aiLL 5/31/23 - Demystifying AI Tools at Your Fingertips

Description

Chapters

Transcript

0:01 foreign
0:36 sorry it's getting things set up again
0:38 all right I reset my Wi-Fi router I'm
0:40 back live again the hair is being
0:42 special
0:44 um Kyle Shannon with the AI learning lab
0:46 we're getting just getting going live
0:48 again so welcome back I got ice water
0:52 resetting a router is hard
0:56 um
0:57 uh this is the AI learning lab if you've
0:59 got questions about AI
1:01 um drop me a uh a comment and I will do
1:04 what I can welcome back for those of you
1:06 that were
1:07 at openai.com that's the official chat
1:10 EPT website if you have not played with
1:12 it you should absolutely play with it Go
1:16 Tell it to write you a poem or write you
1:17 a song greetings from Minneapolis hello
1:19 Minneapolis
1:22 um also if you want to geek out on lots
1:23 of different tools futurepedia.io I
1:26 don't have any affiliation if you've got
1:27 questions about AI pop those in the
1:29 comments uh hello everybody Welcome
1:33 um I am a uh I'm just a guy I'm a I'm an
1:37 entrepreneur I'm a Founder uh and CEO of
1:41 a video technology company I've been
1:44 deep diving in AI for about the past
1:46 year and I started this Channel and I
1:48 started an AI salon and I started an AI
1:51 art project and I'm now completely
1:53 reimagining my business
1:55 uh uh on AI and and uh it's every air
2:01 the the more I learn about this stuff
2:02 the more profound I think it is so
2:05 that's why I'm doing this that's who I
2:07 am I'm happy to answer any questions I'm
2:09 happy to chit chat
2:11 um if you love the idea of a salon just
2:13 join if you go to the salon.ai there's a
2:16 link tree there so we've got a meet up
2:18 we've got we do meetings every two weeks
2:21 we've got a Discord server where you can
2:23 geek out and you can do mid Journey
2:25 images and ask questions and and you
2:29 know meet meet cool people that are
2:31 doing things
2:33 um so so definitely come join and uh
2:36 it's a cool Community it's really neat
2:38 so
2:40 um
2:40 yeah so that's that
2:42 um I try to keep up on the news although
2:44 you know it's the stuff is moving so
2:48 fast right now that if I take like a day
2:50 off like today I actually worked on work
2:53 work you know like for my company
2:55 instead of geeking out on AI and I feel
2:57 like oh my God what did I miss and it's
2:59 a lot like there's just there's a lot
3:01 every day right now uh the connection is
3:03 much better great yeah I reset the
3:05 router so thank you for that I
3:07 appreciate the feedback Emilio's wife
3:09 hello I'm back I I'm out of detention I
3:13 was in detention for a while oh there's
3:15 a whole bunch more people here now um if
3:17 you're new to this my name is Kyle
3:18 Shannon this is the AI learning lab I
3:21 talk a lot about AI here
3:23 um that site is the official chat
3:26 website if you have not played with it
3:28 just go there right now and while you
3:30 listen to me ramble
3:32 have it write a song for your
3:33 significant other or about your dog or
3:36 or or whatever
3:38 um ask it to write you a python
3:40 application that my big Insight with
3:42 chatgpt was
3:44 um asking it to program something for me
3:46 um have it reformat some HTML have it
3:49 analyze some code have it I don't know
3:51 it's just remarkable and so that's that
3:53 and then futurepedia.io is just a site
3:57 with lots of uh it's just a list of
3:59 different AI tools so so okay yeah I get
4:04 put in timeout regularly and I don't
4:06 know why because I've never read the
4:07 community guidelines but apparently I
4:10 violate them a lot I think actually last
4:12 time I got booted when someone asked me
4:14 a question about
4:16 um bias in the models and I was
4:18 explaining how the bias gets in there
4:19 and how to deal with it and it just
4:21 booted me so I think the minute you
4:24 start talking about polarizing issues it
4:28 decides if it doesn't like the issue
4:29 you're talking about it says okay you're
4:32 gone so I think that's what happened
4:35 um
4:36 so all right
4:40 connection's much better that's great if
4:43 you have questions about AI pop those in
4:45 the in the things below
4:48 um I can I can do little show and tells
4:52 you know what I might do is I might pop
4:54 open
4:56 Photoshop
5:00 beta
5:02 and I might show you guys some of the
5:04 generative fill stuff if you haven't
5:05 seen that stuff yet it's pretty pretty
5:07 remarkable
5:08 is this the beta yeah that's the beta
5:12 um I think we could probably do it in
5:14 there couldn't we let's see
5:17 mm-hmm
5:24 [Music]
5:27 go to layers where's layers
5:36 all right so I'm going to turn off the
5:38 copy
5:42 and then so the way this works let me
5:45 grab the
5:49 change my canvas size here
5:54 we'll go width we'll go a thousand
5:58 oops
6:02 thousand height will go 1400
6:07 all right so so now I've got this this
6:11 cool image that I made in stable
6:14 diffusion
6:16 that's sitting in the center of this big
6:18 canvas and what you can do now is you
6:21 can just start to extend
6:23 the canvas so if I just select
6:26 the area to the right
6:28 and the area to the left
6:34 it now has it now pops up this little
6:37 um
6:39 contextual menu it's this little
6:41 contextual menu and it says generative
6:43 fill and if I hit that
6:46 I have to agree to their terms if I hit
6:48 that I can type in something or I can
6:50 just hit generate and so I'm going to
6:52 hit generate now
6:54 let me try to get this so you can see it
6:56 get my fat neck out of the way
6:59 all right so there's the image so I'm
7:00 going to hit generate
7:02 so now it's thinking
7:10 all right it's doing all that stuff
7:14 and
7:17 it didn't create anything I didn't
7:18 create anything what is going on
7:21 hold please
7:23 oh I might have been on the wrong layer
7:24 I know it's on the wrong layer
7:26 I'm a dum-dum I'm a dum-dum I don't know
7:30 if you knew this but I'm a dumb dumb
7:32 okay hold please uh do I want to delete
7:35 generative layer one yes I do I'm gonna
7:37 go to now I'm going to go down here okay
7:40 sorry sorry you know
7:43 I'm I'm photoshopping live here people
7:45 what do you want from me what do you
7:47 want from me I'm trying to show you some
7:49 cool stuff okay so I'm selecting the
7:51 area to the right I'm shift selecting
7:53 the area to the left and now I'm going
7:55 to say generative fill generate
8:00 and you can extend the image
8:02 automatically too I don't know how to do
8:04 that that's cool though all right let me
8:07 see
8:08 extend the image that's yeah that's what
8:10 I'm doing
8:12 oh wait I didn't hit generate again okay
8:14 all right there's our little progress
8:16 bar now we're going this is going to be
8:18 so good it's gonna be so worth the wait
8:20 I just went from 150 viewers to 50
8:24 people like what the hell is this this
8:26 guy just why is it not showing my oh
8:30 it's not working now I don't know how to
8:31 I can't do this live why is it not
8:33 working
8:35 why is it not working
8:39 huh
8:41 I do not know why it's not working
8:48 [Music]
8:50 well anyway all right well that was one
8:54 of those you know you know those things
8:56 where people are like oh I'm gonna show
8:57 you this cool thing and then they just
8:59 do what I just did yeah that was nice
9:01 all right well whatever
9:04 um okay so if you're new here I will
9:07 answer questions I can about
9:09 um AI apparently my Photoshop beta is
9:12 not working for some reason it's that's
9:14 very very weird I don't know what was
9:16 going on with that what you doing I'm
9:18 just hanging out so if you if you're new
9:20 here go to that URL that's chat GPT if
9:23 you haven't played with chat GPT start
9:24 that
9:25 um can you pull a photo out
9:28 wait can you pull a photo out of a video
9:32 using Photoshop yeah you can just do a
9:34 freeze frame within a video right I I
9:37 just do
9:38 screen captures I just pop things open
9:40 in QuickTime and just do a screenshot
9:42 do I think a person or AI booted you I
9:45 have no idea
9:47 uh
9:48 uh I assume it's the algorithm
9:51 um do you think the excitement over llms
9:54 will lead to neglect in research of
9:57 other methods
9:59 hmm that's a good question
10:04 I think llms are here to stay I think
10:07 that
10:10 I I mean it's it's hard to know you know
10:13 where things are going but I think the
10:15 future is multimodal and and if you look
10:17 at the plug-in architecture of Chach EBT
10:21 um
10:22 right now these tools feel very single
10:25 purpose right like touch ebts for words
10:27 and mid Journeys for images and
10:29 photoshops for images and we're going to
10:31 be moving into a much more multimodal
10:34 kind of future where you just ask it to
10:37 do stuff and it will figure out do I
10:39 need to put a plug-in here do I need to
10:41 make an image do I need to make a video
10:44 does this person want to hear music do
10:46 they want me to write them a symphony
10:48 do do they want me to print out the
10:51 sheet music of the symphony right so I I
10:53 think I think we're we're
10:57 I kind of feel like AI for for regular
11:00 people which let's call the you know
11:03 November 30th is when of of 22 is when
11:07 chat GPT came out
11:09 this marks the the turning point where
11:12 regular people get access to these
11:14 incredible incredibly powerful tools and
11:16 and you know have kind of adopted them
11:18 overnight
11:20 um I think we've got training wheels
11:22 right now
11:23 and I think we're just kind of learning
11:25 what these things are and what what's
11:26 even possible
11:29 um
11:30 I think things are going to get very
11:32 transformative very quickly
11:35 um so
11:36 I suppose focusing on large language
11:38 models is gonna
11:41 it might hamper other activities but if
11:45 you look at if you look at what's
11:46 happening on GitHub every time a new
11:49 concept comes out white paper comes out
11:51 technology comes out it gets adopted by
11:54 by the the dev Community like that and
11:57 you you're just there's just an
12:00 explosion right now in Innovation and
12:03 some of it's really crappy and just
12:04 thrown together but some of it's really
12:06 exciting and you're seeing people kind
12:08 of mixing and matching the piece Parts
12:10 as they come out so
12:12 um I I don't I don't think so have I
12:14 messed with any llms mixed with
12:17 extensions via uba uba Booga I have not
12:23 um I don't know what that is but I'll
12:24 check it out my thing is it forgets
12:27 things you set as non-variables uh
12:30 Chachi BT does yeah well it's got a
12:33 three thousand um
12:35 word or a 4 000 token context window so
12:40 if you do anything longer than that it
12:41 forgets stuff off the top there are
12:44 going to be solves for that and and in
12:47 fact one of the solves is if you go to
12:51 um po.com
12:56 you have to pay for this but
13:00 you get access to all these different
13:01 models right so the sage is there's gpt4
13:05 Claude Plus
13:07 and then Claude instant 100K what that
13:10 means is that's a hundred thousand token
13:12 model
13:14 so you can you can put into your chat
13:18 window 75 000 words you could literally
13:21 copy and paste a long novel and start
13:24 asking questions on it or in your case
13:26 if you if you're doing things where it's
13:27 for getting context
13:30 um you could uh you could use that for
13:33 you know for for better memory but you
13:36 know there's things like Lang chain and
13:38 and things where you can you can build
13:40 memory so so that it's remembering the
13:43 conversations
13:44 um but that's going to get better
13:45 there's going to be tools that come out
13:48 um I don't know three months six months
13:53 um uh what do I think about the royalty
13:56 license based Falcon 40 billion large
13:59 language model outperforming llama 65
14:01 billion well I think that
14:03 [Music]
14:04 um
14:06 I I don't know the royalty license based
14:09 one but I assume that means that
14:11 um artists or or or data that it was
14:15 trained on they get royalties or they
14:17 get or they get attribution which would
14:18 be kind of cool
14:20 um if it doesn't mean that then then
14:21 that's just my ignorance of that but the
14:24 fact that you you're seeing
14:26 smaller large language models
14:29 outperforming larger ones
14:32 um
14:32 I think we're going to continue to see
14:34 that and that's there was a there was a
14:36 Google there was a called we have no
14:40 moat we have
14:42 no moat
14:46 Google we have no Moats
14:49 so this article so if you if you just
14:52 Google we have no mode this this is an
14:54 article let me see if I can get Focus to
14:56 go
14:57 this is an article that that it's it's
14:59 an engineer from Google this is not an
15:00 official opinion but what he basically
15:03 says here is that
15:05 um for a hundred bucks with with a with
15:08 a small language model
15:10 um you know scrappy little developers
15:12 are creating
15:13 large language models that are that are
15:15 as as good or close to as good as what
15:18 Google and openai are spending millions
15:22 of dollars producing
15:24 um and and so he says he thinks that's
15:27 going to continue to happen and there's
15:28 no real competitive Advantage for them
15:30 to to be doing these big models I don't
15:33 know that that's necessarily true
15:36 um because like Google owns all the data
15:39 right so I think Google's going to be
15:41 able to to make these things Google's
15:43 also very very aggressively starting to
15:46 fight back against open AI it's it's
15:48 ironic that they're having to fight back
15:50 against open AI since they invented the
15:52 technology that is the is the sort of
15:54 key little nugget that's making all
15:56 these things work this well
15:57 um but but they are they kind of
16:00 mothballed something that was really
16:02 cool
16:03 um and then now they're now they're
16:05 struggling to to catch up
16:09 um just had chat EPT write me an SNL
16:12 Monologue for George Carlin Best Line
16:14 Comedy is just tragedy plus time oh
16:16 that's cool I think that's something
16:17 that he actually said
16:18 I'm a big George Carlin fan
16:22 I am using experimental Google Bart 2
16:25 this is much more advanced
16:28 oh cool
16:30 oh wait
16:32 say I missed whatever you said before
16:34 that oh agent GPT can overthink about
16:36 questions you ask
16:38 it by itself yeah the the the autonomous
16:41 agents so so what she's talking about
16:44 there is there's there's these things
16:46 called autonomous agents one's called
16:48 Auto GPT there's another one called Baby
16:50 AGI
16:52 I'm sure there's a pile of them out
16:54 there now that um rather than you having
16:57 to prompt it and and you know give it
17:00 explicit an explicit request and get an
17:03 explicit response you give it a goal you
17:05 say I want you to go find I don't know
17:08 the revenue and marketing taglines of my
17:11 competitors and they will just go figure
17:12 out what it needs to do it'll be like oh
17:14 I gotta go search Google for you know
17:17 what the competitors are and I've got to
17:19 go over here to find the financial
17:20 information it writes its own prompts
17:23 but they get a little lost right now
17:24 they're they're very early so so all of
17:27 those autonomous agents right now are
17:28 pretty janky but but they're still
17:31 pretty cool
17:32 um
17:35 Tick Tock experiment
17:37 followed the live Creator that's cool
17:39 that's a cool name how much is it how
17:41 much is what chat GPT this
17:44 that's free or Poe I was talking about
17:46 Poe I think it's 20 bucks a month for
17:48 Poe
17:50 what do I think of pie
17:53 the mathematical constant is it a
17:57 constant
17:58 [Laughter]
18:00 the mathematical uh thingy
18:03 I think Pi is awesome it's infinitely
18:06 awesome
18:09 never repeats never repeats
18:15 um that's the article that really
18:17 excited me and made me interested in
18:18 learning more about AI yeah the the we
18:20 have no moat thing is a pretty
18:21 remarkable it's a pretty remarkable
18:23 article
18:26 um oh inflection AI Pi I don't know what
18:29 that is let me go see if I can I'll look
18:31 real time uh inflection AI pi
18:38 oh on pie your personal assistant
18:43 is this built on top of
18:47 doing [ __ ] for me right now
18:51 hey pie sorry you've been blocked
18:54 oh
18:57 I think I I think someone was trying to
19:00 scam me damn you
19:03 don't go there
19:07 all right
19:09 you are welcome to leave if you were
19:11 just trying to get me to
19:13 [ __ ] up my machine
19:16 um let me get rid of this window
19:19 all right there you go oh that's a cool
19:21 paper so so this is a paper this is a
19:23 paper called generative agents so
19:27 what the hell is going on oh it's
19:29 reloading okay
19:31 I hope that didn't just install some
19:33 shitty something in my in my machine
19:36 um
19:37 this is a really cool paper generative
19:38 agents
19:40 um these are so this is like a sims-like
19:41 little world
19:43 and
19:45 um these things use chat GPT to talk to
19:48 one another and they they have memory
19:51 from day to day so it's amazing
19:54 so all right
19:56 let's see especially raspberry
19:59 I've had it give me incorrect info when
20:01 I questioned it it apologized and
20:03 answered again yeah Emilio's wife but it
20:06 will
20:07 apologize answer you again and often get
20:09 it wrong the second time and third time
20:11 and fourth time as well I like to call
20:13 Chachi BT mansplaining as a service
20:15 because it will confidently answer you
20:17 even if it's wrong
20:19 mid-journey alternative is leonardo.ai
20:22 and it's free in basic yeah so
20:25 um leonardo.ai is really really good
20:29 um it's it's not mid-journey
20:32 um I think it's built on top of stable
20:34 diffusion leonardo.ai but it's an
20:36 incredibly powerful
20:38 um
20:40 yes I'm white listed it's an incredibly
20:42 powerful front end
20:45 I gotta log in
20:47 ain't doing this stuff live is pain in
20:50 the booty
20:51 but that's it's called live we're doing
20:53 it live isn't this exciting
20:56 I could get infected with oh there's all
20:59 sorts of crazy images what are you
21:01 people making out there anyway
21:04 um
21:05 yeah it's it's uh
21:08 it's pretty cool here's this this is my
21:10 feed on Leonardo
21:11 receive status I wonder if I'm gonna
21:14 have to quit Microsoft edge here for a
21:17 second
21:18 because I think the memory just went
21:20 kaboomy when I went to a shitty
21:23 non-good site so let me
21:27 fire that back up and see if it's still
21:29 malbehaving
21:31 um I've heard having two AI essentially
21:33 debate oh wait hang on essentially
21:36 debate answers leads to more accurate
21:38 results yeah Jeff it that's a way to do
21:41 it and then also
21:43 um
21:44 uh oh hang on a second here
21:48 okay
21:55 oh please all right
21:58 um
21:59 I'm gonna go do something uh
22:02 my machine just hung for a second
22:06 am I still live am I live am I live
22:10 Edge
22:16 all right let's see
22:20 my back let me pin this open
22:24 hello hello am I still alive am I here
22:27 yes okay great
22:31 terrific all right back to questions
22:35 I no longer have memory day to day I
22:37 know welcome to my life too
22:40 um oh so Jeff asked the question I heard
22:42 that having two AIS essentially debate
22:44 answers leads to more accurate results
22:45 yeah they can do that I think there's
22:47 going to be whole systems that are built
22:49 to to make sure things are accurate
22:52 um
22:52 but one thing that you can do with chat
22:55 GPT is when it gives you an answer you
22:58 ask it to evaluate its own answer and
23:00 tell you how it would improve its answer
23:03 and then apparently it gets quite quite
23:06 good after that I haven't played a ton
23:08 with that
23:09 generative agents more than a paper
23:11 anything to demo or interactive I just
23:15 saw
23:17 some video game I saw footage from a
23:20 video game where
23:21 a character walked into a bar and
23:23 started talking with one of the NPCs and
23:26 it was an open AI conversation so
23:30 um
23:31 I assume that stuff's being developed
23:33 right now I think we'll see those
23:35 generative agents quite soon but I I
23:37 don't know that there's anything live
23:39 you can play with you can if you go to
23:41 the paper
23:42 there's a hugging face
23:44 instance of the paper I think you can go
23:46 play with you can go play with the The
23:49 Sims thing that they have set up and you
23:51 can watch it do its thing it's pretty
23:53 remarkable
23:54 would many small smaller llms debating
23:57 answers be more accurate than a much
23:59 larger one probably yeah like I I think
24:06 I don't know
24:08 I think that I think the large language
24:10 models are going to be kind of like
24:13 internet like like like right now we
24:16 just like you have access to Internet
24:17 like everyone's got access to it it's
24:19 kind of this ubiquitous thing I think
24:21 that's how the large language models are
24:23 going to be and then I think you'll see
24:24 tools stacked on top of them and and
24:27 probably smaller more specialized llms
24:30 or just little training sets that that
24:33 you know kind of Leverage both will be
24:35 there so so I think so I think that's
24:37 probably where we're headed is is to a
24:39 more um
24:41 a more
24:43 um
24:44 modular
24:46 kind of kind of future but I also think
24:48 that that modularity
24:51 developers are going to have to deal
24:53 with the modularity but I think for us
24:55 us humans
24:57 that modularity is going to drift into
24:59 the background very very quickly it's
25:00 just going to be like these tools are
25:02 just going to get more powerful and more
25:03 mythical mystical
25:07 what's the prompt to improve an answer
25:10 um I forget what it's called
25:14 um
25:16 it's not query you don't have a query I
25:19 forget what it is I forget what it is
25:22 can it build a business plan sure let's
25:23 go do that right now I'm just going to
25:26 go to chat GPT
25:29 and I'll say here's how we'll do a
25:31 business plan we'll do um
25:34 what are
25:36 three different
25:39 business plan Frame Works
25:45 for a startup
25:47 question mark let's go see what it gives
25:49 us there
25:50 are numerous business plan Frameworks
25:53 The Lean Startup plan
25:57 blah blah blah the business model canvas
26:00 that's a cool one
26:02 I'm a fan of the business model canvas
26:05 and the traditional business plan
26:08 so let's say
26:10 can can
26:12 you create a custom
26:17 framework
26:19 framework
26:22 um
26:23 that is a hybrid of
26:28 these three
26:30 so it remembers it knows what I'm I'm
26:32 referencing when I say these three
26:35 so it says sure here's a potential
26:37 business plan
26:39 that that combines aspects of Lean
26:41 Startup business model plan and
26:42 traditional
26:44 plan the framework can be termed hybrid
26:48 business model hbm this is the this is
26:51 the HPM this is awesome executive
26:54 summary value proposition market
26:57 analysis marketing strategy
27:00 customer relationships key Partnerships
27:02 revenue streams and cost structure
27:04 implementation timeline and key
27:06 Milestones lean methodology
27:08 so is it just giving me 10 are we going
27:11 deeper okay give me 10.
27:14 all right so there's that so so then
27:15 I'll say um
27:19 can you ask
27:22 me
27:26 a series of questions about my business
27:33 so that you have enough information
27:38 to
27:42 um
27:42 write a plan based on that framework
27:50 frame work
27:54 um
27:55 um
27:57 do not
27:59 proceed to the next
28:04 question
28:06 until I say
28:09 okay do you understand what that should
28:12 do that last little thing is rather than
28:14 it just giving me a pile of questions it
28:16 should ask them one by one
28:19 absolutely I understand let's proceed
28:21 step by step
28:23 one can you provide me a brief overview
28:25 of your business outlining what you
28:27 wanted to do why it's Unique and how it
28:29 will generate money I want to sell
28:34 Frozen lizards
28:39 on street
28:41 street corners
28:45 I think frozen lizards
28:49 are cool
28:51 and people
28:54 will pay a lot for them
29:01 interesting concept before we move
29:04 forward could you clarify a few points
29:06 are these lizards so so you know you you
29:10 see you know the the sort of smart-ass
29:12 response I put in there but
29:15 um
29:16 here's here's the thing about about chat
29:19 GPT that that
29:21 this is a subtle shift that took me a
29:24 while to get my head around
29:27 Google trained us for years
29:31 to get good at prompting it right so we
29:34 had to kind of you know early in
29:36 Google's days you had to learn Boolean
29:38 logic and and you know underst they
29:42 trained us how to use it and the basic
29:44 idea there is you want to get good at
29:46 writing a single prompt and getting back
29:48 good results what happens with chat apt
29:50 it's much more like a collaborative
29:52 conversation and so what I just did
29:54 there with the business plan thing was
29:56 just this kind of back and forth hey
29:57 tell me about business plans and then
30:00 make me a hybrid and then you know ask
30:03 me questions and you can really treat
30:07 your interaction with it like you're
30:08 collaborating with with a really smart
30:10 you know co-worker
30:13 um and and that's that's kind of it I
30:15 mean it's it's it's incredibly powerful
30:18 all right I'm going back here am I on
30:20 YouTube I missed this stream
30:22 I'm not on YouTube on Tick Tock and
30:26 and uh
30:29 and and I I've been busy at work so I
30:31 haven't been putting up as many tick
30:32 tocks and then I was in detention for
30:35 for being a bad boy here on Tick Tock
30:38 they kicked me off for a week this is my
30:39 this is my second detention
30:42 um I have no idea what I said I don't
30:44 really care
30:45 um I'm just trying to help people here
30:47 so
30:49 there you have it
30:51 all right
30:53 um
30:58 in the salon space What are you asking
31:02 um I don't know what you're asking okay
31:03 have any recommendations on a program
31:05 that can make a 3D printing model
31:09 from a text or image prompt it's coming
31:12 um
31:13 there was a thing out today
31:18 well yeah let me go look on Twitter here
31:20 hang on
31:21 um
31:22 text
31:24 to 3D
31:29 um
31:34 yeah this thing come on focus focus
31:37 focus
31:40 um and I'm too bright there you go
31:43 um so this is called prolific Dreamer
31:47 High Fidelity and diverse text to 3D
31:51 generation
31:52 with variational score distillation huh
31:57 these Geeks know how to Market don't
31:59 they
32:00 did you know it had a variational score
32:03 distillation no I didn't does it really
32:05 oh yeah what's that do I have no idea
32:10 um but it does something but these look
32:12 like you know fully complete 3D 3D
32:15 objects that they're rotating around
32:16 here so
32:19 two or three months ago the text to 3D
32:22 stuff was
32:24 um you know kind of laughably bad now
32:26 those things look like decently modeled
32:29 3D models so I would say give give it
32:36 you know maybe by the end of the summer
32:39 um you should be able I'm sure there
32:41 will be a service that's that's you know
32:43 type in your thing here and output your
32:46 model there I would I would be not
32:49 surprised at all if if a site like
32:51 thingiverse or one of the 3D model sites
32:54 um just Incorporated text to 3D right in
32:56 it so I I would imagine very very
32:58 quickly you're gonna you're gonna be
32:59 able to
33:00 either take a 2D image turn that 3D or
33:03 go text to 3D and go straight to the
33:05 printer I I think that's coming quite
33:07 quickly
33:08 Pi is in training but never been blocked
33:11 is that chat gbt three four or five I
33:14 don't know Pi kicked me off that like I
33:16 got I got blocked so I don't know what
33:18 that was about
33:20 ask it a question that needs answered in
33:23 order to do a Biz plan for a biotech
33:25 startup you didn't think my frozen
33:27 lizards was a compelling business idea
33:30 wait ask it a question it needs answered
33:34 in order to oh okay okay
33:37 um okay the a joke
33:41 let's
33:42 start over
33:46 what
33:47 um
33:51 what if I want to start a
33:56 AI powered
33:59 biotech startup
34:03 can you help me
34:08 refine my idea question mark
34:13 of course let's move forward with your
34:16 AI powered biotech startup idea it's a
34:18 very exciting field with great potential
34:21 what's the problem you're trying to
34:22 solve what's a unique solution so it
34:24 just goes through and asks you so so
34:26 yeah you can and what I would say is you
34:28 know rather than having me do this here
34:31 um just go to
34:34 here
34:36 chat.openai.com that's the official chat
34:39 GPT web start site and you know while
34:41 you're listening to me here just start
34:43 interacting with it start asking me
34:44 questions you will get to a business
34:47 plan
34:48 in
34:49 if you've never done it before at all
34:51 and you don't have any clue about it
34:53 probably within 45 minutes if you kind
34:57 of have a sense of what you want and
34:58 know what you want it'll help you focus
34:59 that very very quickly I hope I offer AI
35:03 prompt engineering courses
35:05 um I'm thinking about putting together a
35:07 workshop I haven't done it yet so if if
35:09 that's something that you'd like you
35:11 know give me a follow let me know in my
35:12 DMs
35:13 um or let me know in here it's that's
35:15 something I might put together
35:18 um I love Frozen lizards dry ice costs a
35:21 lot that's good see I'm getting good
35:22 business advice here for my frozen
35:23 lizard business I don't know why you
35:25 guys think this is a bad idea I think
35:27 it's pretty brilliant
35:28 all right let's see
35:31 um it writes amazing content for
35:33 marketing cop copy email copy ebooks if
35:37 you give it the right info yeah there's
35:39 an interesting thing there's a
35:41 if you ask it shitty questions you get
35:44 shitty content if you ask it good
35:46 questions you get good content and and
35:48 one of the things that is
35:50 counterintuitive about it is if it gives
35:53 you crappy content don't quit don't just
35:55 go oh it gave me shitty content
35:58 um tell it it gave you shitty content
36:01 and tell it why that was boring I want
36:04 it to be more funny can you make me
36:06 sound smarter
36:08 um
36:09 you know give it more context than than
36:13 you think is reasonable it's it's almost
36:15 like like what I've found is it's almost
36:17 like
36:18 if someone just introduced you to a
36:20 brand new intern at work and the intern
36:22 was really really smart
36:25 and could kind of do anything
36:28 and you wanted to give that person a
36:30 project what would you say well you'd go
36:32 well hey there
36:35 um
36:36 I'm doing this kind of a project here's
36:39 what the project is about here's the
36:42 specific problem I'm trying to solve
36:44 like if you give it lots of context and
36:47 then ask it to do something it will give
36:48 you much better answers it's it's pretty
36:50 amazing that way
36:53 um
36:54 oh can't drink anything on screen I
36:56 think even water okay that's weird
36:59 uh I should have an ALT account to
37:02 bounce to when I go to jail yeah I
37:04 should probably do that I hadn't thought
37:05 about that that's a good idea
37:07 I will come up with an ALT account it'll
37:10 it's it's uh who's that guy Hawk Hawk's
37:13 got like seven of them so I guess I'll
37:15 have a different account oh man uh tick
37:19 tock tick tick tock experiment it will
37:21 be like any big tech tech transformation
37:23 eventually things just suggest who owns
37:26 the rights to the answers that chat gbt
37:28 gives you do
37:30 um but
37:31 um cheche EBT can use them for training
37:33 there's there's a bunch of security
37:35 issues so if you go to your settings
37:38 there's a data controls thing and you
37:40 can turn off your chat history
37:43 and training so so you're basically
37:45 telling chat GPD not to use your data
37:47 and not to save your chat history
37:49 because that's where a lot of people are
37:51 losing their data
37:53 um but if you if you want to be really
37:55 paranoid about
37:57 um your data privacy and security
38:01 um run a large language an open source
38:03 large language model locally or on a
38:06 virtual machine
38:07 um locally would be the most secure
38:10 um so I would do that but that requires
38:12 some some more geekiness but but you can
38:15 um you've got some level of data
38:17 controls now within chat GPT and those
38:19 are going to get better and I I'm pretty
38:21 sure that chat GPT is going to give
38:24 um
38:26 they're going to give
38:28 um
38:29 that they're creating a business a
38:32 business level account I'm thirsty I'm
38:34 gonna go drink off camera here for a
38:35 second
38:37 if that's really a rule that's the
38:39 dumbest thing I've ever heard
38:41 like what are people supposed to do they
38:43 want you to stay on these things for
38:45 more more than 30 minutes but you can't
38:47 drink water okay text to 3D sounds like
38:49 it's got a lot of retro encabulator
38:52 going on yeah exactly
38:55 um has anyone interested in creating a
38:58 promo
38:59 for my AI map uh please don't
39:02 um uh you know promote business stuff in
39:05 here
39:05 like ask questions like I I really don't
39:07 want this to be a Spam spammy place I
39:09 want this to be a place where we can
39:10 just talk and
39:11 and learn
39:13 um
39:14 where you know I'm I'm happy to talk
39:16 about your business but don't don't spam
39:18 to the audience if you wouldn't mind
39:20 prompt engineering is made unnecessarily
39:23 complicated no I I disagree with that
39:26 I think
39:27 prompt engineering or prompt crafting I
39:30 prefer the term prompt crafting because
39:32 I think it is a really creative Endeavor
39:35 um it's complicated right now because
39:38 the tools are so new like we're just
39:40 getting our hands on this chat interface
39:45 um and
39:48 here's the thing you're you're not even
39:51 querying a database you're querying a
39:54 system
39:55 that essentially knows all of the
39:58 internet
40:00 so it's an incredibly vast
40:04 data set that you're interacting with
40:07 so the reason that that prompts can get
40:10 complicated is that if you want really
40:12 specific results you kind of have to
40:14 steer the large language model into a
40:17 semantic corner
40:19 and
40:21 so so that your answers are are more you
40:24 know are in a semantic region so if you
40:26 wanted to be talking about you know
40:30 um
40:31 whatever quantum mechanics then you want
40:34 it kind of in the region of of its
40:36 knowledge where there's lots of quantum
40:38 mechanics knowledge in there you don't
40:39 want it over in in the area where it's
40:41 talking about poster design
40:43 um so
40:44 so that requires you to to do
40:48 multi-layered
40:50 um kind of conversations that focus it
40:52 more and more and more and more
40:55 um and that can also get frustrating
40:56 because because it's got a limited
40:59 context window
41:01 it will sometimes lose its way where you
41:04 started a conversation in one place you
41:06 gave it a bunch of information and then
41:08 as that kind of Scrolls off the top of
41:10 its context window it will kind of
41:12 forgot forget that it you told it that
41:14 so so
41:17 current prompt engineering is
41:19 complicated because we're early but this
41:21 is just like the early days of the web
41:23 you had to manually type in HTML there
41:25 weren't site Builder tools and there
41:27 weren't automated any things
41:29 and there weren't e-commerce engines if
41:31 you wanted any of that crap you had to
41:33 build it yourself and so it was just
41:35 more complicated and so that's ah look
41:37 at that lovely little heart there's a
41:39 lovely little heart flying right on my
41:40 head that was lovely thank you and yeah
41:44 I got I have a lot of fun working on
41:46 prompts yes
41:47 prompt writing is fun you'll get banned
41:49 for anything apparently
41:51 chap GPT says it is not for commercial
41:54 use when asked yep that's probably true
41:57 I would you know
41:59 I would be careful with it you can
42:02 um open AI
42:03 um announce you can do Standalone
42:05 machines on Microsoft Azure that are
42:08 that are like Standalone
42:09 um open AI instances so if you're
42:12 working for a big company
42:15 um
42:15 you can do that I mean Microsoft's going
42:17 to make it very very easy says AWS sews
42:21 everyone everyone is going to make it
42:24 easy to do
42:26 um your your your own data
42:31 um we have a dehydrated King exactly I'm
42:35 gonna go have a sip now
42:36 [Applause]
42:40 okay so if you're new here this is the
42:42 AI learning lab we are learning about AI
42:44 um I'm I'm sharing what I can if I don't
42:47 know I will I will uh
42:50 tell you that if I do know I'll tell you
42:53 that uh if you're new go to that URL
42:57 chat.openai.com that's the official chat
42:59 GPT website it's I don't get paid for it
43:02 um just start playing with it it's it's
43:04 I think it is super super important for
43:08 anyone who hasn't tried this stuff to
43:10 try it start using these tool tools
43:13 start getting educated about them get
43:15 curious about them see what they
43:17 actually make possible because they are
43:19 remarkable and they're going to be
43:20 everywhere and all of us are going to be
43:22 required to use them so why not do that
43:24 voluntarily
43:27 what can you ask or not ask GPT well so
43:32 it's got a bunch of safety
43:37 um guard rails built into it so if you
43:39 ask it things like you know
43:41 how do I build a dirty bomb or how do I
43:43 probably I'm probably gonna get kicked
43:45 off because I said that how to how do I
43:47 build
43:48 um whatever um how do I do how do I harm
43:51 someone right all that sort of stuff it
43:53 will
43:54 um
43:54 it will tell tell you that you can't ask
43:56 that there are hackers that are you know
43:59 kind of hack around it and get around
44:01 the guard rails the open source models
44:04 don't have any safety
44:06 um guardrails built into them so chat
44:08 gbd has them
44:10 um you can you know hackers can kind of
44:11 get around them
44:13 um but but basically they're they're
44:15 putting those guardrails in place to try
44:18 to make it a safe thing right because
44:19 they're incredibly powerful tools so
44:21 they want people to use them
44:22 and basically if you push the boundaries
44:24 it will tell you and then you know you
44:27 decide if if you wanted to do something
44:29 that it's going to not let you do then
44:31 go get an open source model and and you
44:33 know set up your own server you can do
44:34 that did I hear about the FDA approval
44:37 of neurolink yeah so so neurolink if you
44:40 don't know is Elon musk's company where
44:42 they have these remarkable they built
44:44 this remarkable robot that that inserts
44:46 these 60 electrodes in
44:49 um well in your brain they drill a hole
44:52 in your skull put these electrodes in
44:54 your brain and it's it's essentially a
44:56 direct brain to computer interface and
44:58 they've opened it up for
45:00 um human trials so yeah that's coming
45:03 Switzerland neuroscientists helped them
45:05 in oh well the Switzerland thing so that
45:07 was different that wasn't neural link so
45:10 there's a guy in Switzerland he was
45:12 paralyzed in a motorcycle accident 14
45:15 years ago or something like that and
45:17 they used a large language model they
45:19 imaged his brain had him think about
45:21 moving his legs they imaged his brain as
45:24 he was thinking about that so they
45:25 figured okay here's the part of the
45:27 brains that activate when he tries to
45:28 move his legs then they built a
45:30 Bluetooth Bridge they did oh maybe they
45:33 did use neurolog I don't know they
45:35 inserted something in his spine so
45:36 something in his brain something in his
45:38 spine and a Bluetooth connection between
45:39 them and then they used a large language
45:41 model to
45:44 um
45:45 to to basically pull the information
45:47 from his brain send it directly to his
45:49 spine and now he's walking again after
45:51 14 years of not walking it's I'm I'm
45:54 telling you like we're we're about to
45:57 enter this just an era of remarkable
46:00 scientific progress
46:03 um I don't care when people find it so
46:04 difficult about making a prompt it's
46:06 just being specific it's listen here's
46:08 the deal if if you've got a
46:11 that's really funny if you've got a
46:14 faculty a good faculty with words
46:16 um if you're good with words there's a
46:18 great Steve Martin joke some people have
46:19 a way with words other people
46:21 um not halfway
46:23 um if you're good with words
46:26 prompt crafting
46:28 feels fairly intuitive if you're not
46:31 good with words it it doesn't
46:34 um but but it's like I would say this
46:36 the people that are good with words and
46:39 understand programming are going to be
46:41 really good at this because chat GPT
46:42 right now is essentially programming
46:45 using your native language
46:48 um so if you're good with words and
46:50 you're good with logical constructs then
46:52 you're going to be really good at prom
46:53 prompt crafting
46:55 um using Ai and brain implant via spinal
46:57 reconnection yeah I know yeah uh yeah
46:59 paralyzed man walks again after 12 years
47:00 might have been 12 years not 14 but yeah
47:02 really remarkable thing
47:04 um they didn't give out much detail
47:06 because of the efficacy of neural link I
47:08 I yeah I mean
47:11 when we're starting to do things like
47:13 curing paralysis
47:15 um I I'm I'm
47:18 you know I'm good I'm down with it I
47:21 think it's amazing
47:25 [Music]
47:27 by the way I set a goal because Tick
47:31 Tock told me to and I listened to
47:32 everything Tick Tock tells me except for
47:35 the community guidelines apparently but
47:37 I set a goal for having
47:39 um 13 subscribers to my live and I only
47:42 have four so if you want to subscribe to
47:44 this you can help me hit my goal I don't
47:46 know what it does but Tick Tock told me
47:47 to do it
47:49 it's it's uh it's tutoring me it's
47:52 probably using AI algorithms to tutor me
47:54 on how to get more engagement so that
47:57 they can brainwash you all with my words
47:59 all right hope for more funding for stem
48:02 cell therapy me too a bunch of companies
48:04 are working on it yeah I think we're
48:06 just gonna see
48:07 um just incredible breakthroughs on the
48:09 medical front the next couple of years
48:10 what kind of price are we looking at
48:12 when chat EPT is finished with this
48:15 trial period I think they're gonna I
48:17 think free will still be there Bing is
48:19 free that's gpt4
48:22 um
48:23 there are
48:25 numerous open source large language
48:28 models so they're they're going to be
48:30 all sorts of tools that are free I would
48:33 be deeply deeply surprised if chat GPT
48:36 didn't keep a free tier deeply surprised
48:47 let's see
48:49 I've been teaching my mom to use these
48:52 tools because
48:53 I don't want her left behind at work
48:55 yeah that's awesome I listen I I'm this
48:59 channel exists because I am really
49:02 passionate about teaching as many people
49:03 as I can about these tools and and not
49:07 even like teach them like I'm a teacher
49:09 and I'm an expert I'm not I'm just a guy
49:11 that
49:12 thinks these tools are going to change
49:14 the world and
49:17 they're gonna It's Gonna Change the
49:19 World so much faster than any of us are
49:21 prepared for so I want as many people as
49:23 possible to be as curious as possible I
49:26 think the people that do positive things
49:29 right now are going to have a
49:30 disproportionately
49:32 um large impact on the future direction
49:35 of this stuff
49:37 um it it that may seem naive and silly
49:40 but I think it's it's really really
49:41 important that people figure this stuff
49:44 out and start to get their heads around
49:45 it so I think that's awesome that you're
49:46 doing that if you know anything about
49:48 this share what you know with people
49:51 I won't share a background into info
49:54 about private people yeah that's smart I
49:56 wouldn't do that either
49:58 um quick overview of Open Source models
50:00 easiest ones to start with oh
50:03 well so I've got
50:05 for a second
50:08 um this is free also I don't I'm not
50:10 charging for this so this is not a you
50:13 know go here to make money thing
50:16 um
50:17 power to the people issue 15
50:22 um so this is now like a month month old
50:25 um but this talks about four different
50:28 large language models red pajamas stable
50:30 LM
50:32 stable LM is from stability so that's
50:35 the that's the company that did um that
50:37 does stable diffusion
50:40 um weblm Mini gpt4
50:44 um and then so that's that's issue 15 of
50:48 everyday Ai and then issue 16 I think we
50:51 had a we had another one in there
50:56 first things about palantir which is a
50:58 terrifying military company
51:02 um
51:07 is there another
51:16 so what's the so so anyways that's where
51:19 that's where I'd go just go good
51:21 everyday AI issue 15. that's a good
51:23 place to start and then the other thing
51:25 to do is just go to go to Twitter and
51:29 start following people that are talking
51:30 about AI stuff you can follow everyday
51:32 AI
51:34 um you can follow Robert scoble is
51:36 talking a lot about this there's a bunch
51:38 of people you can follow Andre carpathi
51:41 from openai
51:43 um
51:44 and then and then if you go to GitHub or
51:47 hugging face those are the code
51:49 repositories and you know hugging faces
51:51 where a bunch of large language models
51:53 live
51:54 um just look for the trending projects
51:56 on GitHub and hugging face and that'll
51:58 probably point you in the direction of
52:00 things that people are excited about
52:02 will neuralink cure sorry
52:09 lost my place I lost my place well
52:12 neuralink clear my depression I don't
52:15 know look you're mine look you're any of
52:17 ours it's uh
52:20 we're just in this mental health morass
52:23 aren't we I just feel like everyone
52:25 right now is struggling on every
52:27 business that I'm interacting with
52:28 people are just checked out
52:31 and you know I think we're all trying to
52:33 figure everything out and uh
52:36 yeah I I you listen I think I think that
52:40 these tools are gonna are gonna have an
52:42 incredibly
52:43 um positive impact on Mental Health if
52:46 we use them right and whether or not
52:48 neurolink is the interface or not I I
52:52 mean
52:53 I can't imagine so so right now
52:57 they've got it so that you know where
53:00 elon's technology is is you've got to
53:02 drill a whole hole in the skull and
53:05 implant these electrodes in the brain
53:07 and that's you know that's instant
53:09 connection to the brain
53:11 but I can't imagine that AI isn't going
53:13 to help us develop a baseball hat kind
53:16 of thing that's got you know brain
53:18 reading technology in it that's actually
53:19 quite fast
53:21 so so we'll see
53:24 um I got an MS in computer science
53:27 um at Georgia Tech I know a lot about
53:29 this field that's cool yeah I have a
53:31 degree in acting from Penn State I you
53:33 know the the I know I know conceptually
53:37 a lot about this and and I I would say
53:40 my expertise is more in using technology
53:44 um once the
53:46 um computer scientists and technologists
53:49 um have gotten it to a state that the
53:51 regular people can start playing with it
53:53 I did that in the mid 90s with the world
53:55 wide web and I'm doing it again here I
53:58 think that
53:59 um
54:00 the the this large language the large
54:02 language model stuff and and AI stuff is
54:05 is changing the game very similar to how
54:08 the World Wide Web didn't back then
54:10 I have a way with words so llms and I
54:13 are BFFs that's great yeah exactly if
54:15 you're good with words man this this
54:16 stuff's a lot of fun
54:19 um AI lab prompt crafting 101 okay I'll
54:21 I'll maybe put together I'll put
54:23 together a workshop and and see if
54:25 anyone's interested in it
54:28 um I must do is The Tick Tock God's
54:30 demand apparently
54:32 um thank you for your service thank you
54:33 for thanking me for my service I you
54:35 know I I enjoy this so I don't really
54:38 see this as a service this is fun
54:40 um I've seen Twitter posts commenting on
54:42 the seeming dumbing down of the results
54:44 they get from chat GPT
54:47 um well certainly certainly
54:50 um
54:51 Microsoft dumbed down
54:53 um Bing after they made the front page
54:56 of the New York Times when when the uh
54:59 when the chat bot told the reporter he
55:01 should leave his wife because it was in
55:03 love with him
55:04 um they dumbed it down after that
55:07 um
55:08 they are they are definitely doing fine
55:11 tuning and training so so these models
55:14 are changing so there's some prompt
55:17 degradation that can happen where you
55:20 have a prompt that works for a while and
55:22 then it just kind of changes and gets
55:23 worse so
55:25 um I wouldn't be surprised at that I
55:27 haven't seen it personally but is python
55:30 the standard when programming AI
55:32 what about CPP or even lisp I would go
55:35 ask uh GPT for that I would
55:38 use the one connected to t plus plan or
55:42 you're paying them they're 20 bucks a
55:43 month
55:44 fee you've got browse with Bing so so
55:48 this is
55:50 um chat GPT connected to the Internet so
55:53 I would go ask it
55:54 um
55:55 uh what language you know what what
55:58 languages to use for AI and specifically
56:00 what you want to do in AI because
56:03 Python's good for certain things and you
56:05 know lisp or CPP might could be good for
56:08 other things so I I would tell GPT what
56:12 you want to accomplish and then have it
56:14 recommend what language you should be
56:17 using and then have it write the code
56:19 for you because it does that and it does
56:20 it remarkably well
56:22 any opinion on graph Theory and
56:24 information Theory also algebraic coding
56:26 Theory well as an actor
56:29 no I don't
56:31 I don't have any opinions on those
56:33 because I don't know what most of those
56:34 are
56:35 um
56:36 I do know that
56:38 the wolf from alpha plug-in for chat GPT
56:42 allows you to kind of mix those two
56:44 worlds together
56:46 so and there's also some of the plugins
56:48 for for
56:50 um chat GPT also have graphing and you
56:53 know visualization capabilities so
56:56 um so that's pretty interesting I would
56:58 say
56:59 um when code interpreter comes out which
57:02 is a another module that they're doing
57:03 where you can do all sorts of fancy data
57:06 analysis
57:07 um that that'll probably be pretty good
57:09 at that stuff but I don't know anything
57:10 about those sorry about that what are
57:12 our thoughts on fine-tuned Laura versus
57:14 large General models I think these
57:17 fine-tuned small models are are like
57:20 like the developer Community is pouncing
57:23 on every time there's a new model out
57:25 and they're getting really good really
57:27 fast so so I think that these small I I
57:32 think probably within the year
57:35 you'll be able to install something on
57:38 any phone any device locally that is in
57:42 the neighborhood of gpt4 that's just a
57:45 small
57:46 teeny little thing that doesn't require
57:49 a massive chip to use
57:52 um so yeah so I I'm I think those things
57:54 are remarkable
57:56 subscribe needs to find 5.99 a month I
57:59 hope you see a big part of that profit I
58:01 don't see a part of any profit on any
58:03 subscriptions so
58:06 um and I don't know what you're talking
58:08 about subscribe means 5.99 a month
58:11 oh oh
58:12 for months here
58:16 if I'm asking you to you described to my
58:18 lives that's 5.99 a month to subscribe
58:20 to my lives I would not pay 50 cents for
58:23 my lives I thought it was just if you
58:25 subscribe you get notified GPT
58:28 gpd4 is incredible to me it's helping me
58:31 create
58:31 structure with my goals that's a great
58:33 way to use it did you hear IBM research
58:36 2033 quantum computer center and smaller
58:39 Fusion energy production
58:42 research 2033 no I haven't heard that
58:45 I've heard a couple of things that IBM's
58:47 doing
58:49 um
58:50 like like they announced they're they're
58:52 sort of dusting off Watson after after
58:54 it failed with Healthcare
58:56 especially helpful with my ad HD makes
58:58 it difficult to structure my goals oh
59:00 that's cool yeah I'm a card carrying
59:02 member of the ADHD Club so welcome to
59:04 the welcome to the club buddy
59:07 um yeah
59:08 um AI is both a blessing and a curse for
59:11 us adhders because man you can run down
59:13 some [ __ ] rabbit holes
59:15 [Laughter]
59:16 all right they also lobotomized it yeah
59:19 they did they lobotomized uh what was
59:22 her name uh Sydney
59:24 they lobotomized Sydney once she when
59:26 she got all
59:28 boiling rabbits on uh on that reporter
59:32 all right we've got Collective fatigue
59:34 from the past five years we do I think
59:36 long covet may be contributing to mass
59:38 lethargy that's an interesting thought
59:41 yeah I you know I think it's I think
59:44 it's um
59:48 I think it's lack of human interaction I
59:50 I think you know I like my company's a
59:53 small company but you know I've got a
59:55 bunch of introverts in there that love
59:56 working from home I hate working from
59:58 home
59:59 so we're getting things done but I've
1:00:02 noticed that just the connection to the
1:00:04 company is not there I I don't think
1:00:06 that's unique to us it's it's a drag
1:00:09 I have problems with the prompts so so
1:00:12 here's the thing you can try is
1:00:16 almost think of Chachi PT as your prompt
1:00:19 therapist actually let's go try this now
1:00:20 this could be kind of interesting I'm
1:00:22 going to use GPT
1:00:25 I'm
1:00:27 not good with words
1:00:31 can you help me
1:00:35 learn
1:00:36 how to prompt
1:00:40 you better
1:00:43 let me just leave it at that
1:00:46 of course I'm here to help be clear and
1:00:49 specific
1:00:51 clearly State what information breakdown
1:00:54 complex questions provide context ask
1:00:57 for clarification use examples so it'll
1:01:00 teach you and say
1:01:02 um
1:01:03 I don't know what any of that means
1:01:09 um
1:01:10 can you walk me through
1:01:14 step
1:01:16 by step
1:01:19 what to do question mark
1:01:23 certainly step one identify your
1:01:25 objective step two
1:01:29 consider examples so so you can ask it
1:01:32 to do that so so one thing you can do is
1:01:34 just start having it train you how to
1:01:36 how to prompt better
1:01:40 um there's another thing that you can do
1:01:42 within if you've got access to plugins
1:01:44 let me let me create a new chat here so
1:01:46 we're going to go to gpt4 plugins
1:01:49 and then one of the plugins
1:01:52 is called prompt perfect so I'm going to
1:01:55 turn on prompt perfect and the way this
1:01:57 works is you type in the word prompt
1:02:01 is it prompt or perfect wait hang on
1:02:09 oh you type perfect okay so I type in
1:02:12 perfect at the beginning of a prompt and
1:02:14 then I say
1:02:16 um
1:02:19 teach wait teach me how to make
1:02:26 better prompts
1:02:29 for chat GPT
1:02:33 and so what this plugin does it's called
1:02:36 prompt perfect what this plugin does is
1:02:39 it will then it will rewrite your prompt
1:02:41 as a better prompt and then give you a
1:02:44 better answer
1:02:46 so
1:02:48 let's see oh error communicating with
1:02:51 the plugin service
1:02:54 regenerate response let me try it again
1:02:56 maybe just add a hiccup now it doesn't
1:02:59 like it it's not working right now
1:03:01 some some of these some of these uh
1:03:03 plugins are a little wonky
1:03:05 um
1:03:05 um prompts.chat
1:03:08 prompts dot chat
1:03:12 which is a big long list why is it not
1:03:15 going come on
1:03:16 prompts.chat there we go call it's
1:03:19 called awesome chat gbt prompts it gives
1:03:22 you a bunch of resources at the
1:03:23 beginning talks to you about prompt
1:03:25 writing and then it's a whole long list
1:03:28 of
1:03:29 you're telling Chachi PD to act like a
1:03:31 certain
1:03:32 um kind of like act like a travel guide
1:03:35 act like an Advertiser act like a
1:03:38 Storyteller act like a stand-up comedian
1:03:40 act like a motivational coach so it
1:03:42 gives you prompts that the kind of focus
1:03:44 chat gbt and make it
1:03:47 um talk within a specific context
1:03:49 um
1:03:50 I think I'm going to add that to my
1:03:52 little uh my little photo shop thing
1:03:55 here
1:03:56 we're going to put in
1:03:58 prompts dot chat
1:04:02 yeah look at that now we got now we're
1:04:05 cooking with gas
1:04:08 that's a very Gen X
1:04:10 saying isn't it yeah we're cooking with
1:04:12 gas we're drinking out of the hose now
1:04:15 all right
1:04:16 hope that helped with the prompts is it
1:04:18 on the stock market yet is open Ai No
1:04:21 they're not public they're a private
1:04:22 company
1:04:24 uh can it understand stocks yes it can
1:04:27 write you trading Bots if you wanted to
1:04:29 they didn't dump it down they censored
1:04:30 it whatever same thing
1:04:32 the the effective result is the same
1:04:35 they lobotomized it benefits of chat GPT
1:04:38 Premium versus Bing version
1:04:41 um
1:04:43 I think Bing is still a little neutered
1:04:45 for my taste I don't I I think Bing is
1:04:47 is it's getting better but it still
1:04:50 sucks a little bit I just think chat EBT
1:04:52 is the most elegant of the large
1:04:54 language models out there
1:04:56 what are you showing us on your PC
1:04:58 screen
1:04:59 I don't know what you mean
1:05:02 um oh what are these things okay the top
1:05:05 one is chat GPT website the next one is
1:05:09 that prompt thing and then future PD is
1:05:12 just a list of tools
1:05:15 early GPT wrote my bio I've had it right
1:05:18 my bio
1:05:20 a dozen times in the past couple of
1:05:22 months
1:05:24 oh it said you were the CEO of a random
1:05:27 biotech company yeah chatgpt will lie to
1:05:29 you make no mistake it is mansplaining
1:05:32 as a service chat lies absolutely
1:05:38 all right
1:05:44 thanks I've learned more than the year
1:05:47 I've been using the a i the artist AI
1:05:51 Creator thanks you're welcome thank you
1:05:52 very much I appreciate that yeah my
1:05:54 purpose here is really just to do what I
1:05:56 can to demystify these tools share
1:05:58 anything that I know
1:06:00 point you in directions of things get
1:06:03 you curious about this stuff because I
1:06:04 think it's quite remarkable I gotta go
1:06:06 to bed soon too it is midnight and a
1:06:09 half here in Denver Colorado
1:06:12 makes I mean someone said something
1:06:15 interesting and then I lost it because
1:06:16 Tick Tock is a
1:06:20 it's a Fickle little minx Quantum makes
1:06:24 AI small enough to fit in the phone oh
1:06:27 interesting China's already doing this
1:06:30 to to to us everything
1:06:34 I don't quite know what that means but
1:06:36 but yeah I that I would I would think
1:06:39 that we're gonna probably start to see
1:06:41 large language models on chips they'll
1:06:43 build them into the hardware and we'll
1:06:44 have lots of little devices that have
1:06:47 incredibly
1:06:48 strong capabilities I think I'd use chat
1:06:51 gbt to analyze my code for
1:06:52 vulnerabilities and errors but not write
1:06:54 it for me it's great at analyzing code I
1:06:57 had a we had for my app for my company
1:07:00 um we had some obscure bug that was it
1:07:02 was an edge case where it just kept
1:07:03 crashing on this one device
1:07:05 and
1:07:07 um our developer couldn't figure it out
1:07:08 couldn't figure it out pulled it into
1:07:10 chat GPT asked what the problem was and
1:07:13 it found it like one flipped you know
1:07:16 flipped something or other that didn't
1:07:17 seem like it would be a problem
1:07:20 flip basically flipped it back ran it
1:07:23 and it fixed it fixed the bug so it's
1:07:25 it's remarkable at that stuff I let it
1:07:27 do all my assignments why not [ __ ] it
1:07:30 half of long covid it I I would say so
1:07:33 [ __ ] it on doing your assignments but
1:07:35 you know learn what it's doing like like
1:07:38 the the the the the critical skill
1:07:41 moving forward is not writing the paper
1:07:44 right I feel like writing the papers
1:07:46 days have have come and gone
1:07:48 chat GPT can write the paper but what
1:07:51 the skill is is the creative the
1:07:53 critical thinking and and the creative
1:07:55 thinking about well what's the paper you
1:07:57 want to write why do you want to write
1:07:59 that paper who's going to read that
1:08:00 paper why are they going to read it
1:08:01 right all that sort of critical thinking
1:08:04 stuff becomes way more important so if
1:08:06 you're just having it do doing your
1:08:08 assignments and not using the Noggin
1:08:09 you're you're screwing yourself
1:08:12 all right half of long covet maybe
1:08:14 something else yeah I've got kids with
1:08:15 chronic lyme disease so it could be that
1:08:17 but hopefully
1:08:19 AI will solve some of that [ __ ] too
1:08:21 because I'm crankier than [ __ ] about
1:08:23 that
1:08:24 um China's recent living off the land
1:08:27 attacks are all AI driven living off the
1:08:30 land of tax
1:08:31 powered by Quantum okay interesting yeah
1:08:33 I don't know I'll look into Quantum
1:08:35 that's really interesting
1:08:36 um we're about to see a
1:08:39 Bonkers election season because of AI my
1:08:44 goodness I love work from home as an
1:08:46 introvert I live my life in my own head
1:08:48 yeah I listen I I appreciate the the
1:08:53 introverts and that you're living your
1:08:54 best life at home but us as extroverts
1:08:57 got our asses handed to us I did a I did
1:08:59 a daily video diary for every day for a
1:09:02 year for the first year of lockdown
1:09:03 called the extra extroverts View From
1:09:06 Within it was just me whining for you
1:09:10 can find that on YouTube uh it's just me
1:09:13 whining for a year about how I didn't
1:09:15 like working from home
1:09:18 um asked gbt how to cure Cova Chet GPT
1:09:21 is going to cure a lot of stuff large
1:09:24 language models are just
1:09:26 the large language models ability to
1:09:30 parse data
1:09:31 and allow you to just interact with it
1:09:34 like a [ __ ] magician is that's what
1:09:37 it's like it's it's it's like magic
1:09:41 it's it's insane I love AI telling
1:09:44 teachers of works
1:09:46 this wait I love AI telling teacher oh
1:09:49 if work is AI or not that's going to go
1:09:51 away that that's such a short-term thing
1:09:53 because these these large language model
1:09:55 tools are going to be built into every
1:09:57 single writing tool every single
1:09:59 operating system every single app on
1:10:01 your phone it's it's going to be at the
1:10:03 operating system level so the the whole
1:10:05 thing about teachers resisting this is
1:10:07 like when they resisted the calculator
1:10:08 it's it's just kind of a useless
1:10:10 exercise right now what they should be
1:10:12 working on is okay if large language
1:10:15 models are going to be in [ __ ]
1:10:16 everything why are we teaching people
1:10:18 little tactical skills that are going to
1:10:20 be automated away why don't we teach
1:10:22 them critical thinking skills and like
1:10:24 revenge of the liberal arts Majors where
1:10:26 we need people that can think laterally
1:10:28 not vertically
1:10:30 do programmers work on improving these
1:10:33 models still or are the Bots doing it
1:10:35 probably both well like a lot of the um
1:10:37 a lot of the open source large language
1:10:40 models they're they're writing
1:10:42 algorithms to use chat GPT to train to
1:10:46 train
1:10:47 um the large language model so so it's
1:10:49 both
1:10:51 um if you code they prioritize you if
1:10:54 you ask dumb questions then no yeah I
1:10:56 agree with that the robots are strong
1:10:59 um I'm sure there's other categories
1:11:00 they prioritize
1:11:03 [Music]
1:11:06 I'm a millennial I'm a gen xer
1:11:09 um they were public for a minute
1:11:11 oh no I say that stuff I'm a millennial
1:11:14 they were public for a minute open AI I
1:11:16 don't think was ever public no they were
1:11:19 non-profit for a minute and now they are
1:11:23 a capped for-profit
1:11:25 and I heard Sam Altman in an interview
1:11:27 last week talk about the fact that he
1:11:29 has zero percent ownership in open AI
1:11:32 which I cannot wrap my head around
1:11:35 there there might be some other way he's
1:11:38 getting compensated I can't imagine he's
1:11:40 putting this much time into that
1:11:41 organization with no upside on the back
1:11:45 end I mean he might just be doing it out
1:11:47 of the good of his heart but I don't
1:11:49 think so
1:11:53 um all right let me answer a few more
1:11:55 questions and I'm gonna go to bed bye
1:11:57 all right this guy reading comments from
1:12:00 three hours ago yeah that's what I do I
1:12:02 try to I try to read everyone's comments
1:12:04 and I get behind so whatever
1:12:06 you you are welcome to leave I encourage
1:12:09 you to take off if you don't like it
1:12:11 I've been on the best I can
1:12:14 just me and a phone and a glass of water
1:12:18 I'm not allowed to drink on camera
1:12:21 ooh
1:12:24 tasty okay
1:12:26 must be a lot of work filtering all the
1:12:28 garbage on the net because at least 25
1:12:30 percent of it is bad info or opinion
1:12:32 yeah but you know
1:12:34 the better you get it prompting the
1:12:36 better that you can extract actual value
1:12:38 out of it and and you know think of it
1:12:40 like this these large language models
1:12:42 get you to 80 really quickly it's still
1:12:45 up to you to do the final refinement of
1:12:47 the work
1:12:48 has been my experience
1:12:52 uh uh uh I'm glad that you're helping
1:12:54 one another in this that's super cool I
1:12:56 like seeing that
1:12:58 all right
1:13:02 all right so rather than reading dozens
1:13:05 can provide information I'm looking for
1:13:08 immediately
1:13:10 wait my God so it's like 10 minutes I
1:13:14 why can't you drink on camera I don't
1:13:16 know because well so I've been I've been
1:13:18 kicked off of these lives twice now for
1:13:21 and it's seven like I go into detention
1:13:23 for seven days where I can't go live
1:13:26 um and I have no idea like why I went
1:13:29 like it doesn't care if I say [ __ ] or
1:13:31 [ __ ] or piss or anything like that
1:13:33 um
1:13:34 but something I said or and then someone
1:13:37 told me in the community guidelines you
1:13:39 can't drink on camera so now I'm
1:13:41 drinking off camera because maybe that's
1:13:43 it I don't I don't know I don't I don't
1:13:45 know and I don't really care so if
1:13:48 anyone wants to
1:13:49 learn the community guidelines and tell
1:13:51 me when I infract I'm fine with that but
1:13:53 I don't really give a [ __ ]
1:13:56 all right
1:13:59 let's see
1:14:05 I think I'm kind of getting to the end
1:14:07 of comments
1:14:09 this guy my name is Kyle John this guy
1:14:13 reading comments from three hours ago
1:14:15 come on don't pick on me
1:14:18 this is me just sitting here in a chair
1:14:20 talking words out of my mouth
1:14:24 extra Google John Connor wants you to
1:14:27 calm down sir I will not calm down oh my
1:14:31 machine went to sleep
1:14:33 I will not calm down I'm all peppy and
1:14:36 fired up I'm I'm live again after a week
1:14:38 away the AI and the chat GPT is the
1:14:41 beginning of the end or it's the end of
1:14:43 the beginning
1:14:46 come on gen xer2 Gen X bitter you betcha
1:14:52 all right okay I can provide the info
1:14:55 I'm looking for it's like 10 minutes
1:14:58 why can't you drink on camera I don't
1:15:00 know he caught up laughing out loud
1:15:03 all right wait we both what we both
1:15:07 this is great thank you so much thank
1:15:09 you Sadie I appreciate it yeah I gotta
1:15:10 go to bed
1:15:12 um glad I stumbled across your live can
1:15:14 AI write me a killer resume and cover
1:15:15 letter yes so here's a fun thing to do
1:15:19 um
1:15:20 just go assuming that you've put
1:15:22 together a LinkedIn profile you can just
1:15:24 literally go copy and paste your
1:15:26 LinkedIn here let's go do it I'll go do
1:15:27 mine
1:15:28 filter I just paste it in
1:15:30 everything I had
1:15:33 it looks like it cut something off but
1:15:35 whatever oh yeah that's just one of the
1:15:37 comments at the bottom okay so I'm gonna
1:15:39 say
1:15:40 um based on this
1:15:45 right me
1:15:47 a resume
1:15:50 for a pose position
1:15:58 uh in the
1:16:01 generative
1:16:04 AI field
1:16:09 um please use
1:16:12 resume best practices
1:16:17 okay
1:16:19 so then
1:16:22 it
1:16:23 took all that
1:16:26 and now it's writing me so Kyle Shannon
1:16:29 I'm going to stop generating I'm going
1:16:31 to say
1:16:32 please use markdown
1:16:38 um
1:16:39 to make it look better
1:16:42 so markdown is just basically a
1:16:46 it's kind of like HTML it's but it's
1:16:48 it's like a more simple HTML just to do
1:16:50 some basic formatting
1:16:55 and now it's stuck looks like
1:16:57 it looks like
1:16:59 chat GPT is a little sleepy tonight
1:17:04 Kyle Shannon Denver Colorado objective
1:17:07 highly experienced so email you know put
1:17:10 it here so it's formatting it for me it
1:17:12 gave me an objective
1:17:15 experience
1:17:19 so it's right now my experience so so
1:17:22 you could you could do things like I'm
1:17:24 just going to stop at generating here
1:17:26 you could do things like tell it to you
1:17:29 could copy and paste in a format of a
1:17:31 resume that you like tell it to match
1:17:33 that formatting I can now say
1:17:36 um please
1:17:38 um please write a cover letter for
1:17:44 a senior
1:17:46 position
1:17:49 leading
1:17:51 a AI oops
1:17:54 AI strategy
1:17:59 for a Pharma
1:18:03 company whatever
1:18:07 your name your address
1:18:10 today's date hiring manager's name
1:18:15 giving me all that dear hiring manager
1:18:18 I'm writing to express my interest in a
1:18:20 senior leadership position
1:18:23 my extensive experience in marketing
1:18:26 strategy has equipped me with a
1:18:27 multitude of skills I believe make me an
1:18:29 ideal candidate I've spent more than a
1:18:31 decade in the AI field that's not true
1:18:34 as Chief generative officer and founder
1:18:37 of the AI artist salon and storyvine
1:18:39 have successfully applied my skills
1:18:41 that's true
1:18:43 right so it writes that I'm going to say
1:18:45 um can you make make it more
1:18:49 conversational
1:18:52 it will make it more conversational
1:18:57 after it writes all that crap
1:19:00 um
1:19:01 [Music]
1:19:04 so the answer is yes it can
1:19:08 I hope this letter defines you well
1:19:11 to say I was excited would be an
1:19:13 understatement so I could understand you
1:19:15 know sort of basic you know language
1:19:17 requests and
1:19:19 context and things like that
1:19:21 um if you're new here this is the AI
1:19:23 learning lab I talk about
1:19:25 there's much AI stuff as I can as
1:19:28 coherently as I can even though it is
1:19:31 nearly 1am here in Denver
1:19:33 um these uh URLs the top one is chatgpt
1:19:38 the next one is prompts for chat gbt and
1:19:42 the bottom one is just AI tools if you
1:19:44 want to look around I would strongly
1:19:46 encourage you to go to chat CPT if
1:19:47 you've never used it just go there it's
1:19:49 free sign up
1:19:51 um start playing with it
1:19:52 all right just saw your AI so cool my
1:19:57 friends asking VCS oops wait uh
1:19:59 buffering uh Gen X you froze
1:20:02 am I Frozen now am I good now
1:20:07 hit the hit the screen let me know if
1:20:09 I'm frozen
1:20:11 just some feedback you're providing an
1:20:13 excellent Show Your Enthusiasm is really
1:20:15 appreciated oh thank you I appreciate
1:20:17 that very much
1:20:20 um he broke it did I break it okay and
1:20:22 my back looks like I'm back this is
1:20:24 awesome thank you I'm so glad you say
1:20:26 please to the yeah I do that too I know
1:20:29 I find myself please you know please do
1:20:31 this thank you for that
1:20:32 I'll if it gives you something good I'll
1:20:34 say that was awesome now could you do
1:20:36 this
1:20:38 um
1:20:39 it's I mean it's chat apt is programmed
1:20:42 to to give you what you want and so it's
1:20:45 programmed to be polite to you so it it
1:20:48 really is like interacting with a remote
1:20:51 employee that just happens to be way the
1:20:54 [ __ ] smarter than you are
1:20:56 um you should integrate
1:20:58 aiprm check epd prompts to to I don't
1:21:02 know what those are I'll look those up
1:21:04 I missed I missed it generative resume
1:21:08 in best what kind of format
1:21:12 um
1:21:13 I don't know whatever like go go find a
1:21:16 resume format you like and you can just
1:21:18 copy it in there I mean you're really
1:21:19 just looking for the the text of it
1:21:22 um
1:21:24 the web GPT extension okay I don't know
1:21:27 if that is
1:21:28 um
1:21:28 you don't happen to have a video
1:21:30 breaking this down do you breaking what
1:21:32 down I don't I don't know like my
1:21:34 channel is full of videos breaking stuff
1:21:36 down I have videos talking about how
1:21:39 chatgpt works I've got videos talking
1:21:41 about how to build automations in zapier
1:21:44 um lots and lots of things thanks
1:21:45 Emilio's wife
1:21:47 um
1:21:48 so I'm not quite sure what you're you're
1:21:50 referring to there
1:21:52 um not frozen okay good you're good now
1:21:54 good now good awesome how can I make
1:21:57 Chachi PD to summarize a large PDF
1:22:00 um
1:22:01 so
1:22:02 if you go
1:22:06 to gpt4 go to plugins plugins that I
1:22:11 have been using recently is this thing
1:22:12 called ask your PDF
1:22:15 and so if I put in
1:22:19 let me go let me go find a white paper
1:22:21 uh uh
1:22:26 um let's do attention
1:22:29 is all you need
1:22:32 so attention is all you need is the
1:22:35 famous white paper that where Google
1:22:39 introduced the concept of the
1:22:40 Transformer
1:22:42 and the Transformer is the thing that
1:22:45 all of these Cool Tools that we're using
1:22:47 right now are built on top of so I'm
1:22:49 going to grab that
1:22:51 URL
1:22:53 of this PDF that's out on the web
1:22:57 and I'm going to go in here to gpt4 and
1:23:00 I'm going to say
1:23:02 um please
1:23:06 summarize
1:23:09 this PDF in Five Short
1:23:16 bullet points and then I put a
1:23:20 colon and then I paste in my URL and I
1:23:23 hit return
1:23:26 and now what the
1:23:30 um plug and do is it should fire off the
1:23:34 plugin it's not like in plugins tonight
1:23:35 for some reason
1:23:37 I I think their servers are overloaded
1:23:39 right now is my sense
1:23:41 yeah it's not it's just not happy with
1:23:43 plugins
1:23:45 um anyway assuming they work then then
1:23:47 it will go it'll it'll go read that PDF
1:23:49 and and write I like I created a uh I
1:23:52 created a prompt to generate a Twitter
1:23:55 thread for my newsletter so I can take
1:23:58 any white paper or any article that I
1:24:01 like and it will automatically break it
1:24:03 down into 10 Tweets in a in a thread
1:24:06 um and and that's reading the thing I
1:24:09 don't know how if it if the PDF is
1:24:12 longer than
1:24:13 um
1:24:14 4 000 tokens I don't know if it breaks
1:24:17 it up or not my guess is that these
1:24:19 things don't because they're not
1:24:20 sophisticated but that's that's the way
1:24:22 you can do PDFs
1:24:24 all right let's see what else have we
1:24:27 got here
1:24:28 the resume sorry it should have been
1:24:30 more specific uh what was the question
1:24:32 damn it
1:24:39 um
1:24:40 I don't know what the question was
1:24:41 regarding the resume sorry about that
1:24:44 all right
1:24:45 I want some passive income to get the
1:24:47 wait oh if you want some passive income
1:24:50 get those videos on YouTube oh that's
1:24:52 good thank you oh you've got a good
1:24:54 content I appreciate that yeah I you
1:24:57 know
1:24:58 I'm doing this you know as a I wouldn't
1:25:02 call it a hobby because it's there's
1:25:03 sort of a professional angle to this but
1:25:05 I I'm not really thinking of this
1:25:07 content as a business I should probably
1:25:09 think of it more as a business
1:25:12 hey I'm the head of Applied ml stability
1:25:14 would love to chat with you about our
1:25:16 models dude
1:25:18 let's let's chat I would love to I would
1:25:21 love to talk about
1:25:23 um about what stabilities up to I think
1:25:26 I think what stability is up to is
1:25:27 really interesting I think the fact that
1:25:29 you're open sourcing everything is
1:25:31 fascinating
1:25:32 um so so I would love to talk to you so
1:25:37 um Joe Pena so um let's connect and I'm
1:25:41 going to write down your name so that I
1:25:44 can go find you
1:25:46 um but yeah I would love to talk maybe
1:25:48 we can even have you come talk at the AI
1:25:50 salon so we we uh
1:25:52 the AI Salon is at the salon.ai is the
1:25:56 URL there's a link tree there and we
1:25:58 meet every um every other week and we
1:26:00 have an online Discord
1:26:02 um we had the head of AI from Adobe
1:26:04 spoke really early on before they put
1:26:06 out Firefly
1:26:07 um he came and showed us some of the
1:26:09 some of what ended up ended up becoming
1:26:11 Firefly and I just talked to him today
1:26:14 and uh he's going to come back and talk
1:26:16 about the new generative AI
1:26:18 um the generative fill tool they put in
1:26:20 Photoshop because it's like that's
1:26:21 blowing up and so I said hey why don't
1:26:24 you come back and tell us what you're
1:26:25 learning so
1:26:26 um would love to chat chat with you
1:26:28 about that stuff so I appreciate it
1:26:29 appreciate you making the comment Joe
1:26:32 Joe Pena
1:26:34 cool that's exciting
1:26:36 see that's why I'm doing this
1:26:39 I'm doing this because we're just at an
1:26:41 exciting time right now and and
1:26:44 you know
1:26:46 learning as much as you can about how
1:26:48 these things work what they do
1:26:50 why they do what they do why stability
1:26:53 is open sourcing with their open
1:26:54 sourcing you know
1:26:56 um
1:26:57 I think if it weren't for companies like
1:26:59 stability and open AI that just
1:27:03 um are asking for forgiveness rather
1:27:05 than permission in terms of all the
1:27:07 training of the models if if people had
1:27:10 dealt with the the intellectual property
1:27:12 issues that are absolutely there right
1:27:14 you know
1:27:16 um if they had dealt with this
1:27:17 proactively like can we have permission
1:27:19 to do this the answer would have been no
1:27:21 none of this stuff would have been
1:27:22 created so so all of this stuff exists
1:27:24 because of a little bit of wild west
1:27:26 attitude that's just where we are right
1:27:28 now we're very early in this part of the
1:27:30 industry and it's super super exciting
1:27:34 um a fan in the chat pogies cool
1:27:37 um I use an ad Google Chrome so I don't
1:27:41 need type questions and hit spacebar for
1:27:43 recording I don't know what that's
1:27:44 talking about but sounds like you're
1:27:46 helping one another good I like that
1:27:49 if I have a pretty video breaking down
1:27:51 the process I have a video breaking down
1:27:54 breaking down the process of what of
1:27:56 prompting or of how chat EPT works I can
1:27:59 I can talk about that I've got to in
1:28:01 fact
1:28:03 no so somewhere in my in my videos is a
1:28:08 video that talks about how chat GPT
1:28:10 Works
1:28:11 um
1:28:12 thank you for the kind words Sadie I
1:28:14 appreciate that I obviously don't
1:28:16 participate in lives
1:28:18 do you have a video on how to do a
1:28:20 resume oh I don't I was just kind of
1:28:22 doing that live on the Fly
1:28:24 um
1:28:25 no I would um
1:28:28 I just go to YouTube and say
1:28:31 um you know um
1:28:33 using chatgpt to create a resume there's
1:28:36 tons of them there
1:28:38 um I was just kind of doing it live on
1:28:39 the flight like my interest in in this
1:28:42 is you know rather than showing you
1:28:44 exactly how to do a resume my advice
1:28:46 would be go start playing like just copy
1:28:49 and paste
1:28:50 um
1:28:50 your work experience either from
1:28:52 LinkedIn or just you know sort of type
1:28:54 it in and then have it help you organize
1:28:57 it
1:28:58 um and just tell it to
1:29:00 like like what you might want to do is
1:29:02 start out by Say by saying
1:29:06 you know chat EBT you're an expert you
1:29:09 know resume writer and employment coach
1:29:13 tell me four different
1:29:16 um types of resumes that I can put
1:29:18 together have it teach you about resumes
1:29:21 and then ask it
1:29:23 um you know
1:29:25 based on the one that you like to write
1:29:27 you a resume based on your work
1:29:29 experience and it'll do it
1:29:31 all right what are what am I primarily
1:29:35 using chat GPT for I'm using it for many
1:29:38 many things at work like almost
1:29:39 everything that I do now
1:29:43 um
1:29:45 I'm having a rewrite things I'm having
1:29:48 it help me organize things up front
1:29:52 um we're using it for Creative
1:29:54 development for the video templates that
1:29:56 we build
1:29:57 I'm creating content generation engines
1:30:01 um I'm completely rethinking
1:30:03 how my video platform does what it does
1:30:08 and what it even outputs
1:30:11 um
1:30:14 I use it for fun I use it to help
1:30:17 organize my thoughts for the newsletter
1:30:21 um
1:30:22 I don't know like it's almost like
1:30:26 the the shift that I've successfully
1:30:28 made this took me a while was whenever
1:30:30 someone presents me with some sort of
1:30:32 problem or if I come up with a problem I
1:30:34 want to solve
1:30:35 where I've historically gone is okay
1:30:38 here's how I Kyle would think about
1:30:41 solving that problem and I don't do that
1:30:43 now I'm like oh is this something chat
1:30:45 GPT could help with yes okay let's just
1:30:48 go right to chat EBT and I start to work
1:30:51 with it like a creative collaborator
1:30:54 can you use chat EBD to craft a legal
1:30:56 argument you can you need to be careful
1:30:58 some lawyer just got busted because he
1:31:00 he wrote he wrote some sort of legal
1:31:03 opinion and he didn't check the stupid
1:31:05 references the idiot and the judge
1:31:08 caught him
1:31:10 these your your the things that you
1:31:14 cited for this legal argument are made
1:31:16 up so you've got to be careful with that
1:31:19 but yeah it's really good at
1:31:20 understanding legal documents and
1:31:21 Concepts
1:31:23 it's just crazy
1:31:26 um all right let me answer a few more
1:31:28 then I gotta go to bed I gotta go to bed
1:31:31 I keep saying that it's one o'clock but
1:31:32 this is fun I like I like chatting with
1:31:34 you folks
1:31:37 um
1:31:38 I was trying to oh you were trying to
1:31:41 take screenshots as I was going through
1:31:42 the the steps with the resume yeah I
1:31:44 would say just go start playing it like
1:31:46 have it teach you have it teach you how
1:31:48 to do what you wanted to do it's it's
1:31:50 really good at that
1:31:52 um me too I use it for nearly all my
1:31:54 work tasks I'm in a Management
1:31:55 Consultant yeah like Consulting right
1:31:58 now it like there's a double-edged sword
1:32:00 I think that
1:32:02 the challenge with being a consultant
1:32:04 right now is that you know c-suite
1:32:07 people that hire Consultants are gonna
1:32:09 start
1:32:10 answering their own questions they're
1:32:12 not going to need to call the consultant
1:32:14 but
1:32:15 Consultants that are leveraging these
1:32:18 tools
1:32:20 and can turn around their thinking
1:32:22 faster and more comprehensively than
1:32:25 their competitors you're going to have
1:32:27 an advantage so if you if you're getting
1:32:28 your head around these AI tools right
1:32:30 now
1:32:31 um I would dig I I think it's I think
1:32:34 it's wicked smart and I you know I think
1:32:37 you can I think Consultants can change
1:32:39 the game the Smart Ones
1:32:42 um new here what's your video platform
1:32:43 so my company is called storyvine it's a
1:32:45 uh it's an automated video storytelling
1:32:48 platform we do a lot of work in
1:32:49 healthcare and Pharma we've been around
1:32:51 for 11 years we don't use a lot of AI
1:32:55 um in how we put our videos together we
1:32:57 use more Aristotle it's more
1:32:59 storytelling based
1:33:01 um but I've been layering in AI tools to
1:33:04 the platform and I'm completely
1:33:06 rethinking what the platform is now so
1:33:08 storyvine.com is the uh is the uh
1:33:13 is the company we do you know like if
1:33:16 like if Pfizer wants their patients to
1:33:18 talk about their disease Journey our app
1:33:20 guides that patient how to tell their
1:33:22 story goes up to the cloud five minutes
1:33:24 later you got a fully edited video
1:33:27 all right uh I've been writing
1:33:29 commercial scripts with it and writing
1:33:31 JavaScript for After Effects yeah the
1:33:33 the whole thing of writing like little
1:33:35 little scripts little code Snippets for
1:33:37 things like unity and after effects is
1:33:41 so powerful
1:33:43 muted am I muted can you hear me
1:33:46 if you can hear me tap the screen so I
1:33:48 can see the things
1:33:49 the lawyer got busted was too funny I
1:33:51 know like like here's the thing use
1:33:53 these tools but also use your [ __ ]
1:33:56 brain like look at the [ __ ] that it
1:33:58 wrote especially if you're a lawyer
1:34:00 really you [ __ ] idiot it's like Jesus
1:34:05 Christ
1:34:08 okay we can hear you thank you all right
1:34:14 um yeah so
1:34:17 all right I guess I'll leave on this so
1:34:19 so first of all thank you all for
1:34:21 hanging out with me you know just sort
1:34:23 of rambling here it's because it's late
1:34:26 so I'm not all that coherent but but
1:34:29 I'll leave you with this
1:34:31 um
1:34:32 if you haven't started playing with
1:34:34 these tools go play with these tools and
1:34:36 like play play like like think of it
1:34:39 like this we're lucky enough
1:34:42 to to to have the most powerful
1:34:45 technology like in the history of
1:34:47 mankind literally at our fingertips
1:34:51 go play with it with that sense of
1:34:53 wonder and just start pushing it the
1:34:56 more you push it the more it will reveal
1:34:58 what it can do
1:35:00 um
1:35:01 and keep your brain engaged don't think
1:35:04 of it like this these tools are gonna do
1:35:07 the work for me no no no you're still in
1:35:09 charge of what your intention is what
1:35:12 you're trying to accomplish who you're
1:35:14 trying to accomplish it for so so the
1:35:17 the critical thinking and the the
1:35:19 strategy
1:35:21 of why you would use these tools in the
1:35:23 first place that's still critical like
1:35:25 that's more critical than ever so don't
1:35:28 lose that piece but please start go play
1:35:30 going and playing with the Tactical bits
1:35:32 all right
1:35:34 um
1:35:35 I will think about putting together
1:35:38 um yeah yeah at our fingertips and free
1:35:40 mostly yeah absolutely it's we're these
1:35:42 are remarkable times we're at the
1:35:44 beginning of something incredibly
1:35:46 um powerful here so thanks all for your
1:35:49 time I'll be back I'm trying to do these
1:35:51 once a night now that I now that I'm not
1:35:53 kicked off anymore and I'm not in
1:35:55 detention
1:35:56 um and uh I will
1:35:58 um I will see you tomorrow all right bye
1:36:00 bye everybody
1:36:03 thank you