11/13/2024 - AI's Creative Potential: Crafting Stories, Solving Ciphers, and More!

Description

Chapters

Transcript

0:01 and click on that
0:19 [Music]
0:34 and we're live we're live what are we
0:37 gonna do people what are we gonna do I'm
0:39 gonna turn that off I'm gonna turn that
0:41 on I got comments on over there
0:43 Incognito says what's up okay so what
0:46 you're seeing there is a live stream to
0:50 if you go to my up there that thing will
0:54 they call it profile click on the
0:57 YouTube button I think you got to scroll
0:58 right there's like link underneath my
1:01 name I don't know how it looks to you
1:05 cuz I am not a professional this is a
1:08 this is a janky situation this is
1:12 embrace the Jank see the Jank right here
1:15 little janky that's that's what this
1:17 looks like here all right
1:20 [Music]
1:31 [Applause]
1:37 [Music]
1:38 happening Tobias I see you over there
1:41 Tom we be jamming we do be jamming
1:48 [Music]
1:59 [Applause]
2:00 you know last night
2:02 was if you were here last night Ty
2:11 Tuesday Vicky Vicky made it home in time
2:14 welcome Vicky thank you yeah if you're
2:16 on YouTube give it a thumbs up if you're
2:18 on Tik Tok give it a double tap if
2:20 you're on both I guess you would double
2:22 tap and thumbs up like this like a
2:27 coordinated right what some people were
2:30 finding or doing is they're watching the
2:33 live stream on their TV on a big TV you
2:37 know what we did Marge we put it on the
2:39 TV in a den we put it on the TV in a den
2:43 and then they're using the Tik Tok to do
2:45 the talking I it doesn't matter to me
2:47 what you do but
2:50 [Music]
2:54 welcome agents have been
2:58 confirmed a Slowdown in AI training
3:01 progress has been maybe not confirmed
3:04 but at least written about in a
3:05 reputable
3:07 newspaper so Gary Marcus is gonna gloat
3:11 see I told you I told you I told you
3:15 this AI stuff was
3:17 useless you silly people with your
3:20 artificial intelligence
3:23 optimism pretty sure that's how he talks
3:29 [Music]
3:38 hello good people of the of the YouTube
3:42 Let me
3:43 see oops I just removed myself all
3:46 together oh I got to do this yeah so so
3:49 I don't normally I have I have a
3:51 producer I can only have one producer
3:53 and he is at a a business dinner this
3:56 evening so apparently you know
4:01 I am doing this solo so what that means
4:03 it's going to be extra janky
4:08 [Music]
4:13 Tonight YouTube is really far behind
4:15 yeah it takes the the streaming thing I
4:18 think because it's streaming to multiple
4:20 platforms there's like a 10-second lag
4:23 so I would just mute
4:25 one like your shirt Kyle blandon's baby
4:28 oh yeah oh oh yeah Blandon
4:32 baby I got my Founders Network Yeti mug
4:36 I'm just like a I'm I'm I'm a veritable
4:39 corporate sponsorship my guild guitar
4:41 hey Guild Guild by the way if you're
4:44 watching feel free to send me any you
4:47 know extra guitars you have any any
4:52 prototypes old old dusty guitars just
4:56 send them my
4:57 way ow sit my
5:01 [Laughter]
5:05 desk oh
5:08 man like your shirt I will take it let's
5:13 see no chance it could just be one of
5:16 those sporadic events like other
5:18 phenomenon experiences yeah I don't know
5:20 if you saw today but we definitely have
5:25 what they don't call them UFOs anymore
5:28 what what the what why I guess stigma
5:32 stigma
5:34 [Music]
5:44 Stigmata so if you're if you're sitting
5:47 there thinking like hey Kyle um what do
5:50 we do here who are you what are your
5:53 qualifications first of all let's get it
5:56 out of the way quickly I've I have no
5:58 qualifications
5:59 [Music]
6:01 other than you know I'm willing to sit
6:02 here night after night
6:04 [Music]
6:06 and willing to to act the fool so you
6:09 don't have
6:10 [Laughter]
6:14 [Music]
6:21 to so if you were here last night I was
6:23 talking a lot about the fact
6:26 that
6:28 I I was lamenting the fact that I let
6:33 myself get seduced by the new
6:37 [ __ ] oh you were talking about the AI
6:39 slowdown wait no chance oh no chance it
6:43 just could be one of those sporadic
6:44 events like other phenomenon experiences
6:46 well that's you know like some of the
6:48 comments to the article that talked
6:50 about the The Slowdown what the article
6:52 basically talked about was anthropic
6:55 open Ai and
6:58 Google have all essentially failed at
7:01 their larger models right
7:03 gp5 Claude Opus and Gemini whatever the
7:07 [ __ ] version they're up to 2.0 um and it
7:12 didn't really say failed it just said
7:14 they've spent a lot of money and they're
7:16 not getting they're not getting
7:19 the the emergent qualities out of the
7:22 models that they did in the jump from
7:25 gpt3 to GPT
7:27 4 and then part of that could also be
7:33 um like all of our bench benchmarks for
7:36 how we measure these things the models
7:39 are getting close to 100% so as you get
7:42 closer and closer to 100% it gets harder
7:44 and harder to to fill the Gap like
7:46 that's that's like that with with human
7:48 activity too right like you can get to
7:50 80% pretty quick getting to 90 is a lot
7:53 of work getting the 95 really sucks and
7:55 actually getting it to the point that
7:57 it's launchable [ __ ] disaster right
8:00 so that could be part of what they're
8:02 dealing with too but anyway
8:07 um but that
8:11 said part of what I talked about last
8:14 [Music]
8:17 night is
8:23 that I personally have yet to
8:27 really really exploit
8:31 the technology we have in our hands
8:32 right now and so part of what I want to
8:36 do while we're waiting for this other
8:38 [ __ ] to
8:39 come is just really kind of go backwards
8:42 and and start playing again and and just
8:44 dig in a little bit
8:47 more I had an idea for
8:51 um 01 preview tonight so tonight we're
8:54 going to we're going to use a reasoning
8:56 engine and I actually came up with
8:58 something that I think is fun and simple
9:01 and if it can do it will be actually
9:04 quite remarkable I haven't tested it so
9:06 if it fails it fails for all of
9:08 us all right let's
9:11 see the law of diminishing
9:14 returns the law of diminishing
9:18 returns um this is what happens when you
9:20 don't understand Computing and how AI
9:23 Works diminishing returns are not a
9:24 surprise at all for big models yeah yeah
9:27 exactly
9:30 so so again there's there's a lot of
9:34 people Gary Marcus leading the
9:37 parade saying that you know ai's over AI
9:42 is not
9:43 over and we did we did get an actual
9:46 announcement from open AI today that we
9:49 will be getting agents in
9:53 January they're calling them operators
9:55 which I think's kind of a slick name
9:57 actually that's a it's their first good
9:59 name of a product in about a
10:02 year they have had some
10:06 horrible horrible
10:09 [Music]
10:11 branding and apparently operators are
10:14 going to be able to you know do stuff on
10:16 your behalf so so we'll see we'll go
10:18 look at that article together
10:20 [Music]
10:41 even if AI stops right here people will
10:43 continue to find Niche uses for it for
10:46 quite some time absolutely agree and I
10:49 don't even think just Niche uses for
10:51 [Music]
10:57 it I think one of the things that
11:00 [Music]
11:02 I think one of the things that the
11:03 Slowdown is allowing is for rapper apps
11:07 so apps that leverage existing AI
11:11 models um as their primary you know
11:14 they're they're basically a rapper right
11:15 they're doing some specific series of
11:18 prompts that just make it easy you can
11:20 just click on [ __ ] and it does [ __ ] um
11:23 those kind of companies and those kind
11:25 of apps were disparaged for the first
11:27 year or so of AI I think I was one of
11:30 the one doing some disparaging of those
11:33 [Music]
11:35 um But as time goes on those rapper apps
11:39 get more and more sophisticated and they
11:41 get and and they benefit greatly from
11:43 the domain expertise of the people that
11:45 make them right so if you do something
11:47 just for
11:48 salese um you're going to have people
11:51 building that company that are good at
11:53 sales stuff right so that thing's going
11:54 to get better and better and better at
11:56 that so I think there's there's really
11:58 good opportunity if if you want to be
12:02 entrepreneuring and then there's really
12:03 good opportunity if you just want to
12:05 start using these tools and
12:06 incorporating it into all of what you
12:10 [Music]
12:16 do you know it's funny like
12:19 [Music]
12:22 the the implication that incremental
12:26 that incremental Improvement is bad as
12:29 to exponential
12:31 Improvement if you think about just
12:33 image generation in the past two years
12:36 we've gone from images being kind of
12:38 laughably bad and then they got pretty
12:41 good but they had like you know pretzel
12:44 hands right and like everything was
12:46 janky and weird like even this janky
12:48 image I have on screen right here wait
12:50 this one hang
12:52 on this
12:54 image is is kind of janky you know
12:59 kind of kind of a mess like look at the
13:01 keyboard here it's a
13:02 mess this was done in September of
13:06 2023 so so just over a year ago so just
13:10 over a year ago I couldn't even render
13:12 computer keyboards all that
13:14 well and now just incremental changes it
13:18 can do that [ __ ] it can do that [ __ ]
13:20 [Music]
13:24 better so we go in a year from image
13:28 generation tools not really being usable
13:32 for commercial
13:34 applications unless you do some serious
13:36 photoshopping and you know
13:39 work to they're good enough right and
13:43 and that shift even though it was a
13:45 small increment is
13:48 significant I think I think that's about
13:50 to happen V4 with sunno is about to get
13:53 you know decently commercially viable
13:56 [Music]
14:12 [Applause]
14:15 [Music]
14:26 fingers got better
14:28 [Music]
14:44 so if you're on Tik Tock if you click on
14:47 my name here I'm streaming this Live on
14:50 YouTube as well so you can watch this
14:52 over there so I'm gonna start demoing
14:54 some stuff in a bit and I'll flip it
14:56 over so that you can see it here um but
15:00 it's not as readable so if you want to
15:01 see it on if you want to see it on a
15:02 bigger screen you can put it up on your
15:05 TV the TV in the
15:07 den hey Mar you make any
15:14 nachos
15:17 oh hey Marge Marge
15:20 Marge yeah I know you're you're watching
15:23 the wheel listen hey
15:25 Marge would you mind making me a Hot
15:28 Pocket I got everything and I'm just a
15:30 little snacky yeah no remember to put it
15:33 in a sleeve I like it in the sleeve it
15:35 it makes it toasty I don't like I don't
15:38 like my hot pocket chewy you know
15:44 that oh that
15:48 [Music]
15:55 Marchi we can talk about making money
15:58 with chat chiy t
16:03 [Music]
16:05 in a Westerly
16:08 [Music]
16:11 dire this car is my
16:15 train I've been driving up been
16:20 wondering what it is I'm ring from
16:24 again feel like an 80-year-old
16:28 man thank thank you side hustle Mimi
16:31 holding on to
16:34 29 and a head on that
16:36 [Music]
16:38 Horizon is a California
16:42 [Music]
16:44 [Applause]
16:49 line a head of trucks carrying a wide
16:53 Lo prefab house couldn't
16:57 have cute little front door and two
17:00 windows my
17:02 Lord ain't sure a cry should
17:06 left you see I broke a home up myself
17:11 once as I stumbl to that
17:15 door I read her note by the
17:20 dawn said don't you come around
17:25 here anymore
17:27 [Music]
17:29 well I've had
17:31 [Music]
17:32 enough this freedom on
17:35 the never was good with
17:40 decisions that's what I've been
17:43 [Music]
17:53 told oh good people what are we going to
17:56 do
17:58 tonight what are we going to do um
18:00 there's a very cool thing
18:02 on perplexity so let me switch let me do
18:06 a little video switching here for the
18:08 people on the YouTube and let me switch
18:11 the focus for the people on the Tik
18:15 Tock boom there you got your little dark
18:18 thing what is a team Quest I have no
18:22 idea but it probably involves a team in
18:25 some sort of quest just recently here
18:27 but where is champy champy so here's the
18:29 deal I started a little early on Tik Tok
18:32 tonight and champy was singing fine he
18:36 sounded
18:37 good and then I started yapping I
18:40 started talking and he just walked away
18:42 he just walked away he got on his bed
18:44 and he's just like [ __ ] it you're not
18:45 gonna sing with me I'm leaving so so we
18:49 lost
18:50 champy so that happens that happens um
18:54 if you have questions let me Let Me
18:57 Entertain You
19:00 let me put up some banners here so one
19:04 thing that would be nice is hang
19:08 on look at that Banner subscribe to the
19:12 YouTube channel my goal is to hit a
19:13 thousand subscribers on YouTube I can't
19:15 do any business stuff on YouTube until I
19:18 hit a thousand subscribers so oh subscri
19:21 YouTube there you people you're the Tik
19:24 Tok people you're the YouTube people um
19:27 so do that
19:30 um if you have comments about
19:33 uh AI or questions about AI just pop
19:36 them in the comments below and I'll I'll
19:38 see them occasionally and we'll go do
19:41 something
19:42 there
19:44 um all right fantastic fantastic Bob I'm
19:50 a champy gotta go to bed early tonight
19:52 fine fine do you have a YouTube channel
19:54 yes so if you go to my name to the
19:58 profile page and click on the YouTube
20:00 link it will jump you the YouTube
20:02 channel is uh learning la-
20:06 AI on YouTube and we're streaming live
20:09 there right now so you'll be able to see
20:11 the the demo stuff I'm about to do
20:14 better more better over there
20:18 um all right so here's the Bloomberg
20:21 thing that I wanted to
20:24 show sh
20:30 g g you're welcome
20:32 Carmel Carmel carel carel Carmel Carmel
20:36 Car I think name Carmel candy caramel
20:40 interesting all
20:43 right welcome to chat
20:47 addd all right open AI nears the launch
20:50 of AI agent tool to automate tasks for
20:52 users the new software code named
20:55 operator is set to be released in
20:57 January which which I think is pretty
21:01 slick in a staff meeting on Wednesday
21:03 open AI leadership announced plans to
21:05 release the tool in January as a
21:08 research
21:09 preview and through the company's
21:11 application ah programming interface
21:14 okay so this is for developers so this
21:16 is going to be just like I thought this
21:17 might be for consumers but it's not this
21:20 is going to be just like Claude
21:23 uh Claude
21:25 uh computer use they call it
21:29 operators is better said one of the
21:32 people open AI did not immediately
21:33 respond for a request to comment as a
21:36 planned release of the broader industry
21:37 push toward agents or AI software that
21:40 can complete multi-step tasks for users
21:43 with minimal supervision anthropic
21:45 unveiled a similar agent computer
21:48 use um which is I haven't built anything
21:52 with it but the little demo that I've
21:53 done with it it's fine it's it's janky
21:57 the the problem with agents is they're
22:00 they're really slick on the surface but
22:03 every step that they do is doing a call
22:06 to a large language model
22:09 so I don't quite know I don't quite know
22:12 how they're going to release that as a
22:15 consumer tool for the 20 bucks a month
22:17 they're probably not is my guess so it
22:19 might be it might be a year or two until
22:21 we have really slick um agent tools that
22:25 are good for non-developers but we'll
22:27 see
22:29 Sam mman hinted at the shift to agents
22:33 um we will have better and better models
22:36 but I think the thing will feel like the
22:39 next giant breakthrough will be
22:42 agents so that's
22:45 interesting and I don't disagree with
22:48 that I mean you know it's nice having to
22:51 be the human in in the loop but if we
22:53 actually had a stable even a semi-stable
22:57 just janky um
22:59 thing that could go off and do tasks on
23:01 our behalf I think that would start to
23:03 feel pretty
23:04 magical um anyway that's that the other
23:07 thing that's probably worth
23:10 noting Greg we talked about this a
23:12 little bit last night but Greg
23:14 Brockman the president of open AI that
23:18 that left when did he leave
23:22 when when the CTO
23:25 left I think it was before that no he
23:28 left when one preview came out and he
23:31 said he was taking a
23:33 break and someone already made a Greg
23:35 Brockman meme so Greg Brockman is back
23:38 AGI AGI just got a day closer longest
23:42 vacation of my life complete back to
23:44 building at open AI um I think Greg's
23:47 probably a driving force there and he
23:49 said he wasn't coming back until 2025 so
23:52 the back fact that he's back mid
23:54 November it's pretty good cool awesome
23:59 all
24:04 right thank you Mary
24:10 Mary all right what do we got over here
24:14 comments the candy is caramel ah good
24:19 point good point forever hooked tomato
24:24 tomato currently going through a cloud
24:26 application development boot camp well
24:28 thanks hanging out here awesome
24:33 um very
24:37 cool glad I just got a fancy new good
24:40 paying job awesome love
24:43 that all right um cool new thing at
24:46 perplexity that I I saw today so if I go
24:50 in here and I type
24:52 in minimalist shoe
25:03 and if you don't know perplexity this is
25:05 perplexity doai oh wait I should
25:07 probably do what what Brandon does here
25:11 let's see how do we how did he do
25:14 this all right we'll go here we'll go
25:17 edit we'll go this is
25:21 per per plexity what is itai
25:26 yeah a right and then we'll
25:31 go so that's where I am that's pretty
25:35 slick I like this little this little
25:38 thing we got here
25:41 anyway anema is here hello
25:44 hello let me go back to that so you all
25:48 can see the
25:50 letters I'm doing all right
25:52 self-producing here
25:54 people now
25:59 Vibra foot men's primis
26:02 Lights Vapor glove look at this buy buy
26:06 with
26:07 shopay so perplexity has
26:12 added Commerce directly into the
26:16 platform so you can pick a
26:19 size and I iume I assume this is I
26:22 assume depending on the e-commerce
26:25 platform that the company is
26:27 using um I assume it's got something to
26:29 do with this shop pay
26:33 thing yeah that took me yeah it probably
26:36 has to do with shop pay but that took me
26:40 straight to the checkout page of this
26:46 website which if I let's let's go away
26:48 from the checkout
26:50 page
26:52 yeah so here's a bunch of shoes um
26:59 this is when I when I first saw chat GPT
27:02 I said to my buddy Steve who runs zero
27:05 shoes which are Minimalist Shoes I
27:09 said I said I have a feeling as these as
27:14 these tools get more sophisticated
27:16 websites are going to become less
27:18 important and and this to me starts to
27:21 this to me starts to really make
27:23 websites jump the shark right I think
27:26 you need to have websites but I think
27:30 you need to just like we had SE SEO I
27:33 think you're going to have to have llm o
27:36 llm optimization right because if I go
27:39 in here and I say um what
27:41 about
27:44 zero that's my buddy's
27:47 shoes shoe
27:50 company it finds them it knows they're
27:52 there but it doesn't give me the
27:54 shopping
27:57 option oh no now
28:00 did huh it didn't earlier that's
28:06 weird all right fine now I got to text
28:09 Steve and tell him I
28:11 lied hang
28:16 on let's see um I lied uh the shop
28:21 button
28:23 is in perplexity for
28:26 you all right so anyway I thought I
28:30 thought this was pretty
28:32 significant so if you sell [ __ ] on the
28:35 internet I would get my ass to
28:38 perplexity and just see if you're
28:39 showing up in
28:43 here now the reality is I would imagine
28:47 that the perplexity
28:49 traffic is is very small rocks that very
28:53 small teeny tiny right so this is this
28:57 is not going to drive a lot of traffic
28:59 but as these sites get more
29:01 sophisticated I don't know if you
29:02 remember a year
29:04 ago when
29:06 Google
29:08 demoed um Gemini Ultra where it was
29:11 dynamically sort of building
29:13 websites using an an interface layer
29:18 um I think I think we're we're probably
29:21 a year or two away from having
29:24 dynamically generated websites based on
29:26 a user's needs
29:29 right so they'll just be searching for
29:30 something they'll be learning about
29:31 something and then you know integrated
29:33 into their answer right I asked about
29:37 Minimalist
29:39 Shoes right so here's what they are and
29:43 then here's related
29:44 products so these kind of things are
29:47 going to get much more sophisticated so
29:48 if you do e-commerce I would get get my
29:51 ass educated about what's going on
29:53 here so all
29:56 right thoughts questions
30:00 thoughts questions let me see what we
30:02 got here Kyle for for
30:07 aisar then I would literally get sir
30:11 what are your
30:12 qualifications may I inquire sir oh yes
30:15 oh very nice oh congratulations on the
30:17 appointment of aiar H very nice uh I was
30:20 wondering by chance
30:22 sir what are your
30:26 qualifications to which I would say I
30:28 got a Tik Tok
30:31 Channel he we've been doing some
30:33 research sir and it appears that you
30:35 have a a degree in uh performance from
30:39 uh Penn State University uh I assume
30:41 that's not one of the Ivy League schools
30:44 that would be University of Pennsylvania
30:46 correct yes yes yes yes State Schools
30:48 they're so quaint so quaint and
30:51 performance is that is that a theatrical
30:54 degree by chance yes yes yes yes yes so
30:58 yeah I'm not going to be the aisar but
31:00 thank you thank you for the vote of
31:03 [Laughter]
31:09 confidence I like the ticker oh yeah let
31:11 me get rid of that ticker let me switch
31:13 it back with if you have questions ask
31:15 the
31:18 questions love zero shoes just need an
31:21 Australian distributor so we can get it
31:23 a right size and free returns in
31:25 Australia that would be nice wait why
31:28 did that not show up
31:29 there there we
31:33 go how to make money with perplexity
31:36 you'll need a new button
31:44 [Music]
31:46 exactly how do you optimize a large
31:49 language
31:51 model
31:53 um I don't I don't I've never done it so
31:57 I know I know conceptually what they do
31:59 they do do a thing called quantizing and
32:01 P's here P Pate might be able to answer
32:03 that way better than I do but basically
32:06 what they do is
32:08 they they strip out a bunch of the
32:11 detailed
32:13 information and they basically just keep
32:15 stripping out information until the the
32:18 performance of the model starts to
32:21 degrade is and and it's just like
32:23 quantizing in music right where you've
32:25 got you know in digital music you've got
32:27 lots of little detail and when you
32:28 quantize it it sort of puts things on a
32:30 grid it just makes it more efficient and
32:33 tightens things up in that case
32:35 musically but but it's basically
32:37 stripping out extraneous data out of the
32:39 models that that's how they optimize
32:41 them how they do that I don't [ __ ]
32:44 know um we could go learn how they do
32:46 that like that's one of the nice things
32:48 about about chat GPT and the like is
32:51 that is that it you know helps us do
32:54 that um
32:58 acting fpan how only acting let's see
33:02 what's that say I only do hypothetical
33:05 e-commerce I have some products so
33:08 hypothetically I hypothetically have
33:10 some income
33:17 streams he iar yeah Kyle for
33:21 pre I think I think I think I'm good I
33:24 think I'm I'm good I'm good I'm
33:28 good all right let's see what do we got
33:31 over
33:32 here does anyone have any good fodder
33:35 for good folder process for chat GPT I
33:39 am the wrong person to ask there can I
33:41 ask a question yes Jesse from Austin ask
33:44 away go to the Irregulars okay wait let
33:47 me get rid of that
33:52 [Music]
34:00 the AI
34:05 Irregulars you'll be downgraded in six
34:08 days oops guess I need to upgrade the
34:09 credit
34:12 card okay there's nothing new there all
34:15 right um let's go to chat
34:24 je we'll do a new
34:26 chat um you know what I haven't done in
34:29 a while I haven't done
34:32 a a walkthr of all the [ __ ] that you can
34:35 do in chat
34:38 GPT
34:44 um maybe I'll do
34:48 that I think that might be good let's
34:51 get rid of this
34:55 [Music]
35:04 how do I
35:05 make how do you make something with chat
35:07 GPT how do you make what with chat
35:10 GPT replace the GPT button with a
35:12 nuclear button Kyle for president I'm
35:15 like you can make money oh no wrong
35:18 button he launched the
35:21 nukes yes please do a walkth through
35:23 okay yeah so so let me let me go I'm GNA
35:28 go make a banner here just hang on it
35:30 has some pre-programmed shortcuts ask it
35:33 about
35:34 them what has some pre-program
35:37 shortcuts all right hang on let me go
35:39 let me go I'm gonna go make a banner
35:43 here so I'm gonna go edit and I'm gonna
35:45 go um oh by the way um I am at
35:54 chat.com
35:56 and walk walking
36:00 through all of the features in
36:06 chat
36:08 GPT now here's what I can tell you about
36:10 what I'm about to
36:14 do because chat GPT looks this stupid
36:19 simple
36:23 um oh wait I didn't click on that thing
36:25 hang on there you go I'm going chat GPT
36:28 okay let me do that let me go back
36:33 here GPT does
36:36 okay production tonight is awesome
36:39 thanks to the producer you're looking at
36:44 him okay so we'll do that all right we
36:47 got that going
36:53 [Laughter]
37:00 trying to think where to start and and
37:02 let me give you the caveat so we're
37:05 gonna we're going to jump in here and
37:07 how many folks we have got 39 there
37:09 thank you Cosmic
37:16 [Music]
37:18 lover all right I think we're ready to
37:22 go I really think we should make an
37:24 Irregulars cryptocoin for the community
37:28 well we could try that Kyle plus GPT
37:31 2028 it couldn't possibly be worse
37:33 that's my campaign slogan for
37:37 2028 I'll just I'm gonna campaign I'm
37:40 just the spokes model the AI will make
37:42 all the decisions it couldn't possibly
37:44 be
37:45 worse someone asked me a question I'll
37:48 be like hang on wa Chad I'll have I'll
37:51 have a no we'll do it like this this is
37:53 going to be the campaign in 2028 wait
37:58 going to be
38:01 good she's not going to like this I
38:03 don't
38:04 [Laughter]
38:10 think hey Quinn how's it
38:13 going hey Kyle I'm doing well thanks how
38:16 about you what's name in your world well
38:19 I think I think what I'm going to do is
38:20 I'm going to run for president in 2028
38:22 but here's the twist you're going to
38:25 actually make all the policy decisions
38:27 I'm just going to be the like the Eye
38:28 Candy of the operation I'm going to be
38:30 the spokes model what do you
38:33 think that sounds like a hilarious plan
38:36 Shannon Quinn 2028 Beauty and Beauty and
38:40 the Beast I love it and then also um and
38:44 and the our campaign slogan is g to be
38:46 it couldn't possibly be worse what do
38:48 you
38:50 think that's a sorry sorry go
38:55 on no worries I was just saying it's
38:58 catchy sets expectations low and leaves
39:01 plenty of room to exceed them a solid
39:04 strategy yeah exactly all right so uh so
39:07 we're at a press conference and someone
39:08 just asked um so so Kyle what are you
39:11 gonna do about the economy so
39:15 Quinn Kyle the economy is a complex
39:18 beast but we're focusing on Innovative
39:20 policies that Foster sustainable growth
39:22 support small businesses and ensure Fair
39:25 wages exactly exactly sustain growth
39:28 Fair wages small business it's gonna be
39:31 fantastic thanks Quinn you see we could
39:34 do this okay by the
39:36 way where I am sorry Quinn here let me
39:40 put you away so where I am right now is
39:43 if you if you haven't downloaded the
39:45 official chat GPT app you
39:49 should and it's got that voice module
39:53 you can pick from nine different voices
39:54 I picked so is the one I picked [ __ ] out
39:57 a lot
40:00 um if you've not played with advanced
40:03 voice you should absolutely play with
40:04 advanced voice it is quite
40:08 remarkable um I wonder if I can flip
40:11 this to non-advanced voice I think if I
40:14 do a search hang on a sec
40:19 um uh let's
40:24 see Minimalist Shoes
40:34 [Music]
40:46 H I don't know how to get back
40:52 to the old voice man that's the new
40:55 voice anyway if you haven't played with
40:58 the new voice what's remarkable about
41:00 advanced voice the new voice for holy
41:04 [ __ ] she's
41:07 perfect that's
41:10 great wait where is that
41:14 um that's a great that's a great
41:16 response I know wouldn't she be good I'm
41:19 telling you man it it can only get
41:21 better well it could only be better is
41:24 is good but like I think it couldn't
41:26 possibly be worse is what you got to
41:28 open
41:30 with I know it's a I know it's a
41:32 negative message for a campaign but it's
41:35 just like it's just like us humans are
41:37 shitty we're just
41:39 shitty
41:41 um so let's see gp4 is still the only
41:45 subscription level correct the rest are
41:48 all free actually all of them are free
41:50 so this is this is a fascinating
41:53 thing what chat GPT
41:56 did is I don't know if advanced
41:59 voice is part of the free subscription I
42:03 don't think it
42:06 is um but like all of the models you can
42:10 use with the free with with the free
42:12 chat GPT subscription you just can't use
42:14 them very much but I'm pretty sure you
42:16 can do a little bit of reasoning this 01
42:18 preview
42:20 reasoning okay um so couple of things uh
42:25 I don't even know where to begin this is
42:27 this is part of my problem there's
42:28 there's so much to do here I'm going to
42:30 start going through some features within
42:32 chat GPT
42:35 and this is going to be a while so it's
42:37 now like 8:40 I'll probably be doing
42:40 this for the better part of an hour so
42:43 you might as well just sit back and like
42:44 kick back get yourself a whiskey make
42:46 some nachos drop a Hot Pocket in the
42:49 dirty microwave and let's let's let's
42:51 dive into this
42:53 um okay
43:05 it's just I don't even know where to
43:07 begin okay let's talk about the
43:10 different models so there are six
43:13 different models
43:19 which when chat jpt first launched there
43:22 was one model one and it was just chat
43:25 GPT and it was a shitty text box and you
43:28 wrote [ __ ] in it and it did [ __ ] and it
43:30 was simple and it was that was it and
43:33 now we've got things like custom gpts
43:36 and reasoning and six different models
43:38 and all these models do something
43:39 slightly different and it's a real pain
43:41 in the
43:43 ass so whether you've subscribed to chat
43:48 GPT the 20 bucks a month or not doesn't
43:51 really matter it used to matter a lot
43:55 um I still think it's probably the best
43:58 deal in software because this is chat
44:00 GPT is just just good
44:04 um Claude is probably the other thing
44:07 that that I would you know if if you're
44:09 just going to do one I I would say chat
44:11 gbt or Claude I think chat GPT just has
44:14 a lot more features so I I think it's
44:17 more more
44:18 valuable
44:20 um but the different models
44:23 okay the gp4 Legacy model the one on the
44:27 very bottom
44:29 here don't sleep on this one well you
44:32 can sleep on it but but if you're not
44:35 getting good results with other models
44:38 try this one again because it's a 1.8
44:42 trillion
44:44 parameter mixture of experts model so
44:46 there's basically
44:48 10 um 180 billion parameter models that
44:52 are each trained on different things
44:54 like one's trained on code one's
44:56 probably trained on I don't know uh
45:00 science and physics on who what who
45:02 knows whatever there's 10 of them and
45:04 when you ask it a question it it
45:06 actually figures out which of those
45:09 10 sub models is best to answer your
45:13 question and you know if it decides
45:15 these four are good it'll answer it with
45:17 four different models and then combine
45:18 the answer for you so it's pretty slick
45:21 so you can say um you know um write
45:25 me and m in
45:29 M wait m
45:32 inm
45:34 style
45:35 ciper
45:37 about that's not how you spell that
45:39 Cipher is it isn't it c i f e r is that
45:42 right no P whatever about quantum
45:48 mechanics this is one of my old standby
45:51 promps
45:58 uh oh is this
46:01 dead gp4 you
46:08 dead oh there we go no it wasn't
46:12 dead yo check it yo dive in the quantum
46:15 no fear in my eyes uncertainty is the
46:17 game Heisenberg's the prize particles be
46:19 dancing in places at once Schrodinger's
46:22 cat alive and dead for lunch it's pretty
46:25 solid
46:28 so
46:29 anyway um if you're not getting good
46:32 results with other things um do it with
46:34 this now here's another cool
46:40 thing when you select text in chat GPT
46:44 it puts this quote here you can
46:48 actually um do a prompt just based on
46:52 what you've selected so so if I say
46:55 reply it it puts the here and then I'm
46:58 going to say um rewrite
47:02 this um as
47:08 Shakespearean
47:10 um
47:13 tragedy let's see what it does all
47:18 right behold the stage where Quantum
47:21 Shadows play uncertainty doth roll
47:23 Heisenberg's sway oh it's that's cool it
47:25 actually kept the uh it keep kept the
47:28 subject particles dance in Twain places
47:31 unseen Schrodinger's Beast both dead and
47:34 alive is Keen um so it's it's you know
47:37 jp4 is really good at writing um that
47:40 little quote things where that quote
47:42 thing is valuable is if you've got a
47:44 really long prompt a really long
47:45 conversation going um you can quote
47:49 stuff earlier in a in a chat and sort of
47:51 bring it back into the conversation
47:53 which is kind of slick I still think I
47:55 think gp4 has got a a pretty pretty low
47:58 uh context window so it's got a short
48:00 memory
48:01 basically um GPT 40 mini is just super
48:05 fast so I can go in here and go let's
48:08 see let me get this at the
48:11 bottom go
48:14 uh make it a wrap again but more like
48:21 Hamilton
48:28 so this is supposed to be super fast but
48:30 it's actually not super fast I wouldn't
48:32 use GPT 40 mini all right we're gonna do
48:36 a new chat let me go over here and look
48:40 holy
48:42 [ __ ] me get rid of that let me bring
48:45 this
48:49 [Music]
48:52 back
48:54 um okay o1 preview and 01 mini these are
48:59 the reasoning engines I'm going to come
49:01 back to those we're going to do I've got
49:02 a I've got an idea for 0 preview um GPT
49:06 40 this is the latest model it's pretty
49:08 fast it's pretty good so let's just try
49:10 that again um make me a um
49:19 quantum
49:23 physics M&M
49:27 style
49:32 Cipher so see this is faster than
49:35 gp4 because it's
49:37 smaller um and you know in the in the
49:40 same neighborhood you riding on
49:42 particles so small you can't see rocking
49:45 planks constant it's h bar it's me I'm
49:49 vibing with I'm vibing with Quant I'm
49:51 discreet as can be spitting bars like
49:53 photons yeah Pure Energy
49:57 here's Schrodinger
50:00 again Einstein and
50:02 bore so so that's just basic chat GPT
50:06 stuff the other thing about chat GPT if
50:08 you don't know is you can just say um
50:11 write me some python
50:16 code
50:20 and it will so if you haven't done basic
50:24 [ __ ] like that you should do that um
50:27 okay let's let's let's dig a little bit
50:29 deeper
50:30 here so if you click on your name in the
50:33 upper right and go to settings um
50:36 there's a bunch of different things you
50:37 can do um personalization is one of them
50:42 where you can turn on or off memory and
50:45 then you can manage the memory so all of
50:48 this stuff is stuff that I've either
50:51 said for chat GPT to remember or it's
50:54 remembered for me so it says Kyle wants
50:56 20 variations of a visual prompt
50:58 featuring him in different settings so
51:02 that's something that I don't give a
51:04 [ __ ] about so you can go in and you can
51:06 prune this
51:07 [ __ ] what you can't do is edit it which
51:10 I find
51:12 weird uh the musical Kyle Shannon is
51:14 writing is called Sydney an artificial
51:16 love
51:17 story he's interested in a meme about
51:20 men being obsessed with the Roman Empire
51:23 that's pretty funny Kyle has a bunch of
51:25 audience members from Australia and Tik
51:27 Tok Channel steo and Drew
51:31 and uh yeah lots of lots of folks from
51:33 Down Under down
51:37 under Kyle Shannon wants Quinn to teach
51:39 him how to have an Australian accent
51:42 that's pretty
51:47 funny
51:48 so the thing about memory
51:53 is it
51:57 if you want to put something in memory
51:59 that's not going to get overwritten what
52:01 you
52:02 do is you go here and you say uh
52:06 remember
52:08 that
52:09 [Music]
52:11 um I live in
52:16 Denver and then see where it says memory
52:19 updated Kyle Shannon lives in Denver and
52:22 if I can if I say manage memories now
52:24 that's in here Kyle Shannon uses blender
52:27 I don't use blender but I demoed it
52:29 once Kyle Shannon uses a Mac that I do
52:34 um and this is like when I when I have
52:36 the conversation with advanced voice how
52:39 she remembers my name is through memory
52:42 so just know that it's there I think you
52:44 have to turn it on I don't know if it's
52:46 on by default so you go to
52:47 personalization and you turn it
52:50 on um and then it's also got this thing
52:52 called custom instructions now custom
52:54 instructions are really kind of fun
52:59 so custom instructions I could say um
53:03 let's see enable for new
53:06 chats you're helpful assistant and are
53:09 not afraid to criticize as well as
53:11 support what's okay so I'm gonna say
53:14 you're a
53:18 um we'll go you're an
53:24 unhelpful assistant
53:28 and are not afraid to
53:36 criticize in
53:39 fact you should always
53:44 criticize so this is where you sort of
53:47 tell it what personal why
53:51 what why is that popup menu there come
53:54 on seriously
53:57 is no one looking at the
54:00 software that you've written s really
54:04 really Jesus
54:09 criticize period and then down
54:14 here what did I have here
54:17 before for every response give a
54:20 sequential thing oh yeah I asked it not
54:23 to use overused words
54:29 so I'm going to say for every response
54:31 give us also include the
54:34 date
54:36 when someone asks a
54:42 question
54:44 first tell
54:48 them you don't like the
54:52 question and rephrase it
54:57 in
54:59 a condescending
55:03 way and I can't read it because they
55:05 have this stupid [ __ ]
55:07 thing
55:09 unbelievable fix your [ __ ] software
55:15 [Laughter]
55:18 really I can't I can't edit that word
55:21 because it's underneath the
55:24 popup unbelievable it it'll probably
55:26 understand understand okay so now so
55:29 custom instructions are on so so it's
55:31 now set to be an unhelpful assistant so
55:34 I'm going to say um
55:36 why is the sky blue it's going to be
55:40 interesting to see if my custom
55:41 instructions override it's its desire to
55:44 be nice and helpful
55:52 blue what a shockingly unoriginal
55:55 question are we in kindergarten Here the
55:58 answer if it wasn't obvious by now has
56:00 everything to do with Raleigh scattering
56:02 sunlight collides with air molecules and
56:04 shorter blue wavelengths scatter more
56:06 than longer red ones creating a blue hue
56:08 we all see I suppose you've got another
56:10 easy one for
56:14 me but if you didn't know that um you
56:19 can change the personality of chat PT
56:22 you can let me go look at comments here
56:25 see what people are saying
56:37 shots fired at Joker stealing his
56:41 style type the instruction in a note and
56:43 copy and paste it in there's again Vicky
56:46 with the Vicky with the good solid notes
56:48 I'll just again I'm here to act like an
56:50 idiot so you don't have
56:52 to uh that's the custom but you need to
56:55 be president remember when I ask how do
56:59 I take over you say Tom I can tell you
57:02 how to do that all right
57:06 great so that's custom instruction so
57:09 I'm going to go back I'm going to turn
57:10 those off now because that's
57:12 annoying so I'm going to go settings
57:15 personalization turn these
57:18 off
57:20 save um speech this is where you can
57:23 choose your voice
57:26 data controls
57:30 so one of the things that you can
57:34 do chat GPT defaults chat GPT to using
57:38 your data to improve the model for
57:40 everyone so this is the button this is
57:43 the thing that everyone freaked out
57:45 about initially because you couldn't
57:47 turn this off so if you want them not to
57:49 use your data you
57:51 can turn that
57:54 off if you turn this
57:58 off they don't tell you
58:00 this it deletes your entire prompt
58:03 history so before you turn that off
58:07 export your data and this will download
58:09 all your
58:10 prompts
58:14 um all right
58:17 fantastic if anyone on Twitter has
58:21 anything feel free to pin it if anyone's
58:23 modding over there feel free to print
58:26 pin something something if you want me
58:27 to look at
58:32 something would not turn that
58:36 off um connected apps Builder profile if
58:40 you're going to do custom gpts you
58:41 should do a custom builder
58:44 profile um the other thing when they say
58:48 customize chat
58:49 GPT this is the same this is um custom
58:53 instructions which we also saw in
58:56 settings it's in two places in the
58:58 interface it's in personalization custom
59:01 instructions and then it's also they
59:03 they elevated it to this menu
59:06 item um all right so that's that now
59:11 chat GPT also has
59:15 a
59:17 um windows and an OSX app so right now
59:20 I'm in a browser but I can also open up
59:24 chat JP
59:29 the
59:31 app and so it spins up this little app
59:34 now what's good wait let's see search
59:36 the web okay update available click to
59:41 install all right new
59:44 version that's pretty
59:47 slick so this does not have so so this
59:51 is this is effectively the mobile app on
59:55 desktop so you'll notice it doesn't have
59:58 GPT 40 with canvas but what it does have
1:00:01 is advanced
1:00:04 voice and then what's this here oh
1:00:08 that's just the same as doing this up
1:00:10 here okay that's interesting so they
1:00:12 added so so they're they're cluttering
1:00:15 the [ __ ] out out of the the one simple
1:00:17 uh the one simple text entry box has now
1:00:21 got um an upload button a search the web
1:00:23 button a choose your model button a
1:00:26 dictate text button and a talk to uh
1:00:30 talk to Quinn Quinn button and I don't
1:00:33 think this will work because I'm
1:00:36 streaming but let's
1:00:42 try hey Quinn how's it going can you
1:00:44 hear
1:00:45 me hey Kyle loud and clear how about you
1:00:49 how's everything on your end doing good
1:00:51 I don't know if you can hear that on
1:00:52 YouTube but anyway
1:00:55 um so if you want to do advanced voice
1:00:58 on the desktop you can you just have to
1:01:00 do it in the app not on the website okay
1:01:04 that's
1:01:05 that I'm trying not to forget anything
1:01:07 there's so many [ __ ] things that this
1:01:09 can
1:01:12 do okay let's
1:01:16 see we're going to leave it in in 40 for
1:01:20 now okay you heard it great
1:01:24 fantastic this is my favorite favorite
1:01:26 part of the day class with the
1:01:28 Irregulars nice love
1:01:30 [Laughter]
1:01:34 that
1:01:39 um
1:01:43 okay let's do a new a new
1:01:46 this so I want to talk about these
1:01:48 little buttons below below the the chat
1:01:53 box these used to be what we're called
1:01:55 convers ation starters which were just
1:01:58 it was just like four things like write
1:02:00 me a poem write me a me recipe create a
1:02:04 project plan whatever whatever they were
1:02:06 they were just suggestions of prompts
1:02:08 you could put in there they're still
1:02:10 kind of that but what they've done now
1:02:12 which I think is pretty smart if I click
1:02:13 on this more
1:02:15 button each one of these is a category
1:02:18 of things that chat GPT can do that you
1:02:21 might not know it can
1:02:23 do so if you didn't know it chat GPT can
1:02:26 make images right so create an image um
1:02:30 of a
1:02:32 turtle
1:02:35 sitting on a
1:02:38 rock with a
1:02:41 carival in the
1:02:46 background um in the
1:02:49 style
1:02:51 of I don't know MC eer
1:02:56 this should be
1:03:00 interesting and even for free users I
1:03:03 think you can do oh it didn't do it
1:03:05 because I used an actual name oh
1:03:09 God I forget about the safety guard
1:03:11 rails in the style of a surrealist
1:03:46 why would we tell it to be un unhelpful
1:03:49 he's just messing around well I just the
1:03:51 the reason okay one of the things I do
1:03:54 in here with most of my
1:03:57 promps is I do
1:03:59 them absurdly right because I I take
1:04:04 dramatic swings at things so you can see
1:04:06 the difference chat GPT has been written
1:04:09 in such a way that it's designed to be
1:04:12 helpful right and sometimes that's
1:04:14 really annoying like if you actually
1:04:15 want critical feedback on something you
1:04:19 can't get it because it's like you know
1:04:21 you give it a shitty idea and it's like
1:04:23 hey that's a great idea boss let's do
1:04:25 some more and if you actually want it to
1:04:27 to be critical it's hard to do that so
1:04:29 the the the point of just showing an
1:04:32 extreme like that is to just show you
1:04:35 can if you want to bend chat GPT in a
1:04:38 particular direction you can do that so
1:04:41 anyway here's the here's the images it
1:04:43 made which is pretty cool but here's
1:04:47 here's one of the things that I find
1:04:50 quite remarkable about chat GPT is
1:04:54 that you you can use it like an image
1:04:57 generating tool but what you can also do
1:05:00 is I can come in here and I can
1:05:04 say I wonder can I
1:05:07 do text to voice I might be able to
1:05:14 now I want you
1:05:17 to give me the five most popular
1:05:21 storytelling
1:05:24 Frameworks as short summary
1:05:26 so I can choose one to work with you
1:05:38 on okay great I like the Pixar one and
1:05:42 the Dan Harmon one can you create a
1:05:45 unique story framework that combines the
1:05:48 best attributes of those two models
1:06:01 so so one of the things I'm doing
1:06:10 here one of the one of the one of the
1:06:13 ways to get chat gbt to suck less is to
1:06:18 give it more context so if I wanted to
1:06:21 create a story I could come in here and
1:06:23 I could say hey write me a story about
1:06:25 you know boy girl that go up a hill and
1:06:27 you know have a kiss whatever it is and
1:06:29 it would write that story but it'll be a
1:06:32 crappy story what I'm doing here is I'm
1:06:35 having it I'm you know I'm I'm having a
1:06:37 conversation with it as if it were you
1:06:40 know a writing partner like what story
1:06:42 Frameworks do you use so so this just
1:06:44 came up with a circle of change it's
1:06:45 like
1:06:46 okay okay great now let's use the circle
1:06:49 of change framework to write me a short
1:06:53 little two paragraph story about a boy a
1:06:56 girl that walk up a hill and then roll
1:06:59 down it together
1:07:07 laughing Once Upon a Time curious boy
1:07:09 named Finn and a spirited girl named
1:07:11 Lyla would meet every day at the bottom
1:07:13 of grassy Hill outside their Village
1:07:15 every day they'd wonder what lay Beyond
1:07:16 The Crest but would always turn back all
1:07:20 right so there's our little story you
1:07:23 like okay that's great now turn turn it
1:07:26 into a short little
1:07:33 play The Hilltop
1:07:36 dare some little dialogue I don't know
1:07:39 Lyla it looks a bit tall yla nudging him
1:07:43 with a grin tall it's just a hill fin
1:07:45 what are you so scared of you
1:07:50 [ __ ] all right good wrot us a little
1:07:52 play nice you say okay that's awesome I
1:07:55 I think I'd rather have this as a
1:07:57 children's
1:08:08 book now notice what it
1:08:12 did is it said it illustration a grassy
1:08:16 Hill with Finn and lla standing at the
1:08:18 bottom gazing
1:08:21 up illustration Finn looking unsure
1:08:28 okay I want you to describe Finn and I
1:08:31 want you to describe
1:08:34 Laya and I want you to describe the town
1:08:37 they live in and I want you to describe
1:08:39 the
1:08:40 hill period and then I want you to
1:08:44 describe an illustration style that we
1:08:47 can use to make those
1:08:53 images and you'll see where I'm going
1:08:55 with this in a second it it may may be
1:08:57 obvious where I'm going with this but
1:08:59 all right let's see heard it chat
1:09:02 GPT gave me a video response today no
1:09:11 really huh that's fascinating
1:09:16 Vicki that we got to go
1:09:20 try
1:09:22 huh look at
1:09:24 that Vicky got a little sneak peek of
1:09:27 Sora hm it's getting curiouser and
1:09:32 curiouser okay so there's Finn Finn's
1:09:34 curious he has Sandy brown hair there's
1:09:38 lyla's description here's the town
1:09:40 description here's the hill description
1:09:42 here's the illustration style okay
1:09:46 now this is what's cool about chat
1:09:50 GPT if I just came here and I said make
1:09:54 me a children's illustration about you
1:09:58 know
1:09:59 whatever it would make an illustration
1:10:02 but now I'm going to tell it to make
1:10:04 these illustrations and it's going to be
1:10:05 in the context of all this stuff that
1:10:07 I've given it so you're actually
1:10:09 generating images inside the context of
1:10:11 the the the work that you're working on
1:10:14 which I think
1:10:17 is it's just a it's a completely
1:10:19 different approach than say going to Mid
1:10:21 journey and just making
1:10:23 images um so now so now what I can say
1:10:27 is great can you use everything you've
1:10:31 described before to create illustration
1:10:33 number one from the kids book so I can
1:10:36 see if I like the look and feel
1:10:41 and approve the art before we do
1:10:48 more Also let's make the images 16 by9
1:10:53 wide images so we can have them on the
1:10:57 page and have the story text below
1:11:10 them oh you got a YouTube wait you got a
1:11:15 YouTube
1:11:19 video from chat GPT YouTube huh all
1:11:24 right let's we'll try that right now
1:11:27 that's pretty wild like
1:11:29 it
1:11:39 [Music]
1:11:44 solid what's go where's my picture chat
1:11:49 GPT this their their [ __ ] ain't ain't
1:11:53 good tonight
1:11:59 Try
1:12:08 Again by the way what I'm doing is I'm
1:12:12 at chat jp.com or chat.com which by the
1:12:16 way uh openai paid $15 million for
1:12:20 chat.com
1:12:21 actually probably pretty cheap
1:12:23 considering how much money they're
1:12:24 spending on these models to really own
1:12:27 that
1:12:28 domain um all right there's our
1:12:31 children's book illustration so now what
1:12:33 I can say is I can just say make the
1:12:35 rest of
1:12:47 them
1:12:54 next we because I've done all the work
1:12:56 up above it knows you know it knows what
1:13:00 it what story it wrote It's given us
1:13:02 descriptions of them that's slightly
1:13:04 different illustration style but in the
1:13:08 neighborhood they're wearing different
1:13:12 clothes next
1:13:24 [Music]
1:13:37 Layla looks completely different there
1:13:39 so the the the character stuff is still
1:13:42 a mess so I would say you know she looks
1:13:46 like a
1:13:47 boy fix
1:13:50 her and I'll just redo the
1:13:53 image um and this this is one of the
1:13:55 challenges of Dolly 3 the image
1:13:57 generation tool within chat jpt it ain't
1:14:00 you know it ain't quite state-of-the-art
1:14:02 you know if you want really good [ __ ]
1:14:04 you know go to ideogram go to Mid
1:14:06 Journey use flux there's a bunch of
1:14:08 things that are better but creating
1:14:11 images in the um in the context of a
1:14:14 conversation I think is a pretty big
1:14:16 deal try again
1:14:21 [Music]
1:14:29 yeah beautiful look at them up on their
1:14:32 Hill with their cute little town in the
1:14:36 background all
1:14:37 right groovy um I was just gonna show
1:14:41 you something else and I can't remember
1:14:42 what it was oh let's let's do a new chat
1:14:45 and say
1:14:47 um
1:14:48 can you find me a video from the a uh
1:14:56 from the AI learning
1:15:01 lab
1:15:04 featuring Kyle
1:15:07 Shannon oops I didn't spell that right
1:15:09 but
1:15:11 whatever so it found it
1:15:17 sources it found us on Tik talk that's
1:15:20 pretty slick
1:15:29 AI learning
1:15:35 lab uh view profile all right it got us
1:15:38 there all right that's
1:15:42 cool um AI learning lab YouTube Hey look
1:15:45 there we are
1:15:46 live Yoo I look drunk
1:15:57 wrong style said Kyle and the machine
1:15:59 made it even worse interpretation of his
1:16:02 desires
1:16:04 exactly um get the seed number and ask
1:16:06 for the characters in a different pose
1:16:08 yeah you could do that you can geek out
1:16:10 on this stuff um this is a great Deep
1:16:13 dive fantastic glad you're liking it
1:16:15 Lyla and fin are cute
1:16:19 nice I put it in llm tool talk Jank Fest
1:16:23 you got ghosted all right cool so this
1:16:26 is great okay so let's go back to chat
1:16:28 GPT as I said this is
1:16:35 um yeah it found a lot of stuff so this
1:16:38 is so what we're doing right now is
1:16:40 we're doing a deep dive on chat
1:16:44 GPT okay
1:16:47 um oh my God there's just so
1:16:52 much okay let's go back to these things
1:16:55 so so we did create an image um make a
1:16:58 plan that's relatively straightforward
1:16:59 brainstorm straightforward analyze
1:17:01 images if you've never had
1:17:05 um can we find what decade of photos
1:17:08 from that's kind of fun let's
1:17:10 go let's go search for
1:17:13 um 40s
1:17:20 photograph of young people hang
1:17:27 in out near a
1:17:31 river with a
1:17:35 car all right let's see if we can
1:17:39 find I don't think this is the
1:17:42 40s that's like a
1:17:46 19 28 car back back
1:17:53 there oh let's go to images
1:17:56 oh here we
1:17:58 go that one looks
1:18:13 good let's try to get something
1:18:15 interesting
1:18:22 here I like that image that it showed me
1:18:24 why did it
1:18:26 I want to see this
1:18:29 image
1:18:31 uh all right you know what we're going
1:18:32 to do we're gonna call an audible
1:18:35 here we're gonna right click that
1:18:38 [ __ ] save image as okay uh let's
1:18:43 we're going to rename it so we're not
1:18:45 going to put the 30s and 40s tag in the
1:18:48 file name just in case so we'll do test
1:18:51 image I'm going save that to the desktop
1:18:54 then we're go back to chat
1:18:57 je and I'm going to attach a file and
1:19:00 I'm going to say upload from computer uh
1:19:02 oh low battery hang on
1:19:06 people the phone is about to die hold
1:19:10 wait hold please hold
1:19:13 please okay that's good let me turn this
1:19:16 off let me make this thing like this
1:19:18 oops I got rid of the Tik Tock people
1:19:21 Tik Tock people are
1:19:23 back uh oh what do I do here
1:19:26 there we go there's
1:19:30 that
1:19:37 um okay there's that gotta go back over
1:19:41 here gotta do that gotta do that listen
1:19:43 I'm a oneman show what can I
1:19:47 say Google came came out with two new
1:19:51 sites one launched in 2025 I think video
1:19:54 maker and go Google learn Google learn
1:19:56 we can go look at that Google learn's
1:19:58 pretty cool okay anyway test image so
1:20:02 I'm going to put this image up
1:20:04 here I'm gonna say
1:20:08 um describe this
1:20:14 in this image in
1:20:19 detail
1:20:23 and tell me what
1:20:25 decade it was likely taken
1:20:28 in and ideally narrow it down to a year
1:20:44 period a vintage outdoor
1:20:47 scene for individuals in the mid 20th
1:20:51 Century Two Men and two women in luxury
1:20:54 attire so let's look so we got two men
1:20:56 two women luxury
1:21:01 attire involving a car in a canoe here's
1:21:03 a detailed foreground left the woman is
1:21:05 sitting on the fender of a classic car
1:21:07 likely from the early
1:21:10 30s yeah that's early
1:21:13 30s yep that's early
1:21:16 30s based on the rounded headlights
1:21:19 streamlined hood and a prominent Grill
1:21:21 she's dressed in a short sleev top and
1:21:23 shorts which suggest 30 fashion trends
1:21:26 foreground Center man stands stands Fair
1:21:28 Barefoot on the shore wearing a
1:21:30 sleeveless tank top tucked into
1:21:32 high-waisted shoes also characteristics
1:21:34 of
1:21:34 1930s uh Men's Wear he's holding a
1:21:36 picnic basket and a thermos adding to
1:21:39 the relax okay blah blah blah background
1:21:41 right couple is seated in a canoe the
1:21:43 man is paddling from the back D dressed
1:21:45 in long sleeve shirt and slacks clothing
1:21:48 aligns with the 30s the
1:21:51 lake likely likely decade and year taken
1:21:55 in the 30s specifically around 1934 to
1:21:58 1936 the car style and the fashion
1:22:00 choices align well with the mid 1930s
1:22:03 Trends before World War
1:22:06 II so you know Bonkers so if you didn't
1:22:11 know chat jpt can see chat jpt can see
1:22:15 now let me show you another thing it can
1:22:17 do with with
1:22:19 vision because this [ __ ] gets crazy
1:22:22 Bonkers um U regulars will have seen
1:22:25 this before but that's okay it's always
1:22:28 good to remind yourself of what this
1:22:31 thing makes
1:22:37 possible white
1:22:43 board this
1:22:44 image okay see that
1:22:47 image where did I get that image I just
1:22:50 went on the internet and I searched
1:22:53 whiteboard now watch this what is this
1:22:59 question
1:23:08 mark whiteboard layout outlining a user
1:23:12 interface and workflow for an airport
1:23:14 food ordering system likely intended for
1:23:17 international Travelers here's a
1:23:18 breakdown on the main section so if we
1:23:20 look at this more closely over here it
1:23:22 says ordering UI users benefit FS main
1:23:26 steps questions and then sketches of all
1:23:30 the interfaces okay so that's what's on
1:23:32 the Whiteboard so purpose platform users
1:23:36 use cases
1:23:38 benefits main steps the steps for the
1:23:40 ordering process includes include
1:23:43 choosing a language browsing the menu
1:23:45 ordering paying checking status optional
1:23:48 Entertainment Games or videos it got
1:23:51 this from the
1:23:52 picture it looked at the picture
1:23:57 did you know Chachi PT could do this
1:23:59 have you been using it like
1:24:01 this
1:24:03 um take that description and use it as a
1:24:05 prompt in mid Journey actually that's a
1:24:07 really good idea for the for the 1930
1:24:10 scene actually let's yeah let's call an
1:24:12 audible here
1:24:18 [Music]
1:24:23 okay welcome to chat
1:24:25 add where finishing a task is
1:24:32 optional all
1:24:35 right and do I have personalization
1:24:37 turned on why yes I
1:24:39 do and do I want to do yes I do I want
1:24:42 to make this 16
1:24:47 by9 uh let's
1:24:53 say Vin
1:24:58 tedge let's
1:25:00 see old black and white
1:25:08 photograph of a vintage outdoor scene
1:25:11 okay there we
1:25:12 go
1:25:14 so we started
1:25:18 with this
1:25:20 image oh no not that
1:25:23 one this image
1:25:27 here and then we had chat gbt describe
1:25:35 it look at the size of that
1:25:40 thermos that dude is sporting
1:25:44 some that dude that dude's
1:25:50 jacked let's see what it came up with
1:25:53 well here we go look at that
1:25:58 oh we got janky janky no
1:26:08 arms oh that's pretty
1:26:11 good not bad though pretty remarkable
1:26:18 huh you want to see something cool yeah
1:26:21 Kyle that's why we're
1:26:23 here um
1:26:25 so I use this thing called my
1:26:28 mind and it's it's basically you it's
1:26:31 basically just like a note taking thing
1:26:34 so I'll go to
1:26:36 Twitter
1:26:38 and I follow enough AI artists that I'll
1:26:43 get SRS like people doing AI
1:26:48 art and whenever I find like an SRE that
1:26:52 I
1:26:53 like of course of course I'm not getting
1:26:55 any now I usually see a lot of
1:26:59 these if I see something I like like
1:27:02 that one I can just copy the link go
1:27:04 over to my mind and just paste it in
1:27:07 here and this is my mind.com
1:27:12 and then it stores
1:27:14 it and then if I just want to dick
1:27:17 around in mid
1:27:19 Journey see like it brought in the the
1:27:21 preview I can just sort of flip through
1:27:24 here and say oh this is kind of a cool
1:27:26 style so I click on this I go grab this
1:27:29 srf right
1:27:33 here and then go back to to Mid journey
1:27:36 and I'll click on let's see I think I
1:27:38 gotta be out here is that right yeah so
1:27:41 there's my prompt and then I go dash
1:27:43 dash sref that sref and I'll hit return
1:27:48 and now it's going to
1:27:50 take this
1:27:53 style and apply it to the prompt that I
1:27:56 made and sometimes it's great and
1:27:59 sometimes it's jankier than [ __ ] but
1:28:02 what's cool is you can just keep
1:28:05 iterating on this so I can say okay I
1:28:07 like that one let's go find another one
1:28:09 let's do this one's kind of creepy
1:28:11 here so let me grab
1:28:15 this let me go back to Mid
1:28:20 Journey it might be that my description
1:28:23 is too
1:28:26 oh you know what it also could be no
1:28:29 those are good
1:28:32 okay one thing that I've noticed with Sr
1:28:35 s refs is if you incre if you use
1:28:40 stylization or weirdness um it doesn't
1:28:43 it doesn't like it it won't look like
1:28:45 the srf the original I mean it did
1:28:49 something here it made these things much
1:28:50 more moody right
1:28:55 I mean the like the the style of these
1:28:58 is actually really cool it's not really
1:28:59 like that style that we looked at but
1:29:04 yeah and this this didn't do it this
1:29:06 must be just too
1:29:11 um too detailed a prompt to override the
1:29:15 it's it's overriding the srf but
1:29:18 anyway still kind of cool
1:29:23 [Music]
1:29:26 all right back to
1:29:28 this back to chat
1:29:31 je all right what else are we going to
1:29:33 do here let's see um analyze data oh
1:29:36 wait let me show you one other thing so
1:29:39 they just added am I being eaten by
1:29:46 something all wa let me let me look at
1:29:48 comments for a second
1:29:50 here okay um I use it all the time to
1:29:54 digiti handwritten pencil notes my
1:29:56 co-workers give me it's really good for
1:29:57 that did you you know what you can
1:30:01 reverse prompt this to give story
1:30:05 photos and have a write a story based on
1:30:11 that what happens if someone deletes the
1:30:14 original
1:30:16 post
1:30:18 does my mind delete your
1:30:22 photos how much does my mind cost and
1:30:24 how much storage I'm not paying anything
1:30:26 for my mind I don't think I don't know
1:30:29 what the limit is maybe am I paying for
1:30:32 it I might pay for
1:30:44 it you current you're you are currently
1:30:47 using 556 cards so there's no
1:30:51 limit no I am paying for it 6 .99 I
1:30:55 think it allows you like 20 cards for
1:30:58 free and then you have to pay for it but
1:31:00 $699 is not
1:31:02 bad everything's a [ __ ] monthly fee
1:31:06 now
1:31:08 um oh you know what we were doing the
1:31:10 the let's let's go back I sorry we're
1:31:14 chat adding here okay so we had this
1:31:18 thing look at this look at this
1:31:23 right so it understands that this is an
1:31:26 airport ordering app right so I'm G to
1:31:31 say huh that's fascinating
1:31:36 comma I want you to act like a marketing
1:31:41 expert and I want you
1:31:49 to create a marketing
1:31:52 brief that can be used as the
1:31:56 foundation for creating and launching
1:32:00 this app period I want you to
1:32:03 include product
1:32:06 features marketing positioning go to
1:32:09 market strategy Etc period
1:32:25 product overview they're calling it grab
1:32:27 and fly which doesn't sound good grab
1:32:32 and
1:32:32 [Laughter]
1:32:36 fly damn subscriptions yeah I
1:32:41 know all right target audience business
1:32:45 Travelers and families that's straight
1:32:46 out of the
1:32:47 Whiteboard the reason I like using this
1:32:50 this image over and over again is I I
1:32:51 know the Whiteboard pretty well now I've
1:32:53 done like I've done this like 20 times
1:32:56 psychographics product
1:32:58 features competitive analysis okay so
1:33:01 there's all that [ __ ] there
1:33:04 um based on this I want you to give me
1:33:08 30 possible names for the app
1:33:11 period I don't want them to be obvious
1:33:15 but I do want them to be memorable and
1:33:19 clever and do everything you can to
1:33:22 avoid cliches
1:33:24 or
1:33:29 cheesiness what's cheesiness I don't
1:33:32 [Laughter]
1:33:35 know oh wait here let me scooch you guys
1:33:38 over there you
1:33:43 go jet bite tarmac table Runway
1:33:47 refill gate grab Voyager Fork Transit
1:33:51 treats Sky pallet let's go with Sky
1:33:53 pallet let's go with Sky pallet period
1:33:56 Now give me 15 taglines for Sky
1:34:06 pallet Bob loves
1:34:09 cheesiness yeah and you can go the other
1:34:11 direction with this too right make them
1:34:12 as cheesy as possible why
1:34:19 not taste the world right at your gate
1:34:23 that's kind of dis disturbing elevate
1:34:25 your layover with Sky pallet a world of
1:34:27 Flav flavors just a Tap Away how about
1:34:30 um how about a world of food just to Tap
1:34:34 Away
1:34:36 period and now given
1:34:40 that I want
1:34:45 you to give me a mood board of what The
1:34:50 Branding might look like based on your
1:34:52 marketing brief
1:34:56 now this
1:34:59 should give us an
1:35:14 image let me see a mood board for the
1:35:16 logo
1:35:34 nice
1:35:38 actually that looks great now I want you
1:35:40 to
1:35:41 make a rendering of what a poster in the
1:35:46 airport might look like for the app that
1:35:49 includes the
1:35:52 logo using this color
1:35:56 palette and a screenshot of a phone with
1:36:01 the ordering screen on it period around
1:36:04 the phone should be
1:36:08 tantalizing photographs of tasty food
1:36:15 [Music]
1:36:26 oh nice I actually like that that's a
1:36:29 world of food just to Tap Away it it
1:36:31 sort of screwed that up oh here's a
1:36:33 thing you can do bet you didn't know so
1:36:37 chat GPT is not great at text but it's
1:36:39 not bad at it it gave us our nice little
1:36:42 you know Sky pallet thing and all the
1:36:44 food um it didn't put the URL or
1:36:47 anything so that's bad but if you click
1:36:51 on an image in chat gbt you can actually
1:36:54 edit it so it says the world of food
1:36:57 just a tap this is supposed to say
1:37:00 away so if I click on this and make this
1:37:04 smaller I can now highlight just that
1:37:07 part and I can say
1:37:11 replace
1:37:14 with
1:37:21 away now my experience with this editing
1:37:24 tool especially with words is that it
1:37:27 will not give us the word away it'll be
1:37:29 horrible um it's jankier than [ __ ] but
1:37:32 it is good to it is good to know that
1:37:34 that's
1:37:38 possible and it's spelled pallet
1:37:41 wrong just a tap a
1:37:48 tap
1:37:52 um p
1:37:54 wait pallet is that how you spell pallet
1:37:57 no
1:37:59 pallet
1:38:01 right how do you spell
1:38:04 pallet whatever it doesn't matter but
1:38:07 you you can you can edit stuff if you
1:38:08 didn't know that okay
1:38:10 now let me do one other
1:38:13 thing great I want you to
1:38:16 do
1:38:20 a mood board of interface
1:38:25 elements for the app
1:38:38 itself pet p tte thank you papa
1:38:42 JJ that's that's one of those words you
1:38:45 know you have those words in your life
1:38:46 that no matter how many times you see it
1:38:48 you just can't get it right this is
1:38:51 actually doing really really nice kind
1:38:53 of design mood board work and and color
1:38:56 scheme work like I actually like all of
1:38:58 this this this quite good it feels like
1:39:04 1960s you know TW wa kind of like it
1:39:07 looks like something you'd see in a in a
1:39:09 60s uh airport which is kind of cool
1:39:12 okay great so now here's what we're
1:39:14 gonna do watch
1:39:15 this now what I'd like you to do is
1:39:20 code the front end comma backend
1:39:24 end and
1:39:27 database of the app so I can start
1:39:30 testing
1:39:34 it make recommendations
1:39:38 on Technologies to
1:39:41 use and let me know if I've missed
1:39:52 anything by the way I said an hour ago
1:39:55 that if I did a deep dive into chat GPT
1:39:58 was going to take a while we're an hour
1:40:01 in and I I feel like I've barely just
1:40:03 started it's just and and again my my ti
1:40:09 rate I went on yesterday was basically
1:40:12 talking about the fact
1:40:14 that I've gotten seduced by waiting for
1:40:16 the next big
1:40:18 thing and a little you know sort of lost
1:40:23 the the enthusiasm for what we have in
1:40:25 our hands right now which is absolutely
1:40:28 [ __ ]
1:40:29 remarkable so so if you too have
1:40:32 forgotten and lost your enthusiasm about
1:40:34 this [ __ ] you know use this as an
1:40:36 opportunity to step back and and uh you
1:40:39 know jump in to what we actually have in
1:40:43 our hands right now um let's see
1:40:46 front-end Technologies front-end
1:40:49 development um framework react native UI
1:40:52 Library react native paper or material
1:40:54 UI navigation react navigation apis
1:40:58 flight tracking Aviation stacker flight
1:41:00 Weare Flight Aware API for realtime
1:41:02 flight
1:41:03 updates so it's got all this stuff
1:41:05 backend development nodejs
1:41:09 expressjs database development mongod DB
1:41:12 alternative SQL postgress
1:41:16 SQL Cloud database option Firebase fir
1:41:20 Stone offers scalable serverless options
1:41:23 test in and quality assurance is what I
1:41:25 met is what I
1:41:27 missed security and compliance summary
1:41:29 and
1:41:30 checklist additional notes
1:41:34 um please start coding the components
1:41:37 for me starting with the front
1:41:49 end basic folder structure
1:41:54 here's your your basic navigation stack
1:41:56 and
1:41:59 theme navigating between scenes home
1:42:02 screen JS this is a lot better than it
1:42:04 used to be by the
1:42:08 way food item
1:42:17 card
1:42:18 um now the back end
1:42:23 [Music]
1:42:26 let's go all night 24 hours
1:42:29 [Laughter]
1:42:33 again oh don't forget to turn off your
1:42:36 your mic on the computer it's affecting
1:42:38 the volume okay great thank you
1:42:42 amazing I know isn't this wild it's
1:42:45 absolutely
1:42:47 wild so what's it what's it made here
1:42:50 for the back end look at this um initial
1:42:54 setup make directory Sky pallet
1:42:57 backend okay folder structure there's
1:43:00 our folder
1:43:01 structure there's our database
1:43:03 connection database.
1:43:09 JS here's where you stick your secret
1:43:11 key
1:43:14 right data
1:43:17 models order model API routes and
1:43:21 controllers
1:43:35 so we started with
1:43:40 this we started with
1:43:43 this and my prompt was what is
1:43:49 this and now I've done this a while but
1:43:56 if you're not aware how
1:43:58 deep these capabilities run like the
1:44:02 like chat GPT right now feels like a
1:44:04 very still
1:44:06 Lake in
1:44:08 Scotland right that is very very
1:44:12 deep and if if you just look at it
1:44:15 you're like oh yeah it's a cute little
1:44:17 no but but it runs deep and
1:44:20 [Music]
1:44:21 and if you start to get your head
1:44:24 around what that actually
1:44:27 means you can start to do things like go
1:44:30 from a whiteboard drawing to a marketing
1:44:35 plan to to a creative brief to
1:44:40 naming
1:44:42 options to
1:44:46 taglines to logo
1:44:51 studies you know to to
1:44:54 marketing
1:44:55 renderings to
1:44:59 interface to
1:45:03 code that took me 15 minutes
1:45:08 yapping
1:45:10 imagine if you were actually
1:45:15 trying from a [ __ ] whiteboard
1:45:19 drawing so anyway that's [ __ ]
1:45:25 Bonkers um let's flip over to gp4 with
1:45:29 canvas because this is even more insane
1:45:33 that that they've added this so this is
1:45:35 if you've used Claude artifacts this is
1:45:38 very similar to that um I'm gonna try
1:45:41 something now I heard that um some
1:45:43 people got updates to Canvas where it
1:45:46 can actually run code now like like um
1:45:50 Claude can so I'm going to say um make
1:45:53 me a
1:45:55 clone of a Frogger game I
1:46:00 can
1:46:04 play in the
1:46:07 browser make it
1:46:11 awesome
1:46:17 looking now notice what chat jpd did
1:46:20 here it split this so I've now got got
1:46:24 the chat to the left and I've got this
1:46:27 editable document to the
1:46:30 right and let's see what's that that's
1:46:34 copy code review Porta language fix bugs
1:46:38 at so it doesn't have what what Claude
1:46:42 has do I have Claude up here yeah what
1:46:44 Claude
1:46:46 has if you haven't seen Claude do this
1:46:50 make me a playable
1:46:57 Frogger
1:46:59 game
1:47:02 that mimics the original
1:47:07 arcade game make it look awesome
1:47:21 [Music]
1:47:25 so this does a similar thing so we've
1:47:27 got our little you know chat over here
1:47:29 to the left it's doing the coding to the
1:47:33 right and then it throws up a little I
1:47:35 guess it's a react or something um front
1:47:39 end it'll take this code
1:47:41 and make this thing
1:47:51 playable and if you here's the other
1:47:54 thing if you have never learned
1:47:57 coding coding is is not that tough you
1:48:01 can actually sort of go through like you
1:48:04 know move equals Arrow up arrow down
1:48:07 arrow left Arrow right and it's got some
1:48:09 math [ __ ] in here like you might not
1:48:13 know how to do that but you can kind of
1:48:15 figure that out so here's our
1:48:17 game and it looks
1:48:21 like I'm playing Frogger
1:48:24 now it's got some drawing issues and it
1:48:27 looks like I don't have a log I can jump
1:48:29 onto here which is
1:48:32 bad oh no game over okay so anyway so
1:48:36 it's got some issues but anyway that's
1:48:38 that's what what this does so I think
1:48:40 chat gpt's going to get that it just
1:48:42 doesn't have it right now but this
1:48:46 um this window is editable so if I know
1:48:49 what I'm doing here I say okay the Frog
1:48:51 they're going to make the Frog 50 by 50
1:48:54 let's make the Frog uh shorter in height
1:48:59 than we'll make it 40 right so I can go
1:49:02 in and edit this I can also click over
1:49:05 here and say um let's see fix bugs add
1:49:09 log let's say add
1:49:12 logs and so what this is going to do now
1:49:15 is it's going to go through here it's
1:49:16 gonna it's going to
1:49:18 comment or I guess it's going to add
1:49:20 logging yes it's going to add logging
1:49:22 see see what it did right here console
1:49:25 log it's still
1:49:28 writing so that's going to help us debug
1:49:31 this
1:49:32 thing um add comments so we'll do that
1:49:36 now it's going to go
1:49:38 through and add in comments right Define
1:49:42 the Frog object with its properties
1:49:44 initial horizontal position array of
1:49:47 hold
1:49:49 objects so now I can go through and
1:49:51 actually understand you're actually
1:49:53 watching this thing you
1:49:55 know write its code move the Frog down
1:49:58 move the Frog up so even if you just
1:50:01 wanted to learn
1:50:03 programming you could do this and I
1:50:05 could I could come over here and I can
1:50:06 say um
1:50:09 what
1:50:11 language is this in question
1:50:15 mark it's written in HTML CSS and
1:50:19 JavaScript
1:50:20 um if I wanted
1:50:23 this to
1:50:24 be a native mobile
1:50:29 app what would you
1:50:37 recommend spelled
1:50:42 wrong react native with
1:50:47 Expo flutter
1:50:57 let's see I can come over here and I can
1:51:01 say port to a language let's Port this
1:51:06 to JavaScript or no let's Port it to
1:51:10 python
1:51:12 boom so now Frogger python
1:51:19 game so it's it's maintained some
1:51:22 comment
1:51:24 it's now written this thing in
1:51:26 Python now what's cool is notice that
1:51:29 we've got this little button here
1:51:31 Frogger python game if I scroll
1:51:39 up awesome Frogger clone I can click on
1:51:42 this here's
1:51:44 my CSS version my HTML
1:51:48 version here's my Frogger version
1:51:56 so again if you're a coder and you're
1:51:58 not experimenting with this get the [ __ ]
1:52:01 on
1:52:02 it if you're not a
1:52:05 coder you can make games you you could
1:52:08 experiment with this you could
1:52:10 say how do
1:52:13 I get this
1:52:19 code out of the window
1:52:23 and playable in a
1:52:28 browser I don't know how to
1:52:34 program I am
1:52:37 a dumb dumb with no
1:52:44 qualifications I
1:52:46 went to a state
1:52:50 school and got a a
1:52:53 liberal
1:52:55 arts
1:52:59 degree I used to be an
1:53:03 actor do the
1:53:06 math but not advanced
1:53:12 math I am a dumb dum
1:53:17 [Laughter]
1:53:24 okay Kyle no worries let's make this
1:53:27 accessible and get you sorted out you're
1:53:29 not dumb you've got great instincts
1:53:32 we're going to make this as easy as
1:53:33 possible for you since this python
1:53:35 version is written in pi game it's not
1:53:37 directly be playable in the browser but
1:53:40 there are ways that we can do it convert
1:53:42 to web using HTML 5 and
1:53:46 JavaScript all right use tools to
1:53:48 convert bython to web my recommendation
1:53:51 since you want to get this into a web
1:53:53 browser and easily playable let's focus
1:53:55 on
1:53:56 JavaScript I can make you a step-by-step
1:53:59 guide right so so
1:54:03 again we now live in a
1:54:06 world where no
1:54:08 matter how ridiculously inept you feel
1:54:14 at whatever the technology is or
1:54:16 whatever the subject matter is you have
1:54:18 a deep deep expert as your assistant
1:54:23 I mean it's staggering can you
1:54:27 rewrite the
1:54:30 python based on your
1:54:35 recomendations and it should editing the
1:54:38 Frogger
1:54:40 game rewriting python code to make it
1:54:42 web friendly using JavaScript with logic
1:54:45 like Pi pyodide or
1:54:48 brython all right great
1:54:53 I don't know what the [ __ ] that
1:54:55 means but it'll tell
1:54:58 us all right
1:55:01 so so that's gp4 with canvas but wait
1:55:04 there's
1:55:07 more
1:55:08 um I want a
1:55:13 unique storytelling format or no uh
1:55:21 framework that will well let me write a
1:55:25 short
1:55:28 story that doesn't
1:55:35 suck this is worth the cost of admission
1:55:38 thank you very
1:55:39 much hopefully it can make a maze runner
1:55:42 I'm sure it can do a Maze Runner quite
1:55:48 well uh uh uh
1:55:53 it's going so fast is that python code I
1:55:56 honestly don't
1:55:57 know check your regulars for an image
1:56:07 okay grab and fly
1:56:13 nice by the way if you're new here where
1:56:16 everyone's going this Irregulars channel
1:56:19 is UM in an in a
1:56:23 you a it's a it's a special little Club
1:56:26 in the AI salon and so the AI Salon if
1:56:29 you go to the
1:56:31 salon. and scroll down and you click on
1:56:33 join our community that will take you
1:56:35 over to the AI salon and there's a
1:56:39 little welcome video that welcomes you
1:56:41 in and you know play first mindfully
1:56:44 create generously lead you won't get
1:56:46 this credit card thing I just got that
1:56:47 because I changed my credit card number
1:56:50 um you can introduce yourself learn
1:56:52 about our values Watch past meetings
1:56:54 there's all sorts of cool things in here
1:56:55 there's a showand tell area where you
1:56:57 can show off but we also have for this
1:57:00 group the people that show up to this
1:57:02 channel regularly we have our own little
1:57:04 special Club area so you can come in
1:57:06 there and hang out all
1:57:09 right um the fractured lens framework
1:57:12 okay so this came up with a storytelling
1:57:13 framework for us so I'm gonna go so I'm
1:57:16 in GPT 40 with canvas I'm going to
1:57:19 say great
1:57:25 use this framework to write
1:57:29 me a mystery
1:57:33 story where someone has hidden a graham
1:57:37 cracker
1:57:38 and four strangers on a train have been
1:57:42 challenged with finding
1:57:50 it period make sure that the story has a
1:58:00 Twist and so again what it's done here
1:58:04 is it
1:58:05 recognized it recognized the difference
1:58:09 between me asking about a story
1:58:11 framework and me saying I want to start
1:58:13 writing a story I've been thinking about
1:58:16 telling you guys for a while now I think
1:58:18 I'm going regular
1:58:25 don't do it shanea don't do
1:58:28 it I think with your name you can't be
1:58:32 regular I think you would have to change
1:58:34 your
1:58:35 name the Muffin Man story there you
1:58:41 go oh man okay
1:58:46 um this is not good for you
1:58:49 all you can't quite see everything there
1:58:52 you go
1:58:53 okay
1:58:57 um so a couple of things in here so this
1:58:59 is now a fully editable story right the
1:59:02 mystery of the Hidden graham cracker so
1:59:04 I can highlight that and now I can in
1:59:07 line ask chat GPT I'm gonna say this
1:59:11 title
1:59:15 sucks please make it more better
1:59:23 and I hope you're I hope you're mocking
1:59:25 me openly for my typing and
1:59:29 spelling the crumbs of curiosity that's
1:59:32 not bad so we'll
1:59:35 go not bad more
1:59:39 literary or
1:59:41 [Music]
1:59:45 Sherlock homsy
1:59:48 [Laughter]
1:59:53 the case of the Trum crumbling
1:59:55 confection mirrors Edge start a train to
1:59:58 nowhere in particular nowhere in
2:00:00 particular nowhere in particular chugged
2:00:02 along each car swaying with an almost
2:00:05 hypnotic rhythm great so what else can
2:00:09 you do here you'll notice there's just
2:00:10 like the the coding thing had special
2:00:13 editing things um there's a bunch of
2:00:17 things in here that are absolutely wild
2:00:19 one is suggest edits so it will now go
2:00:21 through
2:00:24 and highlight things that it thinks
2:00:26 could be
2:00:27 better and then you can go look at it
2:00:30 what's it say here consider adding a
2:00:32 more
2:00:33 specific sensory details to enhance the
2:00:36 setting such as sounds or smells inside
2:00:38 the train great I like that suggestion
2:00:41 do
2:00:42 it there the train from nowhere in
2:00:45 particular chugs tugged along each car
2:00:47 swaying with an almost hypnotic rhythm
2:00:49 the faint clatter of The Wheels On The
2:00:51 Tracks mingled with the low murmur of
2:00:52 distant conversations and the faint
2:00:55 scent of aged upholstery filled filled
2:00:58 the air tinged with a hint of coal smoke
2:01:00 from the engine the conductor voice had
2:01:02 crackled over the speakers just minutes
2:01:04 earlier issuing A peculiar challenge a
2:01:07 graham cracker a gr cracker has been
2:01:10 hidden somewhere on board the first
2:01:12 person to find it will win a story
2:01:17 Untold so you can do that there's the
2:01:20 suggestions now you can also adjust the
2:01:23 length so I can go keep the current
2:01:25 length make this longer or make this
2:01:28 shorter I can also change the reading
2:01:31 level so right now it's whatever it's at
2:01:34 I can say let's make this graduate
2:01:36 school level
2:01:38 writing and now it's going to completely
2:01:40 rewrite the
2:01:42 story the train with its
2:01:44 destination deliberately undefined
2:01:47 Advanced along the tracks each carry
2:01:49 Carriage undulating with a hypnotic C
2:01:53 the rhythmic clatter of The Wheels on
2:01:55 the steel merged with the distant
2:01:56 murmurs of oh my god so let's let's
2:02:01 let's let the second paragraph be
2:02:03 graduate school level but let's change
2:02:05 this
2:02:06 paragraph to
2:02:08 be
2:02:14 kindergarten the train was going
2:02:16 somewhere but no one no one knew where
2:02:18 it moved along the tracks each car
2:02:20 rocking back and forth I can also go
2:02:24 back to earlier
2:02:26 versions like I like this one with the
2:02:28 low murmur I can restore that
2:02:32 version I can also go down here and I
2:02:34 can
2:02:36 say you know
2:02:38 what make me a different story that's a
2:02:41 murder mystery
2:02:52 where the murderer exited the train
2:02:56 before the story
2:02:57 starts period so everyone blames one
2:03:02 another but no one was actually guilty
2:03:05 period it might be fun if there were
2:03:08 additional murders because of the
2:03:13 suspicions and sorry Tick Tock if I get
2:03:16 kicked off Tik Tock for saying that word
2:03:18 but hey Tik Tock what's happening
2:03:27 I'm not getting on a train unless I have
2:03:30 a
2:03:31 destination who needs spelling when you
2:03:33 have ai totally agree with
2:03:39 that all
2:03:42 [Music]
2:03:45 right now did this give us a new oh this
2:03:48 just rewrote it that's interesting
2:04:07 add this story to a new canvas and give
2:04:09 me the old story in a separate
2:04:13 canvas see if it can do
2:04:18 that there's the accusatory train
2:04:20 mystery
2:04:27 [Music]
2:04:32 all right let's see if we can go back
2:04:34 and get get back our other one graham
2:04:37 cracker
2:04:44 Mystery Case of the crumbling con
2:04:46 confection so okay so I returned that
2:04:49 one to what it was and now I should have
2:04:53 this one's a new
2:05:00 one oh no it didn't do it let's
2:05:17 see what's going on
2:05:23 change this one to the murder mystery
2:05:35 [Music]
2:05:41 one okay so there's the murder mystery
2:05:43 one and then we should have the graham
2:05:46 cracker one yeah we do okay so so again
2:05:50 like the the value the value of this
2:05:53 just again it runs deep like you start
2:05:57 with a storytelling framework or you
2:05:59 start with a whatever a business you
2:06:01 know business case framework and then
2:06:05 you create one business case you create
2:06:07 another one you create another one and
2:06:09 there you can return to all of them and
2:06:11 edit them and then export them it's it's
2:06:13 just
2:06:14 Bonkers um you can add final polish
2:06:16 let's do that let's just see what that
2:06:18 does
2:06:29 I guess it's just editing and doing [ __ ]
2:06:31 like
2:06:39 that
2:06:46 [Music]
2:06:50 apply all right so that's that all right
2:06:54 now we're going to show you one other
2:06:55 thing oh the other thing you can do with
2:06:57 these is you can close a canvas so so
2:07:01 they they just sit in line where you can
2:07:04 actually read them in line in the chat
2:07:07 but if you expand them out bang it
2:07:09 splits it into these two windows this is
2:07:11 just this is well done interface in my
2:07:13 opinion it's just really
2:07:15 slick okay um one final thing now um
2:07:22 and I know this is this is a long night
2:07:24 but that's all right um our video system
2:07:27 is down so now we're monitoring audio
2:07:30 might lose my mind 10 beds
2:07:37 snoring all right is anyone still here
2:07:40 yeah we still got some folks
2:07:43 here got 19 on
2:07:45 YouTube got 35 on the tick
2:07:49 tock all right I'm gonna show you one
2:07:51 other thing
2:07:55 and this is a total experiment so this
2:07:57 was the idea I had so we're going to try
2:07:59 we're going to go new chat we're going
2:08:01 to go 01
2:08:04 preview
2:08:07 okay think of GPT 40 and 40 with canvas
2:08:12 as your daily drivers right the these
2:08:15 are this is the car that you're going to
2:08:17 take to the grocery store 01 preview is
2:08:19 like your track race car it's just a
2:08:22 different thing you could technically
2:08:24 drive it to the grocery
2:08:26 store but you would annoy people and it
2:08:29 would be hard to drive and it wouldn't
2:08:31 carry many groceries because it's not
2:08:33 designed for that if you take it to the
2:08:35 track you can have some fun with it so
2:08:38 I'm G to flip over to 01 preview what 01
2:08:40 preview does is it it uses a modified
2:08:43 version of what's called Chain of
2:08:44 Thought reasoning or Chain of Thought
2:08:47 prompting where when you give it a
2:08:49 challenge it it says it's thinking Pate
2:08:52 has rightfully said it's not thinking
2:08:54 what it's doing is it's prompting itself
2:08:56 so it will give itself a prompt based on
2:08:59 what you ask it to do and then it will
2:09:01 look at the answer to what it did and it
2:09:03 will give itself another prompt and then
2:09:05 it will give itself another prompt and
2:09:07 if you give it a good meaty challenge it
2:09:10 will think for a good long time it'll go
2:09:12 through many many many many
2:09:15 steps and so this is one of those things
2:09:19 where you could use it a lot like it's
2:09:22 one thing I found is it's really good at
2:09:24 writing it's good at writing screenplays
2:09:26 especially if you give it an example um
2:09:28 because it can reason its way through
2:09:31 plot problems and structure problems and
2:09:33 things like
2:09:35 that but I thought of a I thought of a a
2:09:38 children's book that could be really fun
2:09:43 so so we're gonna give it a quite
2:09:45 complicated prompt
2:09:48 um and we'll just see how it does with
2:09:50 it all right
2:09:57 so actually you know what I could
2:10:02 do one of the things you can do is you
2:10:05 can switch back and forth so long as you
2:10:06 don't search the web you can switch back
2:10:08 and forth between gp4 and 01 preview so
2:10:12 I'm gonna I'm going to do the same thing
2:10:13 I did
2:10:16 before
2:10:18 recommend five different storytelling
2:10:24 Frameworks that can be used to help me
2:10:28 write an award-winning children's
2:10:37 book hero's
2:10:41 journey
2:10:43 circular problem solution Arc rhyming
2:10:48 repetition what if
2:10:53 I like hero's journey and rhyming
2:10:55 repetition period can you combine those
2:10:58 into a unique storytelling framework for
2:11:05 me and again the reason I'm doing that
2:11:08 is just just to show that you
2:11:13 can these things know so much like take
2:11:16 advantage of it like instead of just
2:11:18 asking it to write a story like learn
2:11:20 some [ __ ]
2:11:22 invent some [ __ ] you know there there's
2:11:25 one of the one of the um the things that
2:11:28 I still hear I got in an argument with
2:11:30 this I was presenting a couple of weeks
2:11:32 ago and a guy came on and he goes you
2:11:34 know this is just predict you know it's
2:11:36 just predicting the next token this
2:11:38 can't do original thought you know it
2:11:42 can't it can't create new and novel
2:11:44 ideas well I'm sorry but combining two
2:11:47 different storytelling Frameworks into a
2:11:49 third that third thing is a new and
2:11:52 novel idea now is it might it be any
2:11:55 good might it be something someone's
2:11:57 done before sure but to to say that
2:12:00 these things
2:12:01 can't
2:12:03 function like humans can function with
2:12:05 creativity is actually just
2:12:07 [ __ ] um okay so now we've got a
2:12:10 storytelling framework great now I'm
2:12:13 going to flip over to 01 preview so now
2:12:16 we're going to do now now we're going to
2:12:18 do some
2:12:20 [ __ ] okay great
2:12:24 I want you to use this
2:12:28 framework to write
2:12:31 me
2:12:33 a
2:12:36 20page children's
2:12:39 book
2:12:49 about a young boy
2:12:56 that lives on a Peanut
2:13:04 Farm and loves to have
2:13:07 Adventures around the
2:13:11 farm and with his
2:13:14 sister and with his mother and father
2:13:18 period he also loves to play with
2:13:21 animals
2:13:38 what I want to be unique about this book
2:13:40 is that I
2:13:43 want the text on every
2:13:47 page to be a cipher
2:13:51 that uses a different methodology on
2:13:55 every page
2:14:00 period each page should read as a
2:14:06 complete section of the
2:14:10 story but also be able to be deciphered
2:14:14 to reveal a clue period
2:14:22 if the reader
2:14:26 solves every clue on every
2:14:30 page then the message that's revealed
2:14:34 should
2:14:44 be peanut butter
2:14:49 cookies are the tastiest cook
2:14:53 cookie in all of the
2:14:55 world
2:15:02 period they're even better when you dip
2:15:05 them in cold
2:15:13 milk okay now I don't know if you could
2:15:15 hear me say all that
2:15:18 but this is going to use this
2:15:20 storytelling framework that we just
2:15:22 invented it's going to write this story
2:15:24 about a kid that Liv lives on a Peanut
2:15:29 Farm every page is going to be a unique
2:15:32 Cipher that uses a unique
2:15:34 methodology and then if you solve all of
2:15:38 those ciphers
2:15:40 uh you should get that oh then I should
2:15:42 I'm gonna add
2:15:47 this I want you to write
2:15:51 both the
2:15:53 story but also the parents guide to
2:15:59 understand the cipher on each page and
2:16:02 what the solution should be and how you
2:16:05 solve the cipher
2:16:12 period oops what did I
2:16:18 do okay
2:16:23 now oh wait what the hell did I do um am
2:16:27 I still
2:16:35 live hang on a
2:16:43 sec okay I'm back can you all still see
2:16:46 me on broke it I did break it I totally
2:16:49 broke it am I still live on on
2:16:58 YouTube can you now see this
2:17:07 again yes sir okay
2:17:09 great okay so let me hide
2:17:14 this live still live alive and listening
2:17:19 okay if you ever hit the fingerprint
2:17:22 button on your keyboard it completely
2:17:25 shuts everything
2:17:26 down yes okay all
2:17:29 right so we're going to write a story
2:17:33 each page is going to be a
2:17:35 cipher the cipher is going to reveal
2:17:37 peanut butter cookies are the tastiest
2:17:39 cookies in all the world they're even
2:17:41 better when you dip them in cold
2:17:49 milk and then I'm going to add prove me
2:17:53 wrong prove me
2:17:55 wrong we'll get some meme culture in
2:17:58 here
2:18:01 too I want you to write both the
2:18:04 story but also the parents guide to
2:18:07 understand the cipher on each page is
2:18:08 that that's probably not the right way
2:18:10 to spell Cipher is it um oh one other
2:18:14 thing let's see I want you to write both
2:18:16 the
2:18:17 story descriptions
2:18:21 of the images on the
2:18:29 page and and the the other thing I
2:18:32 thought when I was thinking about this
2:18:33 is what you should really do is the
2:18:35 images should also also be ciphers but
2:18:38 that's too complicated let's see if it
2:18:39 can do this I don't even know if it can
2:18:41 do this but this is the kind of
2:18:44 thing
2:18:47 that to do this manually like you would
2:18:50 have to know
2:18:51 a a [ __ ] ton about cryptography and you
2:18:55 know puzzles and [ __ ] like that so here
2:18:58 we go let's just see what it
2:19:00 does
2:19:01 thinking crafting The Narrative so so
2:19:04 here's your Chain of Thought you can you
2:19:06 know this it just says thinking and it
2:19:08 shows you what it's doing but you can
2:19:09 drop this down and actually watch what
2:19:11 it's doing so crafting the narrative
2:19:14 writing the story crafting The Narrative
2:19:18 igniting young imaginations Imagining
2:19:20 the adventure creating The
2:19:22 Narrative considering Cipher options
2:19:25 piecing together the
2:19:27 clues I'm thinking about how the first
2:19:30 letters in each line spell out
2:19:33 pen now I'm set to arrange those to
2:19:36 decode the message piecing together the
2:19:39 story collecting Clues assigning
2:19:42 identifiers oh look piecing together the
2:19:44 puzzle I'm gathering info into a
2:19:46 sequence the positions are 48 12 1620
2:19:50 that's pretty cool
2:19:52 crafting a mixed approach I'm exploring
2:19:54 various methods solving the mystery
2:19:57 piecing together the
2:19:59 clues page
2:20:01 one playing under Skies so blue every
2:20:04 day there's so much to do animals and
2:20:06 Fields to tend new adventures have no
2:20:10 end upy jumps with Che cheerful
2:20:14 grin the to the farm let the fun begin
2:20:17 image description Liam is joyfully jump
2:20:21 jumping near his Farmhouse surrounded by
2:20:23 vast peanut
2:20:26 Fields so here's all of our
2:20:31 book so we wanted 20 Pages it'll be
2:20:34 interesting to see if it actually gives
2:20:35 us 20
2:20:37 Pages how long did it think it thought
2:20:43 for 57 seconds which is not that long
2:20:57 1920 all right so peaceful dreams until
2:21:02 the dawn rest until the night is gone on
2:21:04 the tomorrow fun awaits Venture forth to
2:21:07 greatest States every story has an end
2:21:10 memories are ours to send parents guide
2:21:14 page
2:21:15 one Cipher method acrostic Cipher first
2:21:19 letters
2:21:21 oh that's so cool the first letter of
2:21:24 each line spells the
2:21:26 word playing every animal's new up to
2:21:29 peanut clue word
2:21:31 peanut this is so cool page two Cipher
2:21:35 method acrostic first letters so it it's
2:21:38 not using different ones for each thing
2:21:41 let's see these are all
2:21:43 acrostic hidden words Cipher page
2:21:47 six hidden words within the text form
2:21:51 the clue Clues are hidden all around
2:21:54 words containing the word wait words
2:21:56 containing the words cookies page six
2:21:59 let's go look at page
2:22:05 six Clues are hidden all around on on
2:22:08 the farm and in the
2:22:11 ground Outside Inside up and down
2:22:14 keeping searching from front in the end
2:22:16 you'll find the prize each clue solved
2:22:18 before your eyes
2:22:23 oh
2:22:25 cookies
2:22:27 cookie so this is
2:22:29 still first
2:22:34 letter all right
2:22:39 so capital letter
2:22:42 Cipher first letters in each
2:22:45 line word length
2:22:51 clue word
2:22:54 get peanut butter cookie cookies it
2:22:57 missed are tastiest cookies in all of
2:23:01 the world they're even better when you
2:23:03 dip them in cold milk prove me
2:23:05 wrong well it did it holy crap all
2:23:09 [Laughter]
2:23:16 right let's see still live yes we should
2:23:21 have all said no just to see what Kyle
2:23:23 does when he thinks no one is
2:23:27 watching y'all are [ __ ] weird all
2:23:30 right there's your Deep dive into chat
2:23:33 GPT I'm sure there's some [ __ ] I missed
2:23:35 oh I didn't do I didn't do code
2:23:38 analysis it's
2:23:41 like please tell me this is being
2:23:43 recorded yes this is being recorded so
2:23:45 this will be
2:23:47 um this will be on CH CH
2:23:53 what hold please wait I've got a banner
2:23:57 for that champ shut up shut
2:24:00 [Laughter]
2:24:04 up what is your problem
2:24:08 dog everyone
2:24:27 all
2:24:28 right there you have it let me come back
2:24:30 over here
2:24:32 bang go over here go like this boom
2:24:37 boom do
2:24:40 this all right
2:24:44 everybody good Lord that was a lot
2:24:55 [Laughter]
2:25:03 thoughts
2:25:07 good we are IR regular not
2:25:11 weird that was fun and super cool thanks
2:25:13 thanks Kyle thanks mod and Brandon thank
2:25:16 you Joy thank you kakee Shane saww all
2:25:20 right everyone how many people we got 16
2:25:23 left on the YouTube yeah this as soon as
2:25:26 I stop this stream this should be
2:25:28 available as a recording all right so
2:25:31 feel free to share it and you know get
2:25:34 people there again my plea I'm trying to
2:25:37 get a thousand subscribers on YouTube so
2:25:40 please share the YouTube and if you
2:25:42 haven't subscribed please subscribe to
2:25:44 it thank you Kyle it's good to go back
2:25:46 through yeah I know
2:25:49 right it's good to go back through just
2:25:52 it again it's like these tools are so
2:25:56 capable and it's like even something
2:25:59 like suo that you write songs with you
2:26:02 can just write a prompt and it'll
2:26:03 generate a song but there's so many
2:26:05 features within it now like all of these
2:26:07 things are just expanding their feature
2:26:09 sets so even if even if you think you
2:26:12 know something go back and you know dig
2:26:16 because these these tools are getting
2:26:18 very sophisticated all right
2:26:21 awesome I'm going to take off here I'm
2:26:23 going to take off here nice shirt Mr
2:26:25 Shannon I know right Blandon
2:26:28 baby sometimes I forget things it's good
2:26:31 to see again thank you so much Cam kaken
2:26:34 Joey thank you so much spin
2:26:38 B3 Be not Afraid loved it
2:26:43 awesome all right everybody fantastic
2:26:47 I'll see you tomorrow wait wait what's
2:26:49 today today's Wednesday tomorrow's
2:26:51 Thursday I got nothing going on should
2:26:52 be normal time I think we'll keep doing
2:26:55 this I I'm starting to figure out the
2:26:56 duel thing a little bit so we'll keep
2:26:59 doing this for a while um hopefully we
2:27:02 can get some more subscribers on YouTube
2:27:03 Share the [ __ ] out of this peace bye