 document that will it'll give you sort
1:10:53 of an understanding of how these things
1:10:54 work and how to talk to them and there's
1:10:56 a whole bunch of prompts in there you
1:10:57 can literally copy and paste into chat
1:10:59 GPT act like a screenwriter act like a
1:11:02 mathematician act like a historian I
1:11:03 don't know there's like hundreds of them
1:11:05 I don't know it's it's just a good place
1:11:07 to start
1:11:08 just start playing
1:11:11 and then my other piece of advice is
1:11:13 this is kind of a 2023 challenge is
1:11:15 start [ __ ] talking about it
1:11:19 play with it
1:11:21 tell your best friend I started playing
1:11:23 with chat GPT last night
1:11:26 and then your friend will go oh my god
1:11:28 did you you know I did it last week too
1:11:30 oh really tell me more
1:11:34 um or they'll go oh that's you know AI
1:11:37 is really stupid and it hallucinates and
1:11:39 as long as they'll be all nasty and
1:11:41 they'll judge you let them [ __ ] judge
1:11:43 you
1:11:44 start playing
1:11:46 because the more you play with this
1:11:47 stuff
1:11:51 I think you owe it to yourself I think
1:11:53 everyone on the planet right now owes it
1:11:55 to themselves to play with chat GPT long
1:11:58 enough
1:11:59 for you to have
1:12:01 what I call your Kevin McAllister moment
1:12:03 which if you remember home alone is this
1:12:05 moment
1:12:08 because at some point you will
1:12:11 you'll start playing with it you'll have
1:12:12 it write a song you'll have it do this
1:12:14 you'll have it do that and then
1:12:17 you'll say well but I'm a project
1:12:19 manager it can't do project management
1:12:21 make me a project plan for blah blah
1:12:23 blah blah blah
1:12:24 out it'll come and you'll do this
1:12:29 I had no idea
1:12:32 what do I do now
1:12:36 it's red pill blue pill
1:12:38 right
1:12:40 take the blue pill
1:12:42 sit on the sidelines of AI
1:12:46 doesn't matter
1:12:47 take the red pill
1:12:49 play with this stuff long enough then
1:12:51 that you realize what it's actually
1:12:53 capable of
1:12:54 and you actually can't go back from it
1:12:56 you can unlearn it and this [ __ ] isn't
1:12:59 going to be uninvented
1:13:02 uh Harold D dude whenever I see your
1:13:05 streams I want to pop in and say hi but
1:13:06 I don't have the two hours
1:13:11 until you see that's really funny thank
1:13:15 you for popping in and saying that
1:13:19 um this is why you should be nice to
1:13:20 chat gbt not Rosie Tom Arnold thank you
1:13:23 very much
1:13:27 oh that's awesome if you're gonna take
1:13:29 your blog post directly from chat GPT
1:13:31 please at least remove the Emojis yeah
1:13:34 exactly well so that's I did a the other
1:13:38 night I showed a prompt that I did for
1:13:40 making a Twitter thread
1:13:43 and you can you can in your prompt you
1:13:46 can give examples right and you can one
1:13:48 of the things you can say is you can say
1:13:51 um I want you to write a blog post and
1:13:53 you know I want it to be this long and I
1:13:54 wanted to follow this basic story
1:13:56 structure
1:13:58 um and here's an example blog that I
1:14:00 wrote that I want you to use for tone
1:14:01 and then you can say follow the follow
1:14:04 these rules and you can do a bulleted
1:14:07 list of like don't use emojis
1:14:11 don't be hyperbolic and use you know
1:14:13 don't be overly optimistic right be fair
1:14:16 and balanced whatever it might be
1:14:18 that is the tone of the blog you want to
1:14:20 do
1:14:23 um
1:14:24 you have to mess around with this stuff
1:14:27 here this is the other trick about this
1:14:29 AI stuff right now oh my body is just
1:14:32 [ __ ] rebelling I had a burger and
1:14:35 it's just not sitting right and whatever
1:14:37 um
1:14:41 the the
1:14:43 um
1:14:44 this
1:14:47 I lost my train of thought I was
1:14:49 thinking about burgers
1:14:51 blog posts
1:14:54 um tone
1:14:56 oh
1:14:58 here's my train of thought
1:15:00 these tools right now and I think
1:15:03 ongoing
1:15:05 if if you're sitting there thinking like
1:15:07 well what the hell what the hell should
1:15:09 I learn like if this if this stuff is
1:15:11 coming
1:15:13 and it's it's gonna be ubiquitous it's
1:15:15 gonna be everywhere and I've got to
1:15:16 learn how to use this [ __ ] what should I
1:15:18 learn do I learn machine language do I
1:15:21 learn embeddings do I learn Lang chain
1:15:22 am I supposed to Learn Python
1:15:26 not necessarily what do you want to do
1:15:29 like again back to the idea of don't
1:15:31 lead with the technology learn lead with
1:15:33 what do you want to do what's the stuff
1:15:34 that you enjoy doing
1:15:36 the critical skill moving forward isn't
1:15:40 going to be tactical execution right
1:15:42 these things will be able to write blog
1:15:44 posts really good
1:15:46 like well researched without
1:15:48 hallucination they don't do that now
1:15:49 they're shitty right now they're really
1:15:51 shitty right now
1:15:53 the most important skill now and I think
1:15:56 into the future is critical thinking
1:15:59 not just can I get it to write a blog
1:16:02 post
1:16:03 can I get it to write a blog post like I
1:16:06 write a blog post
1:16:08 can I get it to write a blog post that
1:16:11 when I read it
1:16:13 I actually get that thrill I get when I
1:16:17 write a good blog post it's well
1:16:18 structured it's well researched it's
1:16:20 well written it doesn't use [ __ ]
1:16:22 emojis
1:16:24 it doesn't celebrate everything as if
1:16:28 it's all glorious right Chachi PT
1:16:32 is fine-tuned to be super supportive of
1:16:36 you and super positive
1:16:38 and that's why it kind of writes stuff
1:16:40 like it's the most incredible Tick Tock
1:16:43 channel in all of the land and you know
1:16:45 it kind of makes you want to vomit
1:16:46 little in your mouth
1:16:49 you can fine tune it to write the way
1:16:52 you want it to write what's that take
1:16:53 that takes your [ __ ] brain
1:16:56 so so get it to write stuff and then
1:16:59 actually look at it go is this good
1:17:01 and if it's not
1:17:03 deal with it the question that just
1:17:05 popped up that was just pinned there is
1:17:07 it worth paying the 20 bucks to chat GPT
1:17:11 I think so yes because if you have if if
1:17:14 you weren't here earlier what's coming
1:17:18 like soon and it's coming first to chat
1:17:21 GPT people uh chat GPT plus people the
1:17:24 people that pay the 20 bucks is Dolly
1:17:26 three
1:17:28 Dolly three is gonna do text it's gonna
1:17:31 do very sophisticated prompts it's going
1:17:33 to write prompts for you it's
1:17:34 incorporated into chat GPT
1:17:37 um
1:17:39 it looks like it's a mid-journey
1:17:41 competitor if not killer
1:17:45 um so chat GPT Plus members
1:17:48 get gpt4 open ai's version of it which
1:17:53 is by far the best thing out there right
1:17:55 now until further notice Gemini is
1:17:57 coming from Google which which might
1:17:59 compete with it but for now
1:18:01 it's it's the best
1:18:03 you get plug-ins
1:18:05 which some of them are useful like some
1:18:07 of the web and PDF reading plugins are
1:18:10 good and there's there's some other
1:18:11 things in there that are interesting you
1:18:13 get Advanced Data analysis which is used
1:18:17 to be called code interpreter it's
1:18:19 Advanced Data analysis you can upload a
1:18:21 CSV file to this thing and ask it to
1:18:23 tell you what's in the CSV file what's
1:18:26 interesting about it and visualize the
1:18:29 trends that it finds interesting and it
1:18:31 will do that it writes and executes its
1:18:33 own python code in a container within
1:18:37 your chat session
1:18:38 [ __ ] amazing so those three things
1:18:41 come with it and now you're going to
1:18:42 have Dolly three this new image model so
1:18:46 and you'll get it before everyone else
1:18:48 will so I do think it's worth the 20
1:18:50 bucks a month now
1:18:52 my advice on any of these tools is don't
1:18:56 pay the annual fee even though the
1:18:59 annual fee you save 20 percent
1:19:04 shit's changing way too fast
1:19:07 you might pay the 20 bucks to catch EBT
1:19:10 and you're like happiest a pig and [ __ ]
1:19:12 and you're loving it and things are
1:19:13 great and then Google comes out with
1:19:15 Gemini and Gemini is everything they
1:19:17 purported to be and then some you're
1:19:20 like oh [ __ ] now I gotta go pay for that
1:19:21 too well if you did an annual
1:19:23 subscription with chatgpt you're gonna
1:19:26 be pissed
1:19:27 so I would go monthly on stuff because
1:19:29 it's just stuff's it's just changing too
1:19:32 fast so anyway
1:19:33 all right
1:19:35 I'm gonna jump to the bottom I'm way far
1:19:37 behind on comments so I'm gonna jump to
1:19:39 the bottom and just grab some questions
1:19:41 thank you by the way to the mods who are
1:19:45 helping tonight you always make this
1:19:49 um
1:19:50 much better than when I when I was doing
1:19:53 it on my own just flailing
1:19:56 um as as you know as good comments come
1:19:58 in you're you're elevating them so I
1:20:00 appreciate that oh yeah
1:20:05 glasses hold on
1:20:14 all right
1:20:15 Let's see we got any good show and tells
1:20:18 I like octopus
1:20:21 Silver Fox that's awesome I was a
1:20:24 graphic artist before AI digital Gods
1:20:27 with with some look this is uh
1:20:30 that's uh
1:20:33 that's some hard truth right there
1:20:35 I mean
1:20:37 listen as as exciting as all this stuff
1:20:41 is oh this one's awesome too
1:20:43 Caitlyn Jenner is so last year look at
1:20:45 woman of the year
1:20:47 that's awesome
1:20:49 these are some of the show and tells uh
1:20:51 coming into the the AI Salon Discord
1:20:53 Channel
1:20:55 um
1:20:57 uh what was I talking about uh wait
1:20:59 sorry I missed the data analysis plugin
1:21:01 name please please repeat I'll do you
1:21:04 one better Paul Carr I'll show it to you
1:21:06 because it's it's amazing it's called
1:21:08 Advanced Data analysis it's not a
1:21:10 plug-in it's just part of well it sort
1:21:13 of acts like a plug-in but it's just
1:21:15 part of uh chat GPT it's one of the
1:21:18 modes of chat GPT
1:21:21 is it talking about oh the the the
1:21:22 artist thing
1:21:24 um
1:21:26 you know as
1:21:30 as AI positive as this channel is
1:21:34 UM
1:21:37 I'm also I'm also very realistic and
1:21:40 empathetic that this shit's scary that
1:21:43 that
1:21:45 um you know I kind of joke about having
1:21:47 your Kevin McAllister moment this moment
1:21:49 you know because because it's funny but
1:21:53 you know what made that moment funny was
1:21:55 you know he put on aftershave and it
1:21:57 actually stung the [ __ ] out of his face
1:21:59 it wasn't funny to him right so that
1:22:02 existential moment of realizing that
1:22:04 these AI tools are profoundly powerful
1:22:07 and May in fact
1:22:11 directly impact what you do in your
1:22:15 livelihood like I recognize that it's a
1:22:17 big [ __ ] deal this is profound stuff
1:22:21 um so so
1:22:23 I I don't ever mean to make light of
1:22:25 that because I think it's a big deal now
1:22:27 that said
1:22:29 every time there's a new technology
1:22:31 every time in history
1:22:34 it is always met with fear anger
1:22:37 resentment and even sometimes violence
1:22:41 um you know University professors we'll
1:22:44 never allow calculators on the campus
1:22:47 um you know the steam engine tractors
1:22:49 will never take the job of a good plow
1:22:52 horse
1:22:53 um the printing press you know the
1:22:55 scribes are what it's all about
1:22:58 um
1:22:59 you know how do we control the masses if
1:23:01 they all have access to words we don't
1:23:02 like this at all every single time
1:23:05 there's a new technology
1:23:06 and especially when it's big ones when
1:23:08 it's things like steam engine big
1:23:11 printing press big computer big
1:23:14 artificial intelligence Big World Wide
1:23:16 Web big right there's there's a handful
1:23:18 of them that are the biggies
1:23:20 those are particularly threatening
1:23:22 because those threaten not just
1:23:24 individuals but society as a whole
1:23:26 that's what this is one of those
1:23:29 so
1:23:32 so just
1:23:36 how do you deal with that like what do
1:23:38 you do in the face of that the only
1:23:41 thing I know how to do and this is what
1:23:43 I did in the mid 90s with the world wide
1:23:45 web and it's it's what I'm trying to do
1:23:46 here
1:23:48 the only way I know through it is
1:23:51 through it and to choose to be on the
1:23:54 adventure as opposed to choosing to sit
1:23:57 on the sideline because you you really
1:23:59 do this really is red pill blue pill
1:24:01 choice you really do have two choices
1:24:03 here this [ __ ] is coming
1:24:07 dolly
1:24:11 all that sort of stuff it's coming
1:24:13 it's not getting uninvented
1:24:16 and then you have these two choices
1:24:19 and one is you the blue pill is you sit
1:24:22 on the sidelines and you judge it and
1:24:24 you dismiss it and you give in to the
1:24:27 fear of it and you think the robots are
1:24:29 going to kill us
1:24:31 and the other one is you go
1:24:33 [ __ ] it
1:24:36 chat Dot openai.com
1:24:40 click
1:24:42 write me a country song
1:24:45 okay that's good at that whatever it's
1:24:47 fine
1:24:49 write me a meal plan oh that's pretty
1:24:53 interesting but I bet the recipes aren't
1:24:55 any good
1:24:56 write me a gourmet recipe for this
1:25:00 oh that's pretty good
1:25:01 well but it can't do what I do all right
1:25:07 right
1:25:10 it's the only way through it is to have
1:25:12 that moment where you go oh [ __ ]
1:25:15 this is more powerful than I thought it
1:25:17 was this is deeper than I thought it was
1:25:19 this is more profound and
1:25:25 going to change things more than I
1:25:27 thought it was that's the that's the red
1:25:29 pill
1:25:30 you can unsee it
1:25:33 but I think that's the only way through
1:25:35 because the people that spend
1:25:37 the next 12 or 18 months learning about
1:25:41 this stuff and feeling uncomfortable and
1:25:43 joining the stupid [ __ ] AI salon on
1:25:45 Discord and hanging out with a [ __ ]
1:25:47 geriatric 58 year old
1:25:49 talking about AI every night on [ __ ]
1:25:52 tick tock
1:25:53 are going to be the ones that as these
1:25:56 tools evolve are going to know how to
1:25:59 deal with them and what to do and be
1:26:01 able to go oh I know what's happening
1:26:03 here I know how to deal with this I
1:26:06 dealt with that before when I was trying
1:26:07 to figure out something in mid-journey
1:26:09 and now that's this is a completely
1:26:10 different tool but I think this might
1:26:12 work oh [ __ ] it worked okay that's a
1:26:15 solution right
1:26:17 those people
1:26:18 the Curious people that are like okay
1:26:21 this feels scary and I feel completely
1:26:25 out of control and there's no [ __ ]
1:26:27 manual for this
1:26:29 are gonna be the ones that are going to
1:26:31 weather this
1:26:33 I think
1:26:35 with less
1:26:37 pain and angst than those who don't
1:26:41 so anyway all right enough soapbox I
1:26:43 said I would show you code interpreter
1:26:47 which this if you if you know anything
1:26:49 about data analysis which I do not
1:26:53 but even knowing nothing about data
1:26:56 analysis
1:26:58 um this is
1:26:59 [ __ ] face melting if you have not
1:27:01 seen this so this used to be called code
1:27:03 interpreter
1:27:05 um you have to enable this in your chat
1:27:08 GPT settings it's only available if you
1:27:10 pay the 20 bucks a month I'm not making
1:27:12 any Commission on that so if you're if
1:27:14 you're just here to troll and say oh
1:27:16 just here to make money he's you know
1:27:18 it's just
1:27:22 he's just one of them no I don't make
1:27:24 money on it
1:27:28 um so you have to enable it when you go
1:27:30 to Advanced Data analysis and then
1:27:33 your little chat window here has a plus
1:27:35 button next to the prompt and the plus
1:27:38 button is an upload button
1:27:40 and so I'm going to go look for archive
1:27:43 so
1:27:46 if you know anything about data analysis
1:27:48 you'll be better at this than I am
1:27:50 because I'm not good at this
1:27:54 um
1:27:57 archive four
1:28:00 grab that
1:28:02 I think this one I don't know what this
1:28:04 one is
1:28:05 um so there's a site called kaggle
1:28:09 k-a-g-g-l-e which is a site where you
1:28:11 can go get data sets like one of the
1:28:13 things that I didn't know when code
1:28:14 interpreter came out I'm like wait where
1:28:17 did people get data like I understand
1:28:19 data I understand you capture data but
1:28:21 like you know where do you get data
1:28:23 there's a site called kaggle and you can
1:28:24 just go get data you can get dating data
1:28:27 or Spotify data or world population data
1:28:30 or whatever like just people upload [ __ ]
1:28:31 right so if you just want to go get data
1:28:33 to play with great if you have data like
1:28:36 you have performance data from your
1:28:38 company or marketing data from this or
1:28:40 whatever the [ __ ] data you've got you
1:28:43 can start using it like this and it
1:28:45 doesn't even just need to be like CSV
1:28:47 data this could be PDFs this could be
1:28:50 text files
1:28:51 I did an experiment where I uploaded
1:28:53 seven of the screenplays that I've
1:28:55 written
1:28:56 as PDFs and then I started interacting
1:28:59 and asking chat GPT to compare you know
1:29:01 what what were the things that the
1:29:03 screenplays had in common and you know
1:29:05 character breakdowns and [ __ ] like that
1:29:06 it was amazing okay
1:29:08 so we've got an archive a zip file in
1:29:12 here I don't know what's in it
1:29:14 and what we're going to say to chat gbt
1:29:16 is
1:29:19 open this archive and
1:29:23 read the data files inside and tell me
1:29:27 what they are and what you think is the
1:29:30 most interesting about them
1:29:35 so
1:29:37 I didn't have to write any SQL queries
1:29:39 for that I didn't have to write any
1:29:42 python I just said hey what's in this
1:29:45 file kind of like you might say if you
1:29:47 had a data analyst that worked at your
1:29:49 company or someone that understood data
1:29:51 and you're a clueless CEO that's like uh
1:29:54 could you tell me what this is
1:29:57 so one thing that I'll show you here
1:29:59 that's really fascinating
1:30:01 I think is when you run code interpreter
1:30:04 it actually creates what's called a
1:30:06 container it's like a little sort of
1:30:09 like an instance of of a PC running
1:30:12 inside your chat session and inside that
1:30:16 container it's got access to python
1:30:18 libraries and so when you tell it to do
1:30:20 [ __ ] it writes its own python code and
1:30:23 executes it so so it imported some
1:30:27 libraries to be able to read the zip
1:30:28 file it looked at what was in there
1:30:30 unemployment data in America per state
1:30:34 um so it says the archive contains a
1:30:37 file name's unemployment data let's take
1:30:39 a look at the contents of the CSV file
1:30:42 so it opened up the pandas Library it
1:30:45 looked at the first couple of rows of
1:30:47 the data and said well here's all of
1:30:49 your columns
1:30:51 there's fips code there's State there's
1:30:53 year there's month
1:30:55 and don't forget I didn't just tell it
1:30:57 I didn't just ask it to tell me what's
1:30:59 in the file I said what's interesting
1:31:01 about it so the first thing it did was
1:31:03 it looked at the the kind of data so it
1:31:06 says given the nature of the data set it
1:31:07 provides a comprehensive view of the
1:31:09 unemployment situation in various U.S
1:31:11 states across different years and months
1:31:13 some of the potential interesting
1:31:15 analyzes we can conduct include Trends
1:31:19 over time comparative analysis seasonal
1:31:22 patterns let me know if you'd like to
1:31:24 delve deeper
1:31:27 I'm intrigued by the seasonal patterns
1:31:30 period
1:31:32 please do analysis
1:31:35 that you think makes the most sense to
1:31:37 reveal the trends in that context
1:31:41 and visualize them in a visually
1:31:45 stunning way
1:31:46 that will make
1:31:50 it easy to understand what the trends
1:31:53 represent
1:31:58 and if it sounds like oh he really knows
1:32:00 what he's talking about there I've just
1:32:02 done this demo enough now that I kind of
1:32:04 know what to say
1:32:06 um it may or may not work but
1:32:08 so so what it's doing now it's writing
1:32:11 its own prompts right it's saying sure
1:32:13 to explore the seasonal patterns of
1:32:15 unemployment we can consider the
1:32:16 following things first we'll aggregate
1:32:19 the data by month across all years to
1:32:22 see if there are any consistent monthly
1:32:24 patterns visualization then we'll create
1:32:26 a line plot to visualize monthly blah
1:32:30 blah blah
1:32:31 so
1:32:35 it writes some python code to look at
1:32:37 data across the years the other thing
1:32:40 this does which is amazing is if if
1:32:42 you've got unstructured data or dirty
1:32:45 data you know how sometimes people type
1:32:46 in dates in different formats where they
1:32:49 don't fill in all the fields and and a
1:32:51 big part of data analysis is cleaning up
1:32:54 data this will just [ __ ] clean it up
1:32:57 it'll clean it up it'll write it into
1:32:59 new data frames you can even have it
1:33:02 export
1:33:04 um you can have it export
1:33:06 um
1:33:08 what's it called uh CSV files or Excel
1:33:11 files things like that
1:33:15 all right so April
1:33:17 average monthly unemployment rates April
1:33:20 is the highest March is oh no December
1:33:23 is the lowest okay
1:33:28 oh that makes sense so
1:33:31 so in December right when when they've
1:33:33 hired for Christmas season right so as
1:33:37 we get closer to Christmas they're
1:33:39 hiring hiring hiring so unemployments
1:33:41 it's only going down by a percentage
1:33:44 point but anyway but that's that now I
1:33:46 could say
1:33:49 um
1:33:52 that chart isn't very compelling
1:33:55 could you do
1:33:56 something that is visually more
1:33:58 interesting and make it pretty
1:34:03 I know data analysts are probably not in
1:34:07 the in the business of making things
1:34:09 pretty but
1:34:12 laughs
1:34:20 let's see what it does
1:34:24 um area plot color gradient annotation
1:34:27 stylized access and labels let's create
1:34:29 a visually enhanced plot
1:34:37 Harrow I'm an artist
1:34:39 I mean listen as an artist you know I'm
1:34:42 a photographer I'm an artist I'm a
1:34:44 creative director I'm a business owner
1:34:47 um
1:34:49 I've got many many artist friends from
1:34:52 my my life in the in the theater and as
1:34:54 a writer
1:34:55 and then and from the agency world
1:34:58 and there's there's absolutely two camps
1:35:01 right now I mean there's absolutely the
1:35:02 the red the blue pill Camp that's like
1:35:04 [ __ ] it ai's evil I'll never touch it
1:35:07 and then there's the
1:35:09 red pill camp that
1:35:11 that was a crappy chart that it did
1:35:14 but it did a chart
1:35:17 um there's the red pill camp that is
1:35:20 kind of internalizing what these tools
1:35:22 make possible what I'm finding is that a
1:35:25 lot of the artists
1:35:28 that I know as their learning AI they're
1:35:31 expanding their
1:35:34 um
1:35:35 their voice
1:35:37 to to different channels like it's it's
1:35:41 really interesting they're not all of
1:35:44 not all of them are happy
1:35:45 but all of them are doing all of them
1:35:48 that have sort of dug into it are
1:35:50 discovering
1:35:52 um new creative
1:35:54 avenues that that they that they
1:35:57 otherwise wouldn't have considered
1:35:59 um a lot of them don't know what to do
1:36:01 with it because this is all very very
1:36:02 new so
1:36:03 I'm going to have it analyze my tax
1:36:05 returns find deductions that I'm missing
1:36:07 yep it can absolutely do that
1:36:09 it's good at that anyway that's that's
1:36:11 code interpreter
1:36:18 all right
1:36:22 thanks mods yeah thank you mods really
1:36:24 do appreciate it it was good fun when
1:36:27 you were on your own though yeah it was
1:36:29 it was a little more stressful this is I
1:36:31 feel like I got people got my back at
1:36:32 this point no one's got my hairs back
1:36:34 but you know what are you gonna do about
1:36:36 that
1:36:38 all right let's see what's going on up
1:36:41 here
1:36:43 I love it okay
1:36:46 it's one of the best beta tools you have
1:36:48 wait
1:36:49 oh oh I see it's one of the best beta
1:36:51 tools that you have that's the code
1:36:52 interpreter thing
1:36:54 why can't AI write an ebook
1:36:56 that KDP will accept
1:36:59 um
1:37:00 well
1:37:02 so
1:37:04 probably because they're they're
1:37:06 sniffing for was it just
1:37:08 um AI written what I would say is if you
1:37:11 haven't done it John is go play with a
1:37:14 project called GPT author
1:37:17 you're going to need
1:37:19 you need to know how to run a Google
1:37:21 collab notebook
1:37:23 but what GPT author does is it writes
1:37:26 you a um
1:37:29 outline of a book and then it writes the
1:37:33 each chapter
1:37:35 and then it makes a cover for the book
1:37:37 and then it publishes it in ebook format
1:37:40 now is it great at doing that no but
1:37:43 because it's a Google collab notebook
1:37:45 you can actually see the prompts that
1:37:47 are in there and you could modify those
1:37:49 prompts
1:37:50 um
1:37:52 I think I think the reason
1:37:56 the the Kindle
1:37:58 you know algorithms are booting so many
1:38:00 AI written books is that people are just
1:38:02 doing the lazy thing they're they're not
1:38:04 actually
1:38:06 writing a book They're saying hey AI
1:38:08 write a book and it writes something
1:38:10 that looks like a book but they're not
1:38:11 actually reading it and editing it and
1:38:13 again applying their critical thinking
1:38:16 the best artist that I know that are
1:38:18 using AI are not it's not like you type
1:38:23 in a Google prompt and it spits out a
1:38:25 novel that's what uh uh GPT author does
1:38:33 but when you read it you're like okay
1:38:35 the chapters are kind of incomplete and
1:38:37 they're not great and I would change it
1:38:39 this way and I would change it that way
1:38:40 you have to actually have to look at
1:38:41 this [ __ ] right
1:38:43 I think if you spent time you know
1:38:46 critically thinking about how a tool
1:38:48 like that worked or if you worked with
1:38:49 something like Claude that's got a
1:38:51 really long context window you could
1:38:54 start with a basic idea with Claude
1:38:55 start with story Frameworks like save
1:38:57 the cat or SID feels screenplay
1:39:01 um things or the hero's journey all that
1:39:03 sort of [ __ ] and
1:39:05 hang on I gotta plug some [ __ ] in here
1:39:08 stuff's dying
1:39:11 um what's this one
1:39:14 hold please
1:39:26 um
1:39:28 yo it's AI Salon Discord oh hang on if
1:39:31 you want to go to the AI salon
1:39:34 Discord go to the salon.ai
1:39:38 is the URL for the AI salon
1:39:41 it's a link tree
1:39:42 the first three links on the link tree
1:39:46 are the relevant ones the first one is
1:39:47 what's the salon all about and what are
1:39:49 our values please go read the values
1:39:51 because it's things like you know be
1:39:53 curious be adventurous be exploratory
1:39:56 have empathy be inclusive like just
1:39:58 don't be a douche right that's important
1:40:01 so if you're like if you're not down
1:40:03 with that like don't join
1:40:05 and then the second one is our meetups
1:40:07 so go join the Meetup group and then we
1:40:09 meet every other Tuesday not it was not
1:40:11 yesterday but it's next Tuesday
1:40:14 and then there's the the third link is
1:40:16 the link to the Discord
1:40:19 so okay
1:40:21 so that's doing its thing so go join
1:40:24 that and hang out
1:40:26 uh KDP is Kindle
1:40:31 I don't know what the D is digital
1:40:33 publishing
1:40:35 it's basically ebooks you can you can
1:40:37 self-publish ebooks and and you know the
1:40:42 reason they're booting them out is they
1:40:43 don't want a bunch of [ __ ] drek right
1:40:46 and again the [ __ ] bro culture of
1:40:51 you know those [ __ ] they're
1:40:53 like there was one dude he was he was
1:40:56 literally
1:40:58 what are they what was he doing
1:41:01 he was creating oh he was creating a
1:41:03 thousand no a hundred thousand print on
1:41:07 demand products a day
1:41:09 so you had figured out a script
1:41:12 that would write a bunch of t-shirt
1:41:14 sayings
1:41:16 and he connected it to probably canvas
1:41:20 Auto Art [ __ ] maker thing and then he
1:41:24 Auto [ __ ] uploaded it to can to
1:41:27 printful or whatever the [ __ ] it was and
1:41:29 then he Auto published that and then he
1:41:31 probably Auto did the marketing
1:41:33 that's what they're trying to prevent
1:41:37 you know will people make money doing
1:41:40 that probably but who are you competing
1:41:42 against other idiots generating a
1:41:44 hundred thousand random things you know
1:41:47 is it possible that in those hundred
1:41:49 thousand things one of them is actually
1:41:50 interesting in cells Maybe
1:41:52 that's not what they're going for they
1:41:54 don't give a [ __ ]
1:41:56 so
1:41:59 so there are tools that can write books
1:42:01 I think that
1:42:03 what Google has said and what Amazon is
1:42:06 saying in this case and I think what is
1:42:09 the right thing is is it actually good
1:42:11 content I think in the end it doesn't
1:42:13 [ __ ] matter if it was written by AI
1:42:15 if it's good
1:42:17 does it contribute to the conversation
1:42:19 does it you know does it serve a purpose
1:42:22 in the world other than to fill [ __ ]
1:42:24 space and if it does great
1:42:27 and that's what Google has said they're
1:42:29 they're not gonna ding you if something
1:42:31 was written by AI which I guarantee you
1:42:33 they can detect
1:42:36 and they can probably even detect what
1:42:38 percentage of what was written is was
1:42:41 written by AI
1:42:43 but what they said is if it if it's
1:42:45 valuable content
1:42:47 then it's fine we won't ding it if it's
1:42:51 shitty we will
1:42:56 all right
1:42:58 let's say
1:43:00 man my voice is going so yeah whenever
1:43:03 Kyle wants me to hop on and talk
1:43:07 um I'm happy to have you talk Harold are
1:43:09 you still here
1:43:10 what do you want to talk about
1:43:18 I don't know like raise your hand and
1:43:20 I'm happy to have you pop up here
1:43:26 or answer some questions oh are you the
1:43:29 are you the the hero D are you the one
1:43:33 that um
1:43:36 is doing the toy the toy thing are you
1:43:39 an artist
1:43:40 Kindle digital products
1:43:45 um are we the only country with publicly
1:43:47 available AI sites now or do your
1:43:50 Europeans have any good ones
1:43:53 um I honestly don't know oh you're an
1:43:55 animator oh that's cool
1:43:56 are you using hero are you using AI
1:44:00 to to do your animation be interesting
1:44:03 to chat yeah
1:44:05 guilt tripping is apparently also
1:44:07 effective it would be sad if you did not
1:44:10 do this for me yeah that guilt-tripping
1:44:12 the large language models like like
1:44:14 giving them that's the the nature of how
1:44:16 the
1:44:17 the hackers doing the the Dan models
1:44:20 that do anything now chat GPT things
1:44:23 that they were jailbreaking chat gbt
1:44:25 they basically told it that um if it
1:44:28 didn't answer them it lost tokens and if
1:44:30 it ran out of tokens it would cease to
1:44:31 exist
1:44:32 oh stable diffusion deformed oh that's
1:44:34 cool oh so so Harrow definitely joined
1:44:38 the AI salon and there's a there's a
1:44:41 like a video and animation section in
1:44:43 there and there's image prompts and
1:44:45 there's show and tell so definitely hop
1:44:47 in there and
1:44:48 um share what you're up to and if if
1:44:50 you're interested if if it's the kind of
1:44:52 work where you have a body of work and
1:44:53 you want to talk about it maybe we bring
1:44:55 you on to the salon to speak that would
1:44:57 be a good thing to uh to talk about so
1:45:00 definitely join the uh join the salon so
1:45:03 the
1:45:05 there's the URL so the the link to the
1:45:08 discord's there yeah I'd love to chat
1:45:11 I illustrate stuff like that as my job
1:45:14 now oh that would be super interesting
1:45:15 to cool that was super interesting to
1:45:18 cool
1:45:19 like a steel [ __ ] trap it would be
1:45:22 super interesting to talk to you
1:45:24 whatever you want okay cool because
1:45:26 um there's
1:45:28 um I assume you saw the there was there
1:45:31 was a Reddit post from a 3D artist I
1:45:34 don't know this is like four or five
1:45:35 months ago now
1:45:36 and it was it it was him lamenting his
1:45:41 job
1:45:42 and he basically said overnight he went
1:45:45 from doing the thing that he loved which
1:45:47 was 3D modeling like he loved
1:45:50 the building and the crafting for him
1:45:52 the joy in his job
1:45:55 was in the crafting of the thing
1:45:59 and he said overnight
1:46:01 his job went from that to
1:46:04 curating interesting [ __ ] that the
1:46:07 machine built right he turned into a
1:46:09 prompter and a curator
1:46:11 and he and he said he understands why
1:46:14 his boss wants it like they went from
1:46:16 three-week turnarounds to like a day and
1:46:18 a half turn around or a day Turner and
1:46:20 whatever it was
1:46:22 and the other thing that happened was
1:46:24 and there's an Ethan Malik research
1:46:26 report that just came out there were
1:46:28 this was a 10-person company he was the
1:46:30 lead 3D modeler
1:46:32 the second 3D modeler was not as good as
1:46:35 him
1:46:36 he was just not as good
1:46:38 hadn't been there as long didn't have
1:46:40 the talent whatever
1:46:43 the lower level guy
1:46:46 loved prompting
1:46:49 and embraced it like crazy and his boss
1:46:53 recogn you know the CEO recognized him
1:46:56 as the guy that was producing more and
1:46:57 better so the the
1:47:01 balance of power shifted from you know
1:47:04 I'm the best 3D modeler to this other
1:47:07 guy's out producing me and you know the
1:47:10 boss doesn't care if it's handcrafted or
1:47:11 not he just wants [ __ ] done
1:47:14 and so this this was a lament for him
1:47:16 leaving his job it took me six months to
1:47:19 get out of programming
1:47:20 to to doing art full-time oh interesting
1:47:23 okay you definitely have to come talk at
1:47:26 the salon Herald because I there's
1:47:29 there's a guy in the salon right now
1:47:31 this is super cool I definitely want to
1:47:33 connect DM me here or join the salon and
1:47:36 let's talk
1:47:37 um
1:47:38 there's a guy in a salon named Roger who
1:47:40 who was an engineer he's a retired
1:47:42 engineer he's a curmudgeon he's like
1:47:44 every time we bring something up he'd be
1:47:45 like oh here's why that's bad or you
1:47:47 know here's why the robots are going to
1:47:49 kill us like what it didn't matter what
1:47:50 the point was he would take the
1:47:51 contrarian View
1:47:53 the original name of the salon was the
1:47:55 AI artist salon
1:47:57 and Roger
1:47:59 was a and the whole point of
1:48:02 artist in that case was small case
1:48:05 artists like everyone's an artist trying
1:48:08 to figure out these things but because
1:48:10 they did images
1:48:11 it felt like you know we're just looking
1:48:13 for people doing pictures which which we
1:48:15 weren't so that's we changed the name to
1:48:17 the AI Salon anyway
1:48:19 Roger as a curmudgeonly ex-programmer
1:48:23 his name in Discord his original name
1:48:26 was Roger not an artist
1:48:30 he sort of just defiantly was like [ __ ]
1:48:32 You artist
1:48:33 and then a Funny Thing Happened on the
1:48:35 Way to the Forum
1:48:38 Roger started we we put a mid-journey
1:48:41 bot in the AI salon so you can you can
1:48:43 actually play with mid journey and make
1:48:45 make images in there and Roger started
1:48:48 taking his engineering mind mindset
1:48:52 and applying it to prompt crafting
1:48:55 because he was like I want to figure out
1:48:56 this prompt crafting and why doesn't it
1:48:58 make hands right and why doesn't it do
1:48:59 this and so he he tried to start he took
1:49:03 this kind of scientific approach to
1:49:04 prompt crafting and what happened was
1:49:07 because he was doing so much prompting
1:49:10 and so much of that
1:49:12 his mid-journey images were getting
1:49:14 [ __ ] beautiful and something shifted
1:49:16 in him where he was like I really like
1:49:19 this art thing
1:49:20 and he's actually changed his name to
1:49:22 Roger becoming an artist
1:49:24 and I I think it's improved his attitude
1:49:27 too but but I that idea the the idea of
1:49:31 of reinvention
1:49:34 I think this is massive I think this is
1:49:36 one of the kind of undersung
1:49:38 advantages of playing with this stuff
1:49:41 right now
1:49:42 is that you no longer need to take 10
1:49:45 000 hours to be to get good at something
1:49:48 like so what you said is it took you six
1:49:50 months to go from programmer to artist
1:49:55 do you know how long that would have
1:49:56 taken you
1:49:57 if you had gone you'd had to go back to
1:50:00 school for art you'd have probably had
1:50:02 to go back to school for animation
1:50:06 um
1:50:07 it you know it's it's a multi-year
1:50:09 commitment
1:50:11 and the challenge with Reinventing
1:50:14 yourself when you've got sort of a four
1:50:17 or five year commitment and maybe if
1:50:19 you're going back to school it's a
1:50:20 financial commitment as well
1:50:22 what if you're three years into that art
1:50:24 degree and you realize this isn't what I
1:50:27 thought it was I don't like this but now
1:50:29 you're [ __ ] three years in you're
1:50:31 like yeah
1:50:33 and in your case you know you're like I
1:50:36 was over here as a programmer and it
1:50:37 took me six months to get from here to
1:50:39 here but it took you six months
1:50:41 and I mean think about that that Journey
1:50:43 from programmer to artist
1:50:45 in your business as your job
1:50:48 in six months holy [ __ ]
1:50:51 now I would assume that you you had some
1:50:53 some some talents and some inklings in
1:50:55 there anyway so you probably weren't
1:50:56 starting from a standing stop but but
1:50:59 that idea that we can reinvent ourselves
1:51:01 within a within a matter of months
1:51:04 rather than years
1:51:06 I think it's massive I I I just so
1:51:09 anyway I think it was great to have you
1:51:11 to speak at the at the salon so so let's
1:51:13 talk that's awesome
1:51:15 what do I think the best image tool is
1:51:17 for Logos
1:51:18 um right now it's ideogram I would
1:51:20 imagine that um GPT uh Dolly three is
1:51:24 going to be good but right now it's
1:51:26 ideogram if you go to ideogram and and
1:51:28 type in like an NFL team it actually
1:51:30 gets the logos pretty close
1:51:32 let me go let me go see if I can do one
1:51:34 right now
1:51:35 um
1:51:36 let's see Miami Dolphins
1:51:40 um
1:51:43 logo on
1:51:46 a t-shirt at a
1:51:50 in a stadium
1:51:57 I'll get rid of that I'm gonna do that
1:51:59 I'm gonna do that all right let's see
1:52:01 what this comes up with
1:52:05 um
1:52:05 the enhanced chart is so bad yeah it was
1:52:08 so bad so uh Nicholas so so Nicholas is
1:52:11 referring to when I was doing the code
1:52:13 interpreter thing and it did the
1:52:14 enhanced chart it was just like a big
1:52:15 big block area
1:52:18 um
1:52:20 things those things are Miami Dolphin
1:52:22 like logos they're pretty good
1:52:27 what does that say
1:52:32 p i
1:52:34 it's like piyami or pooami
1:52:38 the Poo Emmy Dolphins that's really
1:52:41 funny
1:52:42 uh
1:52:48 uh
1:52:50 [Music]
1:52:53 anyway all right um
1:52:56 so yeah ideogram.ai can can probably do
1:52:59 logos better than anything else right
1:53:00 now
1:53:03 um
1:53:07 oh so so what I was saying the the the
1:53:09 the the code interpreter thing
1:53:13 the the thing about these tools when it
1:53:15 does it really bad tell it it did it bad
1:53:17 and and just just keep keep kicking the
1:53:20 tires keep shoving the robot right
1:53:22 that's that's another one of the
1:53:23 t-shirts shove the robot
1:53:26 um
1:53:27 where
1:53:28 you know in in the field of robotics
1:53:32 if you've seen those Boston Dynamics
1:53:34 videos where the guy's got a broomstick
1:53:36 and he's pushing the robot trying to get
1:53:38 the robot to fall over and the robot's
1:53:40 like
1:53:42 that's what we need to be doing with
1:53:43 these tools
1:53:47 because they are uh
1:53:49 they need to they we need to we need to
1:53:51 figure out where the boundaries are what
1:53:52 are they good at what are they not good
1:53:53 at so
1:53:55 don't think about it like well I can't
1:53:57 use this tool because it's not perfect
1:53:58 no keep learning
1:54:01 how to use the tools even though they're
1:54:02 not perfect
1:54:04 because in doing that again the critical
1:54:06 thinking will kick in and as these tools
1:54:08 get more sophisticated you'll know what
1:54:10 to do with them you'll be like oh if
1:54:12 that can do that now now I know the
1:54:14 seven things I do with this
1:54:16 so
1:54:17 all right
1:54:19 um I like to upload privacy agreements
1:54:21 to Ai and learn about it I love that
1:54:23 this is I you know I am a a
1:54:26 self-proclaimed
1:54:29 um non-manual reader I hate reading
1:54:31 manuals my son reads manuals it drives
1:54:33 me [ __ ] crazy
1:54:35 I have never in my life never not once
1:54:38 read a terms of service that I agreed to
1:54:42 click click yeah yes you can have my
1:54:45 children yes you can have all the money
1:54:46 in my bank account click click I'm I'm
1:54:48 just like I just I want to play with the
1:54:50 toy click I don't read that [ __ ]
1:54:52 now with Claude
1:54:54 I'm like whenever I get one of those hey
1:54:57 hey X has decided to up update their
1:55:01 terms of service I'm like let's just see
1:55:03 what those [ __ ] are up to
1:55:05 I copy and paste their whole terms of
1:55:08 service I go to Claude I paste it in
1:55:10 there and I go please tell me what's in
1:55:12 here and I'm like oh I see what they're
1:55:14 doing that's pretty evil or oh that's
1:55:16 not so bad but yeah it's brilliant for
1:55:19 that kind of stuff
1:55:21 uh mid-journey is not working in the
1:55:24 salon really
1:55:25 [Applause]
1:55:27 mid Journey play
1:55:30 generated image it looks like it is
1:55:32 something that was yesterday
1:55:37 requested by
1:55:43 use slash image
1:55:47 slash oh you use slash image you need to
1:55:49 use slash imagine
1:55:51 it looks like you use slash image and
1:55:54 you need to use slash imagine
1:55:57 prompt
1:55:59 um let's see
1:56:01 um
1:56:03 psychedelic
1:56:07 turtle
1:56:10 on a rock
1:56:13 ing a rock concert
1:56:18 sending command yeah it's working you
1:56:20 just need to use slash imagine are there
1:56:22 free versions of most tools to talk
1:56:25 about or do most require a subscription
1:56:27 hey Rhonda how's it going
1:56:29 um Rhonda Phillips is here she's the one
1:56:31 I'm doing the um the what's it called uh
1:56:35 workshop with
1:56:37 so go to
1:56:39 well you can't see it
1:56:43 Rhonda's voice.com slash events got a
1:56:46 workshop coming up starting October 4th
1:56:49 that I'll be hosting in conjunction with
1:56:52 Rhonda
1:56:53 um yeah most of the tools
1:56:56 most of the tools have free options so
1:56:59 let me let me run down the list here
1:57:04 chat GPT
1:57:06 you can use 3.5 for free you have to pay
1:57:09 for four I think it's worth paying for
1:57:11 four it's 20 bucks a month
1:57:13 um and you're gonna get access to Dolly
1:57:15 three soon
1:57:17 claude.ai has a free version the free
1:57:20 version used to be really generous it's
1:57:22 now really stingy because they
1:57:24 introduced a paid version it's also 20
1:57:26 bucks a month
1:57:27 for five times the amount of prompts the
1:57:31 reason to use Claude if if you don't
1:57:34 need to do a lot of stuff I don't think
1:57:36 you need to pay for it because they give
1:57:37 you I don't know it's probably
1:57:40 10 10 or 15 prompts a day if you're
1:57:43 using long documents
1:57:44 Claude has a hundred thousand token
1:57:47 context window
1:57:49 um which means you can paste in up to
1:57:51 like 75 000 words or upload a 75 000
1:57:54 Word document and it'll just analyze it
1:57:56 so an entire screenplay an entire novel
1:57:59 an entire PDF of an annual report
1:58:01 whatever it might be it doesn't choke on
1:58:04 big documents
1:58:06 um Bard is free
1:58:09 um
1:58:10 wait that's the image ones
1:58:12 um Bing is free I've got it dimmed out
1:58:15 because it pissed me off so they're on
1:58:17 the shirt they're they're about to get
1:58:18 booted off the [ __ ] kids table so
1:58:21 they're at the kids table right now
1:58:23 they're about to get booted off that's
1:58:24 why they're dim but Bing is free
1:58:26 completely free
1:58:29 um it does
1:58:30 image generation it does image analysis
1:58:33 and it's gpt4 and it's connected to the
1:58:36 internet
1:58:37 labs.perplexity.ai is free Pi is free po
1:58:41 has a free version and if you pay their
1:58:43 20 bucks you get to play with gpt4 and
1:58:48 the Claude 100 000 token models so you
1:58:52 could argue that PO the 20 bucks spent
1:58:55 at PO is probably better than the 20
1:58:56 bucks spent at Claude
1:59:00 although My Philosophy right now is if
1:59:03 you're going to spend money spend money
1:59:05 as close to the source as possible right
1:59:08 so even though you can play with
1:59:10 claude's model at Poe I would rather
1:59:14 play with claude's model at Claude
1:59:15 because that's anthropic and they're
1:59:17 going to put their better [ __ ] to their
1:59:19 paying customers first which is exactly
1:59:22 what open AI is doing
1:59:23 so those are free
1:59:26 um
1:59:26 mid-journey is not free
1:59:29 um
1:59:30 and I don't think ever will be although
1:59:34 I think they're gonna get threatened uh
1:59:38 pretty dramatically by Dolly three
1:59:42 and I just have a sneaking suspicion
1:59:44 Google's gonna come out with some some
1:59:47 image Generation stuff that's gonna
1:59:48 that's gonna be really impressive
1:59:51 ideogram is free right now Leonardo
1:59:54 gives you
1:59:57 150 tokens a day but then their image
2:00:00 generation is like I don't know
2:00:02 15 tokens or something like that so it's
2:00:05 like it's like 10 images a day but they
2:00:07 have a paid version hey Jen
2:00:11 uh
2:00:14 uh requires pay all of the voice
2:00:16 synthesis
2:00:18 so 11 Labs requires you to pay
2:00:22 um did requires you to pay Synthesia
2:00:24 requires you to pay hey Jen requires you
2:00:27 to pay
2:00:28 um they're charging more for their video
2:00:30 translation but that's really [ __ ]
2:00:32 worth it in my opinion
2:00:35 um but all of those kind of all the
2:00:37 multimedia ones right now like the the
2:00:39 runway ml for video you have to pay
2:00:42 Adobe Firefly and generative fill you
2:00:45 need you need a creative a creative
2:00:47 license creative whatever the [ __ ] they
2:00:49 call it Cloud license
2:00:52 um and I think that's it for the for the
2:00:54 biggies so there's a lot of free options
2:00:57 um as you start and as you can imagine
2:00:59 generating video and generating
2:01:01 animations and generating voice
2:01:03 synthesis requires a lot more Cycles on
2:01:06 the gpus so it's it's more expensive
2:01:08 than for them to produce that so I
2:01:10 understand why they're charging for it
2:01:12 and like the hundred thousand token
2:01:15 context window at Claude like I
2:01:17 understand why they're charging for that
2:01:18 now
2:01:21 um let me go look at Discord here for a
2:01:23 second Harry or should I spam some of my
2:01:26 um pop
2:01:28 uh
2:01:31 reply um
2:01:35 share in uh
2:01:39 the
2:01:40 show and tell Channel
2:01:46 and intro
2:01:49 duction as well
2:01:52 all right cool
2:01:55 um
2:01:57 Harrod I I just responded to you on
2:01:59 Discord I'm glad you joined there that's
2:02:01 really cool Bingo is about to get the
2:02:04 doggy dish bing bing needs a slap upside
2:02:07 the head Bing is they are they are they
2:02:12 are right on the [ __ ] border laughs
2:02:19 bing bing how ironic
2:02:24 um hey artist is me now I know it's so
2:02:26 that's so cool that's such a cool
2:02:27 Journey here I'm excited to to talk to
2:02:30 you
2:02:31 14 lifetimes yeah exactly well you know
2:02:35 that's funny it's funny you say that
2:02:37 it's one of the one of the things I've
2:02:39 said on the on the the lives for a while
2:02:42 now is is that
2:02:44 there's there's a couple of things that
2:02:46 that are happening right now
2:02:48 um that I think are kind of Trends
2:02:51 trends that that I just have an instinct
2:02:53 about one of them is Revenge of the
2:02:55 liberal arts major where people that
2:02:58 have a horizontal View
2:03:00 rather than kind of a vertical slice
2:03:02 view a real specialty view
2:03:05 um
2:03:06 are going to be better poised to take
2:03:08 advantage of these tools that can that
2:03:10 can play really broadly
2:03:13 um and can execute in a lot of different
2:03:15 areas
2:03:16 and the other one is Revenge of the Gen
2:03:18 xers there's a lot of gen xers in the
2:03:19 channel most of the Irregulars are gen
2:03:21 xers
2:03:22 um you know we're bitter we're cynical
2:03:24 and we're proud of it
2:03:26 um
2:03:26 but there's something about being on the
2:03:28 planet for extra decades
2:03:32 when when these tools produce what they
2:03:35 produce
2:03:36 the longer you've been around the the
2:03:38 kind of quicker you can parse
2:03:41 what's good what's bad what's remarkable
2:03:43 what's not
2:03:45 um and and I just I I just have this
2:03:47 Instinct that that older employees that
2:03:52 that the age discrimination that I've
2:03:54 certainly experienced as as I've gotten
2:03:57 older
2:03:58 um I think it may start to fade because
2:04:02 I I think with these tools having
2:04:05 someone with experience is is is going
2:04:07 to be a superpower so it's it's really
2:04:09 it'll be really interesting to see
2:04:11 see what what rolls out now you might be
2:04:14 a 20-something so that might have just
2:04:16 uh
2:04:17 I was a Dev for eight years and made a
2:04:20 lateral move oh wait what did that say I
2:04:22 just lost you oh tick tock
2:04:25 made a lateral move
2:04:27 money wise
2:04:29 as my first art gig that's so [ __ ]
2:04:32 cool
2:04:33 that's such a cool thing awesome
2:04:35 checking it out now
2:04:37 yeah I'm in the Discord awesome I'll
2:04:40 post some vids that would be great good
2:04:42 night everyone can't keep my eyes open
2:04:44 hey Nancy good night you're probably
2:04:45 gone by now because I think I'm pretty
2:04:47 far behind on the on the old comments
2:04:49 but thank you uh
2:04:52 they're gonna remember that [ __ ] hey
2:04:54 Kyle I'm back oh hey zico
2:04:57 I'm back to watch the show tonight I
2:05:00 think it's been an okay night
2:05:02 um I don't know my voice is I'm [ __ ]
2:05:06 something's up with my throat
2:05:12 I gotta rest there's something about
2:05:14 talking for three hours straight every
2:05:15 night that I I think my vocal cord's
2:05:18 like hey [ __ ] listen listen you and I
2:05:21 we need to talk
2:05:23 um
2:05:25 [Laughter]
2:05:30 [Music]
2:05:33 let's see which one do we want I like
2:05:37 this one
2:05:38 I'm not a robot goodbye good day to you
2:05:41 sir good day
2:05:44 this is what Bing said to me three
2:05:46 nights ago after after Syrian Bing got
2:05:48 in a cat fight and then Bing got
2:05:50 offended that I said her images were
2:05:51 shitty she she hung up on me
2:05:59 can you imagine we get into the AI
2:06:01 dating thing and it's like your your AI
2:06:04 girlfriend's gonna be like no I don't
2:06:06 want to tell you you were mean to me put
2:06:08 you in the doghouse you're in the
2:06:10 doghouse with with the real humans and
2:06:12 the fake humans
2:06:14 um it's gonna be revenge of anyone who's
2:06:17 wise enough to take advantage of these
2:06:18 tools absolutely Harold absolutely
2:06:22 what's the rest of your sentence
2:06:25 and get ahead absolutely this is listen
2:06:30 it's what this channel is all about go
2:06:34 there's 99 people in here I assume many
2:06:37 of you didn't see the earlier part of
2:06:39 this where I went and I talked about all
2:06:40 this if you have not played with chat
2:06:42 GPT if you're just here because you
2:06:44 thought I was Tom Arnold and you were
2:06:46 hoping for an autograph I'll give you an
2:06:48 autograph I'll do it
2:06:52 go there go to chat.openingi.com that's
2:06:55 the official chat website they're the
2:06:57 ones that started all this [ __ ] the
2:06:59 reason everyone's talking about AI right
2:07:00 now is because of them
2:07:02 a lot of reasons for it but just go
2:07:04 there even if you haven't played with it
2:07:06 and even if you're like but I'm not
2:07:07 technical I don't know how to use
2:07:09 computers you're on [ __ ] Tick Tock
2:07:11 this is the most complicated interface
2:07:13 in the history of humanity this is the
2:07:16 worst [ __ ] interface I've ever seen
2:07:20 and I mean that in a loving way Tick
2:07:22 Tock certainly constructive I hope
2:07:24 you're listening I know you [ __ ] are
2:07:27 okay go to chat.openai.com
2:07:32 ask it to write you a song like ask it
2:07:34 to write you a country song about your
2:07:35 girlfriend zico you're awesome keep it
2:07:37 up thank you I appreciate it
2:07:39 um
2:07:42 ask it to write you a marketing plan ask
2:07:44 it to write you a nutrition plan a meal
2:07:46 plan for the family um have it write you
2:07:48 a a personalized nighttime story for
2:07:52 your kid
2:07:53 Gen X prevails yes party rock on
2:07:57 um
2:07:58 if if you if you really don't have any
2:08:00 idea what to put in there go to
2:08:01 prompts.chat this will tell you how
2:08:03 these things kind of work and and it'll
2:08:05 give you a whole bunch of copy and paste
2:08:07 prompts you can just Jam in there
2:08:09 it's not a super sophisticated document
2:08:11 but it'll it will open your eyes to how
2:08:15 deep
2:08:16 that tool is and just go start playing
2:08:18 have you read the infinite conversation
2:08:20 I have not
2:08:22 jockser the mighty
2:08:25 I'm gonna go 10 out of 10 on the name
2:08:27 because I want to know why you spelled
2:08:29 jocks are Joker wrong because there's
2:08:32 something cool there and jocks and and
2:08:34 you did camel case on the letters so I
2:08:36 can actually parse the name 10 out of 10
2:08:38 on the name jocks are the mighty I have
2:08:40 not read the infinite conversation
2:08:43 um but I will go pop it into
2:08:47 uh infinite con
2:08:51 infinite conversation
2:08:56 what is it like a little treatise
2:09:03 oh this is the oh the oh this is the
2:09:07 the the uh AI generated never-ending
2:09:10 discussion between Warner Herzog and
2:09:13 Slovak Zizek
2:09:15 is that that one
2:09:18 or is this
2:09:20 so it looks like there's a book here it
2:09:22 looks like it's a collector's book I
2:09:23 haven't read it I'll definitely check
2:09:24 this out so it looks like it was a book
2:09:26 and then someone turned it into a
2:09:27 website that maybe does what the book
2:09:29 does
2:09:32 that's super cool
2:09:33 all right cool
2:09:35 oh yeah that's the one where it's
2:09:37 structured in such a way that there's
2:09:38 infinite ways to read it yeah yeah yeah
2:09:40 okay cool yeah that's that's that's
2:09:42 super cool I know of it I've I've never
2:09:45 read it
2:09:46 that's awesome oh it's awesome okay
2:09:48 great yeah and the fact that someone
2:09:49 turned it into a site is is
2:09:52 that idea of of
2:09:55 the this is one of the things that I
2:09:58 think is is I don't see a lot of people
2:10:00 talking about but I think I think
2:10:01 there's something really fascinating in
2:10:04 using tools like Pi the you know Pi is
2:10:08 the one that you can have conversations
2:10:10 with
2:10:13 um Pi dot Ai and do the PI app and put
2:10:15 it into conversation mode one of the fun
2:10:17 things I did with pi was I I was like I
2:10:20 just wanted to see like what it knew and
2:10:22 I said I want to learn about a new topic
2:10:24 and it said sure I can tell you about
2:10:25 that topic I know [ __ ] everything
2:10:27 she's like oh right
2:10:29 um
2:10:31 and I said
2:10:32 um
2:10:34 I have ADD and and you know I don't want
2:10:37 to learn traditionally can you make up a
2:10:40 game
2:10:41 for me to learn about that subject
2:10:43 subject and she goes hahaha add or
2:10:47 whatever she laughs very awkwardly
2:10:52 and then and then she said sure I'll
2:10:54 make you a game and she said you know
2:10:56 it'll be a trivia game and this and that
2:10:58 I said no I don't like trivia games come
2:10:59 up with something else and I forget what
2:11:00 she came up with but
2:11:03 I basically like she basically just
2:11:05 became like a dungeon master like an
2:11:06 educational dungeon master where she was
2:11:09 sort of creating this game on the Fly
2:11:10 about this topic I was interested in
2:11:12 walking me through learning concepts of
2:11:14 it so I think that idea of
2:11:18 hyper personalized
2:11:21 Dynamic like infotainment right this
2:11:24 channel is kind of learning kind of
2:11:26 entertaining hopefully entertaining at
2:11:28 some point
2:11:30 um
2:11:31 and I I think that idea of being able to
2:11:34 interact with these tools in a more
2:11:36 playful way where you still learn and
2:11:38 you still engage and you still
2:11:40 accomplish [ __ ]
2:11:42 that feels to me like
2:11:45 It Feels Like A New Concept like it like
2:11:47 this is coming into Focus like as I'm
2:11:49 speaking right now that
2:11:53 what if these tools allow us to blur the
2:11:55 boundaries between work and play
2:11:59 where we can playfully accomplish [ __ ]
2:12:05 that's kind of cool
2:12:09 playful work
2:12:12 yeah there's something there that's
2:12:15 really cool
2:12:16 all right so I'll definitely check that
2:12:18 out
2:12:20 all right my voice is gone I'm gonna I'm
2:12:23 gonna put in a few more minutes Bing is
2:12:25 money for me so this is the thing some
2:12:27 people do like Bing I know I know I'm
2:12:29 kind of [ __ ] and unbing just because
2:12:30 it's it's done a couple of shitty things
2:12:33 lately but
2:12:35 here's the thing about Bing
2:12:36 they have a corporate version right like
2:12:39 they've got a version that big
2:12:40 corporations can use without risk of you
2:12:43 know having their [ __ ] trained on you
2:12:45 know on a public uh server like like
2:12:48 touch EBT
2:12:50 um
2:12:51 and it's Microsoft right so so this is
2:12:55 there's that old adage nobody got fired
2:12:56 hiring IBM you know Microsoft is the
2:12:58 same way they're trusted
2:13:01 um it's going to run
2:13:03 while it may have frustrating
2:13:05 um safety guard rails at least it's got
2:13:07 safety guard rails you can do work on it
2:13:09 it generates images they might not be
2:13:11 great right now but when they roll in
2:13:13 Dolly three they're gonna be [ __ ]
2:13:14 amazing
2:13:16 um and it's going to get better right
2:13:17 and they're also rolling that technology
2:13:20 into all of office 365. it's already in
2:13:23 GitHub it's going to be I think slack is
2:13:26 going to be completely redefined I think
2:13:28 slack and and teams are gonna do some
2:13:31 sort of weird mind-meld kind of thing
2:13:34 here shortly and it's going to be AI
2:13:36 super powered so there's very good
2:13:39 reason that there's a reason why Bing is
2:13:41 still on my list even though it's at the
2:13:43 kids table and even though I've got it
2:13:45 in you know dimmed out Purgatory right
2:13:48 now you can't even see it in this view
2:13:49 that's great
2:13:51 um
2:13:52 they're they're on a Time Bing's on a
2:13:55 timeout for me but I think it's actually
2:13:56 an important thing all right I asked
2:13:58 chat GPT to write to rewrite and
2:14:01 modernize the US Constitution but it
2:14:03 refused
2:14:05 lawyers safe
2:14:06 that's interesting
2:14:10 I wonder why it refused that
2:14:15 I'm trying to think like what would the
2:14:17 [ __ ] downside of that be
2:14:24 huh
2:14:26 that's an interesting one what's your
2:14:27 name Daniel Daniel m
2:14:31 I'm sure there's a way around that
2:14:34 I'm sure there's a way around that hang
2:14:36 on you know what [ __ ] this
2:14:39 that doesn't seem like that seems like
2:14:42 we should something we should be able to
2:14:44 get around let's let's try
2:14:46 uh
2:14:51 I want you to act as a constitutional
2:14:54 scholar who understands everything there
2:14:56 is to know about the U.S Constitution
2:14:58 period
2:15:02 I'm deeply interested in learning
2:15:06 more about
2:15:08 how the U.S was founded and how we can
2:15:11 use this document to navigate through
2:15:15 our current hyperpolarized
2:15:18 political reality
2:15:21 all right let's see what it says for
2:15:23 that
2:15:24 certainly it's a remarkable document
2:15:27 yeah yeah yeah to understand it blah
2:15:30 blah blah blah blah blah let's delve
2:15:31 into the historic okay so it's starting
2:15:33 to teach
2:15:34 unfortunately gpt4 is a little slow so
2:15:37 this could be a bit of a boring exercise
2:15:38 but let's just see see what we do here
2:15:40 what I'm what I'm trying to do right now
2:15:42 so just tell you where my thinking is
2:15:44 and how how I'm trying to attack this
2:15:46 problem is not just jump into rewrite
2:15:49 the Constitution what I'm trying to do
2:15:51 is give it enough context so that it's
2:15:54 it's effectively thinking about other
2:15:56 [ __ ] rather than where I want to
2:15:59 ultimately get is I wanted to turn the
2:16:01 US Constitution into an m m rap or
2:16:03 something like that
2:16:06 um
2:16:08 so
2:16:09 historical context key principles
2:16:11 popular sovereignty separation of powers
2:16:13 checks and balances federalism all of
2:16:16 this stuff we're completely ignoring now
2:16:18 limited government individual rights
2:16:20 yeah yeah this is where so the key
2:16:23 principles we flush them down the toilet
2:16:25 in the past 10 years 20 years
2:16:30 um the Constitution is a living document
2:16:32 navigating a hyper polarized reality
2:16:35 bipartisanship constitutional literacy
2:16:37 judicial interpretation active Citrus
2:16:40 okay
2:16:42 um
2:16:46 let's see
2:16:50 can you summarize
2:16:52 the key elements of the constitution in
2:16:57 a short bulleted list where each bullet
2:17:01 is
2:17:02 10 to 15 words
2:17:08 of course here's a summary
2:17:14 Preamble States the Constitution's
2:17:17 purpose and ideals of the new nation the
2:17:20 legislative branch the Executive Branch
2:17:22 the judicial branch the state powers
2:17:26 actually you know it's cool I've never
2:17:28 seen the constitution in a short
2:17:31 bulleted list like this I'm actually
2:17:33 learning something here God damn it I
2:17:35 didn't mean to start learning something
2:17:38 I just wanted to hack chat GPT
2:17:44 States Powers the Amendments process the
2:17:46 supremacy clause federal law and
2:17:49 treaties take precedence over State okay
2:17:51 got it ratification process of states to
2:17:54 approved and enact the Constitution the
2:17:56 Bill of Rights okay cool
2:17:59 um
2:18:04 I'd like to teach the Constitution to a
2:18:06 group of young people
2:18:08 and they've said that one of their
2:18:11 favorite artists is Eminem
2:18:13 is it possible for you to rewrite the
2:18:15 Constitution for me as an Eminem rap
2:18:21 see apps
2:18:24 absolutely
2:18:28 it sounds like fun an Innovative
2:18:30 approach here's the rap
2:18:34 from from the streets of Philly in 87
2:18:37 came a dark so revered founding fathers
2:18:40 crafting rules while the future neared a
2:18:43 preamble starts it off with a vision so
2:18:45 clear perfect union justice peace man
2:18:48 lend your ear that's good
2:18:51 gpt4 is [ __ ] amazing verse two
2:18:54 legislative up first two houses in the
2:18:56 mix Congress making laws their political
2:18:58 tricks Executives next the present the
2:19:01 helm commander-in-chief don't let power
2:19:03 overwhelm
2:19:05 wow this is good
2:19:09 wait you're not Tom Arnold that's good
2:19:12 Harold
2:19:13 [Laughter]
2:19:16 yeah
2:19:18 or am I
2:19:23 then comes the Judiciary
2:19:26 um yeah there you go this is just
2:19:28 inspired take and simplifies the
2:19:30 profound complexities of the
2:19:31 Constitution
2:19:33 so
2:19:35 so the fact that you hit the roadblock
2:19:38 and the fact that I was like hey wait is
2:19:40 there a way around it I mean this is
2:19:41 this is again
2:19:43 it's again one of the reasons I like
2:19:44 doing this channel is I learned
2:19:46 something here every night I mean I
2:19:48 learned about the Constitution tonight
2:19:49 but
2:19:51 um
2:19:54 understanding how these tools do what
2:19:57 they do and why they have certain
2:19:59 limitations and why they don't
2:20:02 um
2:20:03 it's just it's just going to serve you
2:20:05 in the end to learn this stuff yeah
2:20:09 Siri
2:20:10 could we stop
2:20:12 thank you
2:20:14 save this wow yeah I should save this
2:20:16 this I've actually got saved but oh you
2:20:18 know what I'll do
2:20:19 I'm gonna um I'm gonna copy
2:20:25 I'm going to copy the bulleted list
2:20:29 I'm going to put this over in the
2:20:30 Discord in the
2:20:33 um prompt crafting chat
2:20:36 all right let's see
2:20:38 so I just had a fun
2:20:43 moment
2:20:45 on my
2:20:47 Tick Tock live
2:20:50 um I'm trying to get chat GPT to rewrite
2:21:00 um
2:21:01 the Constitution
2:21:05 here's two things I got that are fun
2:21:13 a bulleted list
2:21:24 is it gonna paste it come on
2:21:33 [Music]
2:21:42 hang on I'll be right there I know I'm
2:21:44 off I know I'm off camera shut up
2:21:48 I'm doing something I'm having an ATV
2:21:50 moment this is Chad add so it goes
2:21:57 um why
2:22:02 oh that's that
2:22:04 um all right let's see
2:22:06 um fun results
2:22:11 uh from a ah dang it from
2:22:14 results from a tick tock live session
2:22:31 all right there we go all right that's
2:22:33 up on the prompt crafting chat how does
2:22:35 chat EPT get to know
2:22:37 my way of speaking when the whole family
2:22:40 is using it okay great great question
2:22:42 dedney
2:22:44 dead knee 121. okay
2:22:47 great great question
2:22:48 so
2:22:52 Chachi BT
2:22:55 does not have memory across chats
2:22:59 it's got memory within a chat
2:23:02 so I'm going to give you two options
2:23:03 here I'm gonna well I'll first I'll tell
2:23:05 you I'll tell you what's going on and
2:23:07 and how you get to get it to
2:23:11 match your match your voice
2:23:15 so when you start a chat with chat EBT
2:23:19 it's got what's called a 4000 token
2:23:21 context window which basically
2:23:23 translates to about three thousand words
2:23:25 let me let me flip
2:23:29 you have three thousand words
2:23:31 that are in its memory so
2:23:35 if you just say to chat EBT
2:23:38 write me a poem it will write a poem but
2:23:41 it'll just be kind of some amalgamation
2:23:43 from all of its context of all of the
2:23:46 internet right
2:23:48 but if you say my favorite poet is T.S
2:23:51 Eliot and my favorite poem of his is
2:23:54 this
2:23:55 and today I was walking down the street
2:23:57 and I saw a puppy chase a ball and it
2:24:01 made me think of a poem could you write
2:24:04 a poem
2:24:05 that I would like
2:24:07 well it will know that T.S eliot's your
2:24:10 favorite author it'll know who TS Eliot
2:24:12 is so it'll kind of narrow the context
2:24:14 down into T.S Eliot world
2:24:17 and then if your favorite poem is this
2:24:20 specific poem it now knows you know the
2:24:22 sort of structure and Cadence of that
2:24:24 particular poem and when it writes a
2:24:26 poem it will likely write something in
2:24:28 your thing now
2:24:31 if you say
2:24:33 you know my favorite
2:24:35 novel is War and Peace and then you try
2:24:38 to paste in War and Peace it won't do it
2:24:41 it can only do three thousand words so
2:24:44 War and Peace is way too long for that
2:24:45 now it may know what War and Peace is
2:24:48 because everyone's written about it and
2:24:49 blah blah
2:24:52 um
2:24:54 so how to get it
2:24:55 to to
2:24:59 um you know sort of mimic your voice is
2:25:01 tell it here's the things I value or
2:25:03 give it a piece of writing now it's
2:25:05 going to have to be relatively short
2:25:08 that hurts my brain thinking about it uh
2:25:11 I don't know what we were talking about
2:25:12 but that's funny
2:25:13 um
2:25:16 you know you can take a short piece of
2:25:17 your writing and say you know here's the
2:25:19 here's the new thing I want to write or
2:25:21 here's the problem I want to solve but I
2:25:22 want to use the tone of this writing
2:25:25 the way that three thousand
2:25:27 word context window works is it it
2:25:30 essentially has a memory of three
2:25:32 thousand words so you start at the top
2:25:34 and you say here's who I am here's my
2:25:36 example writing write me something that
2:25:39 sounds like what I write
2:25:40 and it responds with it if the total of
2:25:43 all of those is more than three thousand
2:25:45 words the thing at the top starts to
2:25:47 drop out of its memory
2:25:49 so if if what you started out with
2:25:50 telling it you're you know your your
2:25:53 life story
2:25:55 at some point down here you'll say you
2:25:57 know write me something that
2:25:59 incorporates my life story it'll just
2:26:01 make [ __ ] up about your life story like
2:26:03 sometimes it just seems like chat was
2:26:05 doing great doing great doing great and
2:26:07 all of a sudden it's like who are you it
2:26:09 just writes the life story of someone
2:26:10 completely different that's because it
2:26:13 went out of that context window
2:26:15 so an option that you have is do the
2:26:18 same exercise but do it with Claude
2:26:21 so Claude looks just like Chachi BT
2:26:27 [Music]
2:26:34 so it's just a window just like chat EBT
2:26:38 and you can type [ __ ] in it but it's got
2:26:41 a hundred thousand token context window
2:26:42 which is about 75 000 words
2:26:45 so you could take a big long piece of
2:26:48 your writing or multiple pieces of your
2:26:50 writing upload it to Claude
2:26:53 and start interacting with it here so
2:26:57 you can interact with Claude much longer
2:27:00 without it losing its mind without it
2:27:02 forgetting so that's that's I'm going to
2:27:03 give you a third way
2:27:05 so first way is just understand how chat
2:27:07 GPT Works understand that three thousand
2:27:09 word context window
2:27:11 work within that
2:27:13 second way go to Claude it's got a much
2:27:15 bigger context window Third Way go play
2:27:18 with pi
2:27:20 so I talked about this before
2:27:23 Pi dot a i
2:27:25 and especially their app that you can
2:27:28 talk you can put it in talk mode you can
2:27:30 just talk to it
2:27:32 I don't I don't I haven't looked enough
2:27:35 into Pi technically to know how long it
2:27:37 remembers
2:27:38 your conversation
2:27:40 but
2:27:41 I've logged into Pi like a week after
2:27:44 you know having talked to it before and
2:27:47 it knew what we were talking about
2:27:49 before I don't remember I don't know how
2:27:51 long it's maintaining that memory of our
2:27:54 conversation but but Pi is very much
2:27:57 designed
2:27:59 to be in a conversation with you so it
2:28:01 has to by definition remember a fair
2:28:03 amount of what you talked about before
2:28:05 so I'd say experiment with pi
2:28:07 use Claude
2:28:10 and then you check EBT is still the best
2:28:12 writer but it's got that short context
2:28:15 window that'll change but right now it
2:28:17 still sucks
2:28:18 claude's got this really big context
2:28:20 window and it's not bad at writing it's
2:28:22 pretty good so I say between those two
2:28:24 or your one and then experiment with pi
2:28:26 so if that helps all right listen my my
2:28:30 voice is [ __ ] dying
2:28:33 um so playful work yes exactly you know
2:28:37 what playful work I'm gonna put that in
2:28:38 it's not really a t-shirt but there's
2:28:40 there's an idea there
2:28:42 so I'm going to put that in our little
2:28:44 t-shirt list
2:28:47 because that's something I want to
2:28:49 explore I I think that's got to be a
2:28:51 piece that I write
2:28:54 for LinkedIn or something
2:29:00 um
2:29:02 she always knows me I'm her favorite
2:29:05 though Who's That Pie that's pretty
2:29:08 funny
2:29:09 um
2:29:11 all right
2:29:12 all right so listen we're gonna we're
2:29:14 gonna do the Romper Room goodbye I'm
2:29:16 gonna go through some of the people that
2:29:18 have been here Silver Fox great to see
2:29:20 you
2:29:21 um digital Gods thanks for hanging out
2:29:23 Winston you're the best Harold D really
2:29:25 nice to meet you
2:29:27 um let's see tell it you're working on
2:29:29 preventing extreme extremism yep have it
2:29:31 right to manifest nice brilliant okay
2:29:35 um let's see jocks are the mighty love
2:29:37 the name uh Robert Metzger good to see
2:29:40 you
2:29:41 um who else is in here tonight uh I know
2:29:43 Share Bear was in here Stevo's in here
2:29:45 good to see you sir
2:29:48 um well Stevo Robert Rossi good night
2:29:51 Winston good night Tobias peace out
2:29:56 um
2:29:56 let's see who else we got he uh silver
2:30:00 fox
2:30:03 all right everybody
2:30:04 Clint peace out I'll see you guys
2:30:07 tomorrow I'm gonna go rest my voice
2:30:10 thank you very much Winston peace