Jan 24, 2024 - Create Your Own GPTs and Transform Content Creation

Description

Chapters

Transcript

0:00 some good singing oh yeah you're ready I
0:02 can tell oh you're ready you ready we do
0:07 a little Jesse
0:08 Winchester
0:12 [Music]
0:17 [Laughter]
0:19 [Music]
0:22 huh the boys was singing
0:26 shingling the summer that we met
0:31 you were 10 and
0:34 17 how could I
0:40 forget every Star from near and far was
0:44 watching from
0:47 above watching two
0:50 teenagers fall in
0:53 [Music]
0:56 love the way we danced was not a dance
1:01 but more a long
1:04 Embrace held on to each other and Flo
1:09 there in
1:12 space and I was shy to kiss you while
1:17 the whole wide world could
1:19 see so shinging said everything for
1:25 [Music]
1:27 me and all the poor old old folks they
1:31 thought we' lost our
1:34 minds they could not make heads or tals
1:38 of the Young Folks funny
1:41 [Music]
1:43 rhymes but you and I knew all the words
1:48 and we always sang
1:50 along to Old
1:54 Cham ding
1:57 sh
1:59 dang
2:01 [Music]
2:09 [Laughter]
2:12 so after years and after tears and after
2:16 Summer's
2:18 passed the old folks tried to warn us
2:22 wait the old folks Tred to want us our
2:26 love would never
2:28 last but all we yet was soaking wet from
2:33 walking in the
2:35 rain and singing shammer
2:39 shingling
2:42 again and all the poor old old
2:48 folks they smile and walk
2:51 [Music]
2:53 away but I bet they did some shamama
2:57 Lama a ding dong in the
3:02 [Music]
3:07 day I know I should learn this song
3:09 better right that's a beautiful Jesse
3:11 Winchester
3:12 [Music]
3:19 song the boys were singing
3:23 Shing the summer that we
3:29 met
3:32 I don't even deserve to play
3:34 [Music]
3:43 [Music]
3:44 [Laughter]
3:45 [Music]
3:59 today
4:00 [Music]
4:05 you and I here all
4:09 alone Sunday morning here at
4:13 home sky blue and a cough is strong it's
4:18 true thank you Robert Rossy good Lord
4:21 very generous thank you so
4:23 much good day good day Cindy [ __ ]
4:27 sitting in SFO she boards for as in a
4:30 couple hours and this is how she spends
4:32 her time good Lord a singing dog an
4:36 attitude
4:37 [Music]
4:38 [Laughter]
4:41 [Music]
4:57 guitar you and I all
5:01 long Sunday morning hair at
5:06 home sky blue and the CU is strong it's
5:13 true then I open my eyes to dream realiz
5:18 in front of
5:21 me and I hav't got a clue what in the
5:25 world is happen to
5:28 me think I think I'm
5:32 happy like first day summer vacation
5:36 happy got to get a little rest and
5:39 relaxation
5:40 happy like to quiet on Sunday morning
5:43 singing
5:45 [Music]
5:58 true
6:00 [Music]
6:12 who you good that
6:16 [Music]
6:24 enough I think he's
6:26 done I think he's digging this singing I
6:29 I think he's digging
6:31 it he comes he comes up here now he lays
6:34 right by
6:35 me he knows what's coming he knows he's
6:38 singing for his
6:40 supper this is this is all for Pate this
6:43 is for chicken
6:44 sticks this is all for p subscription
6:47 pays for your chicken
6:49 sticks so you got to do it you can't not
6:53 sing welcome
6:55 everybody happy
6:58 Wednesday
7:00 had kind of a nothing day
7:02 today just messed
7:04 around messed around all day didn't see
7:08 anything too remarkable out there like I
7:11 said I think we are be being lulled into
7:13 a false sense of
7:15 security I think some shit's about to
7:18 hit the fan with this new with some new
7:20 AI stuff but uh but we'll have to see
7:23 we'll have to see maybe nothing's going
7:24 to happen in 2024 maybe 2023 is not the
7:28 last normal year I think maybe 2023
7:31 maybe was just a
7:32 year 2024 is going to be non- remarkable
7:36 very possible and then we'll just sing
7:42 [Laughter]
7:47 songs all right
7:50 people let's get this show started for
7:54 the
7:55 chicken champy singing for the
7:58 chicken only music itions don't follow
8:00 the
8:03 singer what are you smoking did you wait
8:06 did you not see mid Journey replacement
8:09 for
8:10 free no I
8:12 didn't wait what's that what's mid
8:15 Journey replacement for free what the
8:18 what the
8:24 [Music]
8:28 heck
8:31 oh that was wrong
8:38 right mid
8:42 journey
8:49 journey mid Journey just pushed out some
8:52 updates to the web
8:58 Alpha can you you make can can
9:02 non wait pix art-
9:08 Alpha but is it is it mid Journey pix
9:13 art-
9:25 Alpha Transformer based text image
9:28 diffusion system
9:33 hm no I don't know
9:37 this I do not know
9:40 this it looks like it's a paper with a
9:43 little but it looks like you can play
9:44 with it
9:52 yes is this the one you're talking about
9:55 Sherry D oh you're not looking there but
9:58 so it's a it's it's a hugging face
10:00 spaces right
10:02 now yeah I mean it like it quality-wise
10:05 it might be there but what I'm looking
10:07 forward to is mid
10:09 Journey like mid journey I actually
10:11 meant to show this the other night mid
10:14 Journey um they're improving their their
10:18 website so if I sign
10:22 in well let's get started we'll get
10:24 started here we'll come back to
10:28 this oh I got to do multiactor
10:30 authentication what's the latest on
10:32 Gemini I don't think
10:34 anything but maybe I missed some [ __ ]
10:37 today
10:40 Gemini uh
10:47 Ultra news about
10:51 Gemini nothing
10:53 honey nothing
10:57 honey nope
11:01 nothing we ain't got we ain't got no
11:03 news all right
11:05 everybody so welcome welcome welcome
11:07 this is the uh this is the AI learning
11:11 lab I'm stumbling around in the dark
11:13 here bunch of Google texted video papers
11:16 came out yeah I saw a couple of those
11:18 those were okay you need to be using
11:20 Suzanne's GPT for AI news yeah I should
11:22 do that although the last time I tried
11:23 to use
11:24 it she did not work what's it called
11:27 again Source camp
11:30 [Music]
11:35 let's see
11:39 suzan um Welker is it Welker is that
11:43 right no
11:45 w e
11:48 MK
11:49 [Applause]
11:53 no AI daily
11:56 rundown AI daily
12:01 run
12:06 down s Welker Jurgens okay today's AI
12:12 news today's AI news let's see if
12:15 anything comes up yeah Google put out a
12:18 couple of papers today on some video
12:20 stuff that looked pretty cool the the
12:22 problem with the with the Google papers
12:25 coming out as they do the it's it's a
12:28 preview of what's to come but it's not
12:29 really what's there like they've got
12:31 some really exciting papers they put out
12:34 that you can get to if you build using
12:36 their API
12:38 stuff
12:40 Suzanne I don't know if you're on here
12:42 oh now it's working error creating
12:47 image this this GPT breaks for me for
12:50 some reason it's
12:53 weird Google back's National AI research
12:56 pilot okay cool Google Google's gradient
13:00 invests in send AI for data
13:04 extraction for
13:07 complex documents all right so they
13:10 invested in a
13:12 startup service now rise in cyber
13:14 attacks yeah I don't think there's
13:16 anything super massive that I saw anyway
13:21 so welcome uh my name is Kyle Shannon
13:23 this is the AI learning lab so what
13:25 we'll do here is we'll get started we're
13:26 not really started yet we're just kind
13:28 of just kind of of babbling just kind of
13:30 babbling back and
13:32 forth
13:35 um and then we'll show some stuff we'll
13:37 talk about some
13:39 stuffs 30 minute hellos exactly Danielle
13:42 Danielle's got it figured out she
13:44 understands how this works let me get my
13:47 little mod Channel up in case there's
13:49 any panicked Kyle you got to go look at
13:54 something SpaceTime text to video
13:56 diffusion model by Google research
14:01 all
14:02 right so oh Lumi Lumiere didn't we look
14:05 at this last night Lumiere Dov video.
14:08 GitHub is this new hasn't this been out
14:11 did they update
14:22 it oh they're
14:26 using oh looks like they rolled a bunch
14:28 of things
14:30 together this looks like they rolled
14:32 rolled style drop into some video
14:36 stuff so let's this is actually probably
14:39 worth looking
14:42 at
14:46 H let's see
14:49 Loom
14:52 Year
14:55 Dot video.
14:58 GitHub
15:03 Doo that's spelled right oh Dash
15:10 video lumere DV video.
15:15 github.io let's watch this video so so
15:18 it looks like they're they're rolling
15:19 some tech together I don't know if I
15:20 don't know if uh P's in here
15:23 um but
15:26 um but let's let's just watch the video
15:28 it's text to video
15:29 so they got text to video image to
15:36 video stylized generation so this is
15:39 that this is that style drop thing
15:44 although this doesn't look nearly as
15:46 good as the when it was just images on
15:48 their
15:49 own and then there was video stylization
15:52 so you can take a source video and just
15:55 describe different styles and it'll
15:57 recreate it then there's cinemagraphs
16:01 which are just animated gifts and
16:03 there's video in painting which is
16:05 cool like change her
16:09 outfit huh let's see what they say about
16:16 [Music]
16:28 it
16:29 [Music]
16:33 is it all
16:39 [Music]
16:49 [Applause]
16:49 [Music]
16:57 square
17:01 [Music]
17:27 w
17:33 [Music]
17:35 [Applause]
17:38 [Music]
17:39 [Applause]
17:42 [Music]
17:47 [Applause]
17:48 [Music]
17:54 all right that's interesting I mean but
17:55 if you if you remember there's this
17:57 thing called style drop
18:01 style drop
18:04 Google that they're using they're using
18:07 some of the same Tech in
18:09 it it looks
18:11 like this style transfer technology but
18:15 this looks really slick where where is
18:17 it oh this was just an article oh
18:24 God are you going to give me a hyperlink
18:26 to the actual site you dumb DS
18:30 anyone no I'm screw you I can't I can't
18:34 deal with I can't deal with um what are
18:38 they called
18:40 um media sites that don't link you out
18:43 of their site they're like well we don't
18:45 want them to leaving to site or they'll
18:47 never come back well guess what if you
18:48 don't link me out of your site I'm not
18:49 going to stay there in the first place
18:51 you big fat Dum Dums
18:54 okay so this is this is a
18:57 paper
19:00 this is a paper that this came out this
19:03 was ages ago this came out like June
19:07 maybe June or
19:10 July and then they've got this you know
19:13 you put one image in and then it makes
19:15 these static images that are identical
19:17 to that input image and they have a
19:19 bunch of different styles and these
19:22 things look great and then what they
19:23 just showed in the video like it was
19:25 this same input image and it sort of L
19:28 looked like this stuff but it looked
19:30 like
19:38 [Music]
19:42 [ __ ]
19:45 [Applause]
19:45 [Music]
19:53 [Applause]
19:54 [Music]
19:57 here
20:00 [Music]
20:13 see like
20:15 the I don't know the there's something
20:17 about the video it just it looks it
20:19 looks really
20:21 janky but
20:23 the static images look [ __ ] perfect
20:27 like there's the
20:29 how do I go left
20:32 here can I go
20:34 back where's the left
20:40 Arrow there like look how good those so
20:43 there's the input like sticker and look
20:46 how good those stickers look and then if
20:49 we go to this this here
20:52 VI there's the
20:56 sticker see how Jang they look they just
20:59 don't look they don't look good so
21:02 anyway all right whatever and it only
21:04 looks like they're making Square video
21:06 the research stuff is rarely able to go
21:09 into production immediately that's why
21:10 it's research that's from Pate from
21:13 Google thank you Pate yeah I know and
21:15 and that's like as I'm looking at this
21:17 this isn't ready to go anyway like it's
21:19 it's Square only video but but it what
21:22 what's really cool here is you can see
21:24 where it's going right you can see where
21:27 it's going to it's going to end up being
21:29 this tool that is you know you can
21:31 generate it you can start with video you
21:33 can start with nothing you can start
21:35 with an image you can start with a style
21:37 you can in paint you can out paint you
21:38 can right so so you can see where this
21:40 is headed but what we don't know is when
21:44 are mere mortals going to get their
21:45 hands on this [ __ ] right and that's
21:47 that's again that's more what this
21:48 channel is now if you're a developer you
21:52 should definitely be if it was open AI
21:56 you'd be using it well if it was open AI
21:58 have a I'd have a consumer interface I
22:01 could use it
22:02 on not a development environment like
22:06 like Pate I don't I you'll notice I
22:10 don't hang out in in open ai's
22:13 playground I I hang out in chat GPT
22:17 right I could hang out in Google's
22:18 development environment but but it's not
22:22 hey I'm not a
22:25 developer and you you did a Freudian
22:27 sleep slip if it was open AI you'd be
22:31 losing oh I'd be losing it no I wouldn't
22:34 I no I would still be bitching about
22:35 this the fact that it can't do 16x9
22:38 video it's only showing Square video and
22:40 the fact that the the degradation in
22:42 image quality when you animate these
22:44 things is so dramatic no I'd be [ __ ]
22:47 on open AI I dude I [ __ ] all over the
22:51 open AI store the open AI store is a
22:53 piece of steam and poo it's not even a
22:57 store have you looked at V v. yeah I
23:00 looked at I looked at that one that was
23:02 pretty good it's okay there's a I mean
23:05 here's the problem with all these
23:06 individual little tools is like there
23:08 are too many of them right there's just
23:10 there's thousands and thousands and
23:12 thousands look this is the the GPT store
23:15 is is [ __ ]
23:17 horrific it's it's not have they have
23:20 they added more than seven SE sections
23:23 and is this more than just a an anchor
23:25 tagged web page no it's
23:27 not and do any of these go more than 12
23:30 featured things see more 12 no no this
23:36 is a steaming pile of poo
23:39 like I think the the plugin store was
23:42 better than
23:47 this so I I uh but you know chat GPT is
23:53 still better than Bard so until until
23:58 that that changes Pate that's where
23:59 we're
24:00 starting we're starting with this is
24:03 state-of-the-art and when they put out
24:05 piles of [ __ ] I'll say it they're they
24:08 this store is really
24:12 [Music]
24:13 bad what's the name of the website that
24:16 that Google website was lumere l u m i e
24:21 r e
24:24 lumere vid or- video. github.io
24:30 and this is chat. open.com yeah there
24:33 you go thank you Brandon thank you thank
24:36 you thank
24:38 you did you see the GPT to
24:41 search the GPT store someone created it
24:45 no of course they did cuz someone had to
24:47 create something that sucked less than
24:49 the the [ __ ] site that they
24:52 built I can't believe we had to wait
24:55 three months for that Steam and P
24:59 laoo Sam Alman announced that at the
25:02 beginning of
25:04 November and then they didn't launch it
25:07 in November because they decided to fire
25:09 Sam Alman and then they waited an
25:12 additional two months to launch
25:15 something that
25:17 like that this must have been what they
25:19 coded in
25:22 November and then they just got busy
25:24 trying to you know lick their wounds
25:26 after F firing Sam Alman or
25:28 deal with AGI or whatever the [ __ ] was
25:30 going on over there but they they have
25:32 they they're like the they're clearly
25:35 not taking the the GPT store very
25:38 seriously right now and it might be
25:41 listen who knows why right it might be
25:44 that um they know what's coming and they
25:47 don't want to put time into this CU
25:48 they're doing something different I like
25:50 I don't know but but it's it's it's a
25:52 steam and pile of poo all
25:55 right let's see let's get going going I
25:58 haven't even gotten started yet we're
26:00 all we're all running down all these
26:01 rabbit holes
26:02 already welcome everybody if you're new
26:05 to AI um we will go look at [ __ ] um I'll
26:09 give my my two cents on it um I tend to
26:14 be uh an AI Optimist but I try to be
26:19 decently realistic um other than the
26:21 fact that I do refer to what it does as
26:24 magic more than Pate is comfortable with
26:26 Pate works at Google he like it's math
26:29 this is a probability engine dear
26:32 sir we shall not refer to mathematics as
26:37 magic unless it works like
26:43 [Music]
26:45 magic yeah I you know it's it's close
26:49 enough it's close enough
26:51 for for
26:54 me check out GPT finder in the GPT store
26:57 okay I will check check that out all
27:00 right
27:01 fantastic fantastic Bob Hey Hey listen
27:04 what are you doing tonight what are you
27:06 guys doing tonight are you going to go
27:08 look on the internet for some of that AI
27:10 stuff huh are you go hey hey Kyle you
27:14 going down to the mall today you going
27:17 to the orange julius hey Kyle you going
27:19 down to the mall
27:21 today I grew up in central Pennsylvania
27:24 that's what I listen to for most of
27:27 my schooling years was that
27:33 accent I had to get out I had to get out
27:36 of there you going down to the mall
27:38 today Kyle hey Kyle you going down to
27:40 the
27:41 mall got to change oral in the
27:47 car I've only ever seen in my lifetime
27:52 I've only seen one thing on TV that had
27:56 that accent and it was the Tracy Olman
28:00 show and she did a skit called the
28:02 Baltimore steps and it was about she
28:05 played a like a girl leaving home and
28:08 going to college and everyone was crying
28:10 and she was packing her stuff and trying
28:12 to leave and they wouldn't let her leave
28:15 and that was the accent they all had and
28:16 it was it was [ __ ] perfect I was like
28:19 holy [ __ ] somebody got
28:22 it you going down in a mall
28:26 today d
28:32 [Music]
28:36 you can take the magic out and it'll
28:38 still work fine take the math out and
28:40 it's useless that's a very good point
28:43 Pate very good
28:49 [Laughter]
28:55 point letter Kenny hi AI people hello
28:58 ISU what's
29:01 happening welcome welcome
29:04 welcome I used to
29:07 sound just like Blake station he's from
29:10 my hometown I fixed
29:15 it my Uncle Earl lived in
29:18 Pittsburgh Uncle Earl Earl would change
29:21 change my car but magic magic makes it
29:24 pretty well I I'll tell you what makes
29:26 the the AI stuff magic is what it was
29:28 trained on right it's the it's the p in
29:31 GPT
29:34 pre-trained what what makes chat GPT so
29:37 remarkable the math makes it possible
29:40 but it's the data that's in there that
29:43 it's what has been trained on right and
29:45 so that's why it you know it responds to
29:48 you in ways that are like seemingly
29:51 human and you know you you're kind of in
29:53 this weird have we passed the Turing
29:55 test I'm pretty sure we're passed the
29:57 Turing test right we're kind of in that
29:59 land um and that's like it's funny like
30:03 why I refer to it as magic is it's just
30:05 like it it's just
30:08 um I've spent my whole [ __ ] life
30:10 looking at sci-fi where it's like that's
30:12 not going to happen in my lifetime and
30:14 then I'm sitting here playing with the
30:17 exact same computer I've been using you
30:19 know for 30 years well not same computer
30:21 but same
30:22 basic interaction and all of a sudden
30:25 it's acting different and it's it's it's
30:29 wild I'm so ashamed of the GPT store
30:32 that I don't want to make a follow-up
30:35 demo I promised my audience so Cena
30:38 here's the
30:40 deal keep building keep showing stuff
30:44 just because the the GPT store is not
30:47 what it it needs to be um don't stop
30:51 sharing
30:52 [ __ ] talk about the store and you can
30:55 talk about hey this has got a long way
30:56 to go right now it's basically you can
30:58 get stuff up there and people can find
31:00 it are people going to come to that
31:01 store because of the store no but if you
31:05 build really good gpts that people can
31:07 use put them out there as a calling card
31:09 build calls to action for your services
31:12 into the gpts just keep building keep
31:14 building keep building keep talking
31:16 about it if you've got an audience keep
31:20 going and if you're [ __ ] bored with
31:22 chat GPT show them image stuff make
31:24 videos do a suo song
31:29 and I think that's what what I'm going
31:30 to do I think what I'm going to do for
31:31 the next next few
31:33 days
31:36 um is is I'm going to take the the list
31:40 of of recommended sites that I tell
31:42 people to go check out and I'm just
31:44 going to pick a random one or two and
31:46 we'll go look at them and see how they
31:48 are um because these things are changing
31:50 really really fast and so we'll we'll
31:52 we'll just sort of putter around to to
31:55 different things um if you got questions
31:58 about AI pop them in the comments below
32:00 the mods do a really good job of of you
32:02 know sort of floating floating questions
32:05 up so I can see them because as as I
32:07 start demoing stuff I get I just you
32:09 know I I kind of lose my mind and can't
32:12 look at everything so so just be patient
32:15 and your your your questions will will
32:18 be floated to the top oh that's really
32:21 nice who did that Kevin
32:22 Vasquez I'm
32:26 graduating
32:28 someone's got their AI learning lab
32:34 certificate if you're really nice if
32:37 you're really
32:38 nice uh I'll give you an Irregulars I
32:41 think we have to make some Irregulars uh
32:43 diplomas I think I think it's
32:48 important question I get wait question I
32:52 get is Will these gpts and info be
32:55 available for those not on a play paid
32:58 plan not until further notice Lori I
33:02 mean as far as I can tell um gpts right
33:06 now
33:11 are what did Sam mman say I wrote it on
33:13 my
33:21 whiteboard he said something about chat
33:24 GPT and he
33:26 said
33:28 or cat gp4 and he and he he basically
33:31 said chat gp4 is just for practice like
33:35 he he kind of disparaged chat
33:38 gp4
33:41 as that's just to get people ready for
33:43 what's
33:44 coming and so I think that gpts kind of
33:47 live in that space so what I take from
33:51 gpts is we have now we have now moved
33:55 into a world where non-programmers are
33:57 going to be able to generate
34:00 sophisticated applications like anyone
34:04 will be able to gen generate
34:05 sophisticated
34:07 applications
34:10 um the form of gpts right
34:15 now my guess is
34:19 that they won't be anything like that 6
34:22 months from
34:25 now and
34:29 in in my Olden
34:31 timey
34:33 mentality I'd have been like well [ __ ]
34:35 it why would I build
34:37 it but in my in in you know knowing what
34:41 I know now or you know have having been
34:44 having been through this before with the
34:45 worldwide web what I know what I know is
34:48 that any time that you spend right now
34:51 building
34:53 stuff building prompts building gpts
34:56 figuring out how to do do images
34:57 figuring out how to make videos that
35:00 have music behind them
35:02 and all that sort of
35:05 stuff is training you to be ready when
35:10 the cool [ __ ] comes and you're like but
35:12 Kyle it's already so cool it's already
35:14 so amazing yeah we you don't I don't
35:18 know if you follow Matt wolf Matt Wolf's
35:20 the one that put together futurepedia
35:23 doio he said um he said in a tweet like
35:27 two days
35:29 ago he said
35:32 um like he
35:34 was he was thinking that the ability to
35:39 do like a 30 or 60 Minute video from a
35:44 prompt was probably two or three years
35:47 away and and he said I was
35:50 wrong wait till you see what's coming in
35:53 two or three
35:54 months so I just think I think on all
35:57 levels I think when when Gemini Ultra
36:00 from Google comes out I think when GPT 5
36:03 comes out which I mean that could be
36:04 later in the year but we don't
36:07 know when whatever the next generation
36:10 of these video tools come
36:12 out um we're going to have to relearn
36:18 everything and you should be building
36:21 today because here's what you're doing
36:24 you're establishing yourself as an
36:27 expert in generative
36:29 AI You're Building skills and and more
36:32 than building skills you're
36:35 building an awareness that something new
36:38 is
36:39 possible and not an awareness like oh
36:42 something new is possible yeah ai's
36:43 going to do cool stuff no like
36:46 internalizing like oh [ __ ] I'm a
36:50 horrible illustrator but that no longer
36:52 matters if I want to make art oh [ __ ] I
36:55 can't sing or write music super good but
36:59 now I can oh [ __ ] I wasn't a
37:01 cinematographer but now I am oh [ __ ] I
37:04 wasn't an application developer but I've
37:06 made if your name's Daisy Thomas more
37:09 than 80 gpts 80 applications she's put
37:12 out in the
37:15 world Suzanne's done 40 an's probably
37:19 done 40 or
37:21 50 right so
37:25 so that's why to do this
37:28 now but I wouldn't get too caught up in
37:34 yeah but no one's going to use my GPT
37:36 because they don't have access to GPT 4
37:38 at some point open AI will probably do
37:40 that just understand the game you're
37:42 playing and do the best you can within
37:45 the game you're
37:46 playing and so open AI puts out a shitty
37:49 store
37:51 okay the fact that non-programmers still
37:54 can still make thank you very much the
37:56 fact that fact that non-programmers can
37:58 still put together these incredibly
38:00 sophisticated
38:02 gp4 custom
38:05 gpts is
38:09 miraculous people are discovering whole
38:11 new ways to express themselves because
38:13 of
38:15 this and so what those people are going
38:17 to have an advantage on the rest of the
38:20 people is when the next version of these
38:23 things comes
38:24 along they're going to be like I know EX
38:27 exactly what I want to do now it can do
38:29 that now holy [ __ ] let me take what I
38:32 learned in my
38:33 gpts and jump over to whatever the next
38:36 version of them becomes and blow the
38:38 doors off the thing Source Camp I hope
38:41 what I'm learning gives me a bit of
38:42 foundation for the next oh it absolutely
38:45 will
38:46 Corey see see I think Corey's doing it
38:49 right Cory a lot of people in here are
38:51 doing it right what a lot of people in
38:53 here are doing is is
38:58 they're taking who they are as human
39:00 beings and they're finding new ways to
39:04 manifest their talents with these new
39:07 tools and it's really [ __ ]
39:10 exciting and it's really empowering and
39:12 it's really
39:16 inspiring and so yeah Corey I think you
39:18 I think I think all of what you're
39:20 learning is
39:22 transferable who should I follow to
39:24 learn prompting um
39:34 here's why it's a complicated question
39:35 it depends what you want to do with your
39:37 life if if you're just talking about
39:39 learning prompting as you know using
39:42 chat
39:43 gbt type tools
39:46 um you can learn structured prompting
39:49 structured prompting is not a bad thing
39:51 to learn um you can you can follow re
39:53 Rachel Woods she's one of the people I
39:56 recommend you follow
39:59 um there are all sorts of videos on
40:02 prompting the the challenge with it is
40:06 like like this this is one of
40:08 those I feel like it's it's important
40:11 right now to
40:13 play um so you can so what I would say
40:16 is follow some people to learn some
40:19 fundamentals but don't get to locked
40:22 into there's a way to
40:25 prompt
40:27 because if you look at what's coming
40:30 with Gemini
40:32 Ultra where you've got images are are as
40:36 important in the prompt as
40:38 words or if you even in chat gp4 right
40:42 now if you use GPT Vision you can upload
40:45 a picture of a whiteboard and just say
40:47 what is this and it analyzes what was on
40:51 that whiteboard that's that's not really
40:53 a prompt but that is like You' you've
40:55 got to understand
40:57 that prompting is just an input to the
40:59 computer system and there's going to be
41:02 lots and lots of ways to prompt so so
41:05 learn the concept of prompting more than
41:08 the Tactical
41:09 details because these systems as they
41:12 get better at
41:13 reasoning and as they get better at
41:16 here's a good here's a good word for you
41:18 get your get your notepads out people as
41:21 they get better at a gentic
41:24 behavior where they're taking action on
41:27 your behalf they're going to prompt
41:29 themselves all you have to do is give
41:31 them a
41:33 goal so if you want to be a developer
41:37 and work on the back end of these
41:39 systems where you're putting the prompts
41:40 in that make these things do their
41:43 stuff then it probably makes sense to
41:45 learn a more sort of technical approach
41:48 to prompting but if you're just a human
41:50 being trying to figure out how to do AI
41:53 just start playing follow Rachel Woods
41:56 she's got a very she she sort of attacks
41:58 this stuffff in a much more structured
42:00 way than I do right I'm this is Chad add
42:03 this Channel's Chad
42:06 add like my prompting style is fast and
42:09 loose like can I get it can I get it
42:13 there it's 80% awesome I'm
42:16 done other people are like I'm going to
42:19 sort of prompt the [ __ ] out of this
42:20 thing and I'm going to do details and
42:22 I'm going to she's more like
42:24 that there's no real white right way
42:27 right now and it's changing like it's
42:29 changing dramatically
42:33 fast in
42:36 fact a thing that would be really
42:39 interesting actually let's go to chat so
42:41 here we are chat GPT right and so you're
42:44 like well I don't know anything about
42:45 prompting okay fine let's go over to the
42:48 internets and look up
42:52 uh
42:54 uh Civil War
43:00 letters
43:13 images that's a cool
43:24 one okay I'm going copy that image I'm
43:27 going to go back to my chat GPT I'm
43:29 going to paste that image and say what
43:33 is
43:44 this appears to be an old postcard with
43:46 both an illustration and handwritten
43:48 text it depicts a Native American scene
43:51 with an individual individuals in
43:53 traditional clothing some on Horseback
43:55 with the American flag
43:57 present the handwriting seems to be a
43:59 personal message from someone named
44:01 Henry who is a
44:03 recipient name isn't entirely clear
44:05 because it overlaps with the
44:07 illustration uh so you can say can you
44:11 read
44:17 it best wishes Henry Sneed dear sister I
44:22 received your
44:25 something
44:27 let me say try again try again you can
44:30 do
44:31 it apparently um giving uh chat GPT
44:38 encouragement look dear sister I
44:40 received your loving letter on the 26
44:42 and was glad to hear from you once more
44:45 that you getting along so well I'm in
44:47 the best of health thank
44:49 God I'm not all Idol but have the sweet
44:52 feeling of success and cheerful thought
44:55 of honest days
44:58 toiling toiling Throne at large
45:02 something my only earning and saving I'm
45:04 doing fairly well and and am quite well
45:07 known here in somewhere and have a good
45:10 many
45:11 friends the message continues but it
45:14 comes in increasingly difficult all
45:16 right let's go find another one because
45:17 that wasn't really Civil War let's see
45:20 if we can get a Civil War
45:25 One that's got a
45:28 date
45:30 oops see if I can read this one copy
45:35 image and you're you're like what are
45:38 you what are you talking about I ask you
45:40 about
45:41 prompting what is
45:55 this
45:57 Dear Mother thinking of your whereabouts
45:59 look for me home
46:01 concluded to write you a few lines to
46:04 inform you I shall go to Winstead today
46:07 it's Glaston Berry from
46:10 1863 and I shake my account settled
46:14 letter appears to be a son informing his
46:16 mother of his travel plans and
46:18 whereabouts it also me mentions settling
46:22 accounts the letter provides a glimpse
46:24 into personal Communications from from
46:26 the Civil War era um what was happening
46:32 at this time and where might this person
46:55 be give the locations mentioned in the
46:57 letter Glastonberry Winstead Hartford
46:59 and
47:01 Middletown likely a person was in
47:03 Connecticut a state that was part of the
47:05 Union in the Civil War in October of
47:07 1863 significant several event
47:10 significant events were
47:13 happening and then it talks about the
47:15 aftermath of Gettysburg the Battle of
47:18 chicka malga which I've never heard of
47:21 and the siege of
47:23 Chattanooga Connecticut played a
47:25 significant
47:26 so so all of a sudden here with so what
47:29 what prompts did I write I said what is
47:32 this and then I said what was happening
47:35 at the time I don't need to use
47:37 prompting and then so so my idea was
47:39 well so maybe let's go back out here
47:41 let's go to this thing go
47:48 whiteboard
47:55 images
47:56 whiteboard
48:17 notes let's get a crappy whiteboard
48:20 thing going
48:21 here whiteboard displaying
48:25 notes
48:27 oh wait that's good this is good okay
48:30 copy copy
48:32 image chat
48:34 JP I'll do a new
48:37 chat I'll do a new chat
48:42 paste what is
48:48 this so why I'm doing this is if you're
48:50 enjoying this conversation join us at
48:52 the salon. thank you thank you thank you
48:54 thank you um
48:57 so this is figuring this thing out so
49:00 you could take a course to learn
49:02 prompting but what if you spent a
49:06 week doing nothing but non-traditional
49:10 prompting right use images as prompts
49:13 use
49:17 um take stuff from chat
49:21 GPT the chat jpt comes up with go over
49:24 toso and just say write me a song about
49:28 whatever the topic was that chat GPT
49:31 came up with based on an image
49:36 um take some image that was made from
49:40 chat
49:42 GPT um over to Runway ml turn it into a
49:46 movie like I
49:48 like I wouldn't get too wrapped up impr
49:51 prompting I made a GPT yesterday
49:54 fantastic you can upload an image and
49:56 find the font yeah oh yeah I know it's
49:59 so amazing isn't it I know I
50:03 know we are getting to play with actual
50:06 machine learning that's what that's
50:08 what see the the data scientists have
50:11 been excited about this [ __ ] for a long
50:12 time but they could do all the geeky
50:15 math stuff to make it do its magic and
50:17 that's why it remained narrow
50:21 right chat GPT you know marks the new
50:25 era of General general knowledge tools
50:29 that the rest of us can
50:32 use um let me say um make a
50:39 table
50:43 organizing the
50:54 notes
51:00 so let's say you did this brainstorming
51:02 session
51:04 here and you were assigned with creating
51:06 notes the table is simplified it looks
51:08 like
51:10 looks like you
51:17 truncated do it
51:21 all I something's up with something's up
51:25 with chat Che
51:26 it's it's being
51:28 weird I saw I saw some notes on on
51:30 Twitter
51:33 today that chat GPT is just being weird
51:37 and like people's gpts aren't
51:38 functioning right and [ __ ] like
51:44 that I'm always afraid to go over my GPT
51:47 prompt word limits so I don't prompt it
51:50 how
51:51 does how does it
51:54 work wait I'm always afraid to go over
51:57 my GPT prompts words
52:00 limits um I don't know apparently if you
52:04 go to
52:07 teams if you go to GPT
52:09 team your context window goes up to
52:12 32,000
52:14 tokens um and normal chat GPT 4 is 8,000
52:21 tokens so it looks like it looks like
52:23 when I told it to do it for real let's
52:25 do it it for
52:31 real all right so that's good so just so
52:35 now I'm going to
52:36 say
52:39 um can
52:42 you group the
52:45 notes into
52:48 categories in each of the three
52:54 sections
52:57 [Laughter]
53:05 [Music]
53:09 lunar stick what's
53:11 happening you must your ears must have
53:13 been burning I was talking about all the
53:15 all the dang gpts you've been
53:18 making I also sent you a note um I had
53:21 time today but uh just send me some
53:24 availabilities can
53:26 chat empowered pixels I'm inspired going
53:29 to start building some marketing
53:30 strategy gpts tomorrow no no no no
53:34 empowered
53:35 pixels
53:37 no you can make money
53:40 with you can make
53:42 money all right I want you to go make a
53:45 GPT right now right now right
53:49 now thank you for the dog butt I dog
53:52 butts always make me smile um I want you
53:56 to do it right now so here's what we're
53:57 going to do why is this oh because I
53:59 asked it to to categorize these
54:02 things oh wow this is oh by the way if
54:05 if you've been watching the screen while
54:07 I'm talking yeah Emilio's wife empowered
54:10 pixels today so we're gonna We we'll go
54:12 build one here
54:14 um notice what's notice what's happening
54:17 here
54:18 um so I threw in that image I said what
54:22 is this and then I said put the notes in
54:24 a table and it screwed that up and then
54:27 I said no do it do it completely don't
54:30 get lazy and then it did it and then I
54:33 said can you group the notes into
54:34 categories in each of the three
54:37 sections and
54:39 so it's it's just it went off and did
54:42 that so it's kind of prompting itself
54:44 here I mean I gave it this this little
54:48 goal and then it broke the content up
54:50 into sections and created categories and
54:54 made Separate Tables for each each of
54:55 the three
54:57 areas each suggestion represents a
55:00 community provided
55:03 idea I mean
55:06 dang let me see this I'll let me we'll
55:08 try this
55:12 um is
55:16 there any idea Within These
55:21 Concepts that is visually
55:24 interesting
55:28 the answer will probably be yes heated
55:31 seats napping
55:33 pods box
55:36 Fort hot Community cocoa check-in system
55:40 at the front desk so this is Library
55:42 stuff I
55:50 guess great I'd like you to make a 16x9
55:54 wide image of of a library
55:56 concept that
55:58 features these seven
56:01 visuals that you've
56:08 identified and again what's my prompting
56:11 here am am I learning how to like have I
56:14 learned prompt strategy from Rachel
56:17 Woods for how to do high-end fancy
56:20 structured prompting nope I'm talking to
56:23 it like I have an
56:27 intern and I'm giving it really crappy
56:32 directions and
56:34 it's responding with better than I
56:38 deserve so don't get hung up on
56:41 prompts can you incorporate a YouTube
56:44 video in a GPT you could certainly
56:46 incorporate links in there I don't think
56:49 you could embed one but you could
56:52 certainly you could certainly take a CSV
56:55 file of uh it made a bunch of different
56:58 things all right let's
57:00 do
57:03 um I'd like a single
57:07 room in the
57:10 library that has these elements in it
57:14 period as opposed to each element has
57:17 its own room which is what you created
57:20 please fix
57:24 this
57:26 [Music]
57:29 um what was the question the question
57:31 was oh YouTube videos you could you
57:34 could do a uh a CSV file with a bunch of
57:39 YouTube descriptions and links to them
57:42 and use that as your
57:43 source and then God dang it this new
57:48 dictation thing from OSX it just keeps
57:52 typing
57:54 um
57:56 so you could certainly build something
57:57 that jumps you out to specific YouTube
57:59 videos but you could also just go to
58:01 Bard and Bard does that pretty good so
58:06 so yeah I mean this is the other thing
58:09 start learning your tools all right
58:10 there's our there's our fancy uh there's
58:12 our fancy library with our napping pods
58:14 and whatever the hell else was in there
58:16 our hot cocoa here's our hot
58:19 cocoa
58:20 um but I mean I I so for for
58:26 me I could go learn prompting strategy
58:29 and prompting structure or I could learn
58:33 what this tool can actually do and just
58:35 play with it and just learn holy [ __ ] it
58:38 can do all this without me having to
58:40 learn [ __ ] well I had to learn that it
58:42 could do
58:43 this right basic question can you
58:46 explain tokens
58:48 absolutely um the good news is we've got
58:51 paid in here from Google so if I screw
58:52 this up he will chew me out and he will
58:55 fix it
58:57 okay
58:59 so to a computer there's no well there's
59:04 a difference there's no difference
59:05 between a letter and a space and a comma
59:08 and a space so tokens are either a word
59:12 a space a punctuation mark or some
59:15 fragment of a word right so so when when
59:20 um when you hear this has got an 8,000
59:23 token here let me flip let me flip here
59:27 here when you hear chaty BT has an 8,000
59:30 token context
59:36 window you shouldn't have to think about
59:38 that that is for people like Pate to
59:41 think about so tokens are basically just
59:43 the
59:44 way these machine learning models handle
59:50 information there are about it's about
59:54 there are about 70
59:55 5% words available for the number of
59:59 tokens that you're given so if you're
1:00:01 given 8,000
1:00:03 words you or if you're given 8,000
1:00:06 tokens you have about a 6,000 word
1:00:09 context window now what's a [ __ ]
1:00:11 context
1:00:12 [Laughter]
1:00:15 window so your phone is actually a
1:00:18 really good example if you if you're
1:00:20 viewing a text document on your
1:00:22 phone and you make the text really small
1:00:26 whatever you can see on your screen
1:00:29 that's kind of your context window right
1:00:32 it it's it's it's a similar metaphor to
1:00:34 chat gpt's context window so the the
1:00:37 text starts at the bottom and then it
1:00:38 Scrolls up and it Scrolls up and it
1:00:40 Scrolls up and so so let's say the first
1:00:42 thing that you put there is
1:00:46 um uh I want you to act like a silly
1:00:49 goose right and so the the phrase I want
1:00:52 you to act like a silly goose is there
1:00:54 and then you start typ typing and it's
1:00:55 acting like a silly goose and you're
1:00:57 having silly goose interactions and
1:00:59 you're having silly goose interactions
1:01:01 and then all of a sudden chat GPT stops
1:01:04 acting like a silly goose and you're
1:01:05 like what the [ __ ] just
1:01:08 happened well silly goose went out of
1:01:10 the context window and if you didn't
1:01:12 remind it somewhere in this conversation
1:01:14 that you wanted it to act like a silly
1:01:15 goose it essentially just
1:01:18 forgets um so so in chat gp4 you have
1:01:22 about 6,000 words how do you voice chat
1:01:25 with it so how you voice chat with you
1:01:28 can do it right now with pi and with
1:01:31 chat
1:01:32 GPT you do it with chat GPT with their
1:01:36 app only and same with same with pi for
1:01:40 for whatever reason I got I got Pi's
1:01:43 website to do it once and then the next
1:01:45 time I went back I couldn't get it to do
1:01:46 it I think they took the feature away
1:01:48 but
1:01:49 anyway uh Mr K day so essentially if you
1:01:53 use most of your token it will start to
1:01:55 forget the conversation correct so and
1:01:58 this is why Mr
1:02:00 K
1:02:03 um why a lot of people use Claude
1:02:06 instead of chat gbt is that Claude has
1:02:09 120,000 token context window so what's
1:02:14 the what's the famous book um I don't
1:02:17 know like like famous novels are you
1:02:20 know could be in like the 50,000 word
1:02:23 range so if you have 120,000 token
1:02:27 context
1:02:28 window that means you've got about 80
1:02:31 85,000 words okay so with Claude you can
1:02:36 literally paste an entire novel or an
1:02:39 entire screenplay or an entire research
1:02:43 paper into the prompt and what what's
1:02:45 cool what TD what what Claude does which
1:02:48 is really
1:02:51 cool so if I do that let me see do I
1:02:54 still have my screenplay up yes I do so
1:02:58 here's a screenplay I have so it's 120
1:03:01 pages so if I go to chat JP so here's
1:03:04 fancy chat GPT that can do everything
1:03:06 and P claims that I all I am is a is a
1:03:09 sick aant for Sam Alman it's not true
1:03:12 Pate because watch I paste in my
1:03:15 screenplay and I say read
1:03:19 this and it goes oops the image you did
1:03:23 was so long finally understood why it
1:03:25 forgets what we're talking about yeah
1:03:27 exactly so now watch this I go over to
1:03:30 cloud
1:03:32 cloud claw. Claude
1:03:38 doai I'll continue with my little
1:03:41 Google my little Google account that's
1:03:44 Google single sign
1:03:47 on you know why I use Google single sign
1:03:50 on because I want Google to know
1:03:52 everything about
1:03:53 me
1:03:55 so I just pasted it and look what cloud
1:03:58 does it it turns it into a pasted
1:04:01 document it says 145k
1:04:06 4,833 extracted lines and then I can say
1:04:11 um
1:04:14 please summarize
1:04:18 this and it's going to go now read my
1:04:21 screenplay
1:04:23 now you can do this with chat
1:04:27 GPT but you you what you do with chat
1:04:29 GPT is you upload it as a file
1:04:32 right but that's different than having a
1:04:36 larger chat context window Claude has a
1:04:39 larger chat context window so one of the
1:04:41 ways that I use Claude is when I take um
1:04:46 these
1:04:47 videos I've got a really long prompt
1:04:49 that I wrote to turn the YouTube
1:04:53 transcript from my two and half three-
1:04:55 hour long videos into a YouTube
1:04:57 description in chapters and so I go to
1:05:00 YouTube I copy the transcription I come
1:05:02 to Claude paste it in there
1:05:06 and then paste in my prompt that knows
1:05:10 how to you know I've figured out ways to
1:05:13 make it do that and then it makes me a
1:05:16 YouTube description title description
1:05:17 and and chapters it's really cool this
1:05:20 is really great info thank you you're
1:05:22 welcome here's a summary the script
1:05:24 centers around Jack a Vietnam War
1:05:26 veteran who owns a fishing boat in
1:05:28 Washington state in a flashback Jack is
1:05:30 s shown serving as a helicopter pilot in
1:05:32 Vietnam along with his close friend and
1:05:34 partner snake after snake apparently die
1:05:38 apparently dies in a helicopter crash
1:05:41 and another friend is killed Jack
1:05:43 returns home with money from snake to
1:05:46 buy a fishing
1:05:52 boat so I read my whole [ __ ]
1:05:54 screenplay I wrote a little synopsis of
1:05:55 it how cool is that so how did I know
1:06:00 that you just got to get curious about
1:06:02 this [ __ ] you've got to stumble on to it
1:06:04 if you go to something like
1:06:08 perplexity I don't know what
1:06:09 perplexities context window is I assume
1:06:12 it's not that big um
1:06:15 please Su Ariz
1:06:20 this
1:06:23 paste
1:06:34 huh I guess it read
1:06:46 it yeah Reddit all right so perplexity
1:06:48 has got a big context window want more
1:06:51 AI resources join us at the AI the salon
1:06:54 AI yep I'll I'll talk about that in a
1:06:56 second here Brandon that's awesome thank
1:06:57 you
1:06:59 um so perplexity surprised me there and
1:07:02 actually ate that screenplay and you
1:07:05 know chewed it up and spit it out it did
1:07:07 it really fast too um there's a lot of
1:07:11 people in this channel that are huge
1:07:13 fans of perplexity um I was just given
1:07:16 access to the pro version of it for a
1:07:19 year because I bought a rabbit R1
1:07:20 because I know how to live and make good
1:07:22 life
1:07:23 choices
1:07:27 but one of the really cool things about
1:07:28 perplexity it's got this thing called
1:07:30 Focus so if I just want to search if I
1:07:32 want it to answer me questions but I
1:07:34 only want it to use academic research as
1:07:38 the The Source material I could say um
1:07:41 how can I
1:07:44 chapiz
1:07:46 video
1:07:49 using um AI for the
1:07:53 visual
1:08:00 elements understanding the
1:08:04 question let's
1:08:06 see scene changes object
1:08:10 recognition look at this so so it says
1:08:12 what specific visual elements do so this
1:08:16 okay not for nothing people this is now
1:08:19 I'm getting excited I I I know I've
1:08:21 known for a while I need to do more with
1:08:22 perplexity because people seem excited
1:08:25 about it um and I didn't know why so
1:08:29 perplexity now has this thing in it
1:08:30 called co-pilot which is actually
1:08:32 unfortunate because that's Microsoft's
1:08:33 whole [ __ ] product name is co-pilot
1:08:35 so they shouldn't have named a feature
1:08:38 within their competitor of co-pilot the
1:08:42 name of their competitor
1:08:47 anyway these are the people making this
1:08:49 technology are research Labs they're not
1:08:51 marketers people we've got to give them
1:08:53 some Grace but that's really dumb
1:08:57 anyway they knew it was coming they knew
1:08:59 it was called GitHub co-pilot they knew
1:09:01 it was coming anyway so so I asked it
1:09:06 how can I Chapter IE video and then
1:09:08 co-pilot says what specifical V uh what
1:09:11 specific visual elements do you want to
1:09:13 use for chapter zation scene changes
1:09:16 object recognition emotional detection
1:09:19 so I just want to do the visual things
1:09:21 right now I might want to do object
1:09:23 detection later but I can go I can go
1:09:25 back and try
1:09:26 that and then if yes um you know throw
1:09:30 it in here what's exciting about this to
1:09:33 me is that this essentially just wrote a
1:09:37 little mini application for us so it it
1:09:41 took my question it thought about it and
1:09:45 then it said Hey in order for me to give
1:09:47 a better question here let me put some
1:09:49 interface together so it put together
1:09:51 you know
1:09:52 multi-choice um you know
1:09:56 checkboxes asked me that and now it's
1:09:59 now it's going to go respond
1:10:09 right it'll be curious to see if it if
1:10:12 it brings up any
1:10:14 papers that's what I was hoping it would
1:10:16 do we'll see we'll
1:10:18 see related um so it didn't give us any
1:10:23 of that stuff how accurate what benefit
1:10:25 so I can say um please show me any
1:10:31 papers
1:10:33 that
1:10:35 uh talk about tech that does
1:10:43 this searching scholar found eight
1:10:47 sources show five more so here's eight
1:10:53 papers
1:10:54 robust and realtime handling of
1:10:57 something real-time hand detection and
1:11:00 tracking of monocular video
1:11:03 so so the the thing to understand here
1:11:11 folks why does he sing like that that's
1:11:13 really annoying isn't it yeah that's
1:11:16 what he does it's annoying is he trying
1:11:19 to chase people out sometimes he does
1:11:22 yeah sometimes people away
1:11:25 scared but they come back he's harmless
1:11:29 he's got a nice
1:11:30 dog
1:11:37 [Music]
1:11:48 dude um okay
1:11:51 so at the AI Salon which I'm going to
1:11:53 talk about in a second we have this this
1:11:55 page called AI salon tool
1:11:59 recommendations and in this list the the
1:12:02 thing at the top is chat GPT because
1:12:04 chat GPT is still the damn
1:12:06 best um when that changes it will no
1:12:09 longer be at the top if that changes no
1:12:11 one's caught them yet um but then right
1:12:14 below chat GPT I've got what I consider
1:12:18 the four biggies of large language
1:12:20 models Claude Bard perplexity and
1:12:23 co-pilot from Microsoft see here's a
1:12:26 whole product called co-pilot perplexity
1:12:28 you might want to rechange your feature
1:12:30 name unless that's because Microsoft
1:12:33 invested money in you and then it's on
1:12:36 purpose which maybe it is who knows um
1:12:40 and then you've got specialty large
1:12:42 language models like Pi Gro llama 2
1:12:44 which llama 2 is actually a whole world
1:12:47 of open-source large language models so
1:12:50 if you
1:12:51 are frustrated with the safe safety
1:12:54 guard rails of big
1:12:57 commercial tools like chat gbt and
1:13:00 perplexity then you can go build your
1:13:02 own [ __ ] without any safety guard rails
1:13:04 on it um and there's there's po so I've
1:13:07 got right now and this list I I
1:13:10 personally maintain this list this is my
1:13:12 curated
1:13:14 list in this is in my
1:13:18 opinion [ __ ] you should know
1:13:22 about and what I can promise you is if
1:13:26 it's on this list it won't be a waste of
1:13:28 your time with the sole exception of
1:13:30 grock which which which is just stupid
1:13:33 right now but it's got a sense of humor
1:13:35 so if you want to see a chat GPT that
1:13:37 can make jokes grock is kind of funny uh
1:13:40 but it's just not useful and it cost you
1:13:42 16 bucks a month to be able to watch it
1:13:44 make stupid jokes um but it's going to
1:13:47 get better right Elon Musk is going to
1:13:49 make sure that that thing turns real so
1:13:51 so it's it's it it's there it's it's
1:13:56 there just in anticipation of it not
1:13:58 sucking at some point in the future
1:14:00 but at some point I'm going to go play
1:14:03 with grock and it's going to be really
1:14:05 good right at some point I'm going to
1:14:08 like like what I just discovered with
1:14:10 perplexity that it's kind of writing
1:14:12 this this little mini computer
1:14:14 application for me to be able to you
1:14:17 know refine my search um that's pretty
1:14:20 slick every time I go use one of these
1:14:23 tools some new thing shows up like
1:14:25 Microsoft
1:14:27 co-pilot
1:14:29 um
1:14:31 yesterday I
1:14:33 discovered that it's got co-pilot
1:14:37 gpts so it's got a vacation planner a
1:14:39 cooking assistant a fitness trainer
1:14:42 so Microsoft invested $1 billion in open
1:14:46 AI open ai's got these things called
1:14:48 gpts Microsoft has this thing called
1:14:51 co-pilot which is like chat GPT
1:14:54 and now it has these things called
1:14:55 co-pilot gpts and we knew this was
1:14:58 coming but like this wasn't part of the
1:15:02 site two weeks ago and now it is and
1:15:08 so if you want to be AI literate for the
1:15:14 next 18
1:15:18 months you just need to be
1:15:20 curious and you need to be
1:15:24 adventurous and you need to be in
1:15:29 community because there's so much
1:15:31 happening so fast and it's changing just
1:15:36 dramatically that if you're not in a
1:15:38 community of
1:15:40 people that is figuring this [ __ ] out
1:15:43 with you it it gets I what I can promise
1:15:46 you is it gets very overwhelming very
1:15:50 stressful very um
1:15:54 like you just feel like you can't ever
1:15:56 possibly learn enough and you're like oh
1:15:58 my God everybody knows more than I do
1:16:01 help what to
1:16:05 do if you can find a
1:16:07 community that you can connect with so
1:16:10 you can learn these
1:16:12 tools you're going to be okay that's
1:16:15 what this group is the salon. so I want
1:16:17 you to go to that website now this
1:16:20 doesn't cost any money so I'm not trying
1:16:21 to sell you anything like
1:16:24 you can make money with that's not this
1:16:27 if I turn this channel channel into a
1:16:29 sleazy Commerce Channel I'll let you
1:16:31 know right I won't I won't make fun of
1:16:34 it with a gold button I'll actually just
1:16:35 go you can make money with chat jbt just
1:16:37 send me 1995 a
1:16:39 month we we might go there I might turn
1:16:42 this into an NPC NPC Channel where I
1:16:45 just say hot dog hot dog rabbit rabbit
1:16:49 and then make lots of
1:16:52 money I might
1:16:54 twerk right if I need money I I might
1:16:57 twerk we could turn this into a twerking
1:16:59 channel do you see all those Minnesota
1:17:00 dance dancers doing their
1:17:02 Twirls I could turn it into a dance
1:17:05 competition Channel where all I do is
1:17:07 Just Dance for
1:17:08 hours but it's not there right now and
1:17:11 neither is the salon so the salon this
1:17:13 is a group that I started the week after
1:17:16 chat gbt came out because I've been in
1:17:19 this movie
1:17:21 before I was an entrepreneur in the
1:17:24 early days of the worldwide web in the
1:17:26 mid
1:17:27 90s in New York
1:17:30 City and started a similar group called
1:17:32 the worldwide web artist
1:17:34 Consortium and
1:17:37 um what I learned back then is that when
1:17:41 you're this early when you're at the
1:17:43 beginning
1:17:45 of the thing that is going to change the
1:17:49 world and you recognize that before
1:17:52 before other people do one it's really
1:17:57 exciting and two you will feel like a
1:18:01 clueless idiot for a lot of years
1:18:04 why because you can't get a degree in
1:18:07 this right now well but you can go to
1:18:09 Stanford and get a degree in machine
1:18:11 learning yep but that's not
1:18:13 this chat GPT November 30th 2022 so just
1:18:18 over a year ago marks the beginning of a
1:18:21 new era just just like the worldwide web
1:18:25 marked the beginning of a new era for
1:18:27 the internet the internet had been
1:18:29 around for
1:18:31 decades and Tim burner Lee comes up with
1:18:34 this hypertext transfer protocol
1:18:38 thingy that allows you to underline a
1:18:41 word and do some magical prompt behind
1:18:46 the scenes that when you click on that
1:18:49 word it takes you to some other server
1:18:51 Halfway Around the World
1:18:55 and so the worldwide web went from a
1:18:57 command line interface that only
1:18:59 researchers and scientists and Engineers
1:19:03 used because they love command line
1:19:07 interfaces to click click click click
1:19:12 click and the rest of us got to use it
1:19:15 and so being there at the beginning of
1:19:17 that was absolutely remarkable
1:19:20 but nobody knew how to build websites
1:19:23 nobody knew what websites were like like
1:19:27 browsers didn't exist like the browsers
1:19:29 that that had images in them did not
1:19:33 exist I I I created one of the first
1:19:36 online magazines a magazine called Urban
1:19:38 desires and and the first couple of
1:19:41 issues all of the images are left
1:19:43 Justified on the page because the center
1:19:46 tag didn't apply to images it was that
1:19:50 early right that's where we are with AI
1:19:52 right now
1:19:54 the difference with AI is the tools are
1:19:56 significantly more sophisticated and
1:19:58 there's an entire infrastructure in
1:20:00 place to be able to deliver them
1:20:02 right so that's what allowed when when
1:20:05 chat GPT was
1:20:07 launched it went from zero users to 100
1:20:10 million users in six
1:20:16 weeks Word of Mouth
1:20:19 only open AI didn't think Chachi was a
1:20:23 big deal they thought maybe it would get
1:20:25 a blog post or two in 5,000
1:20:28 comments that's what they said the night
1:20:30 before they launched it six weeks later
1:20:32 they had a 100 million
1:20:34 users Word of Mouth
1:20:40 only holy [ __ ]
1:20:44 right so what I know is nobody knows
1:20:47 what they're doing right now so the only
1:20:48 way to to navigate this space is do it
1:20:51 with others right so like why is there a
1:20:53 crew here heading off into the promised
1:20:55 land through all the Adventure and toil
1:20:57 of of the
1:20:59 adventure cuz it's it's the it's that
1:21:02 that's it this this I know like this I
1:21:05 know there's a lot of stuff on this
1:21:07 channel I don't know right there's a lot
1:21:08 of stuff on this channel I'm just
1:21:10 bullshitting my way through because
1:21:11 nobody knows anything including me but
1:21:14 what I know is that we're so early in
1:21:17 this thing and it is so profound and so
1:21:20 different and we and we so don't know
1:21:22 what what these tools are how they work
1:21:25 what the implications of them are what
1:21:27 work is going to look like what our jobs
1:21:29 are going to look like if we're going to
1:21:30 have jobs how we're going to have jobs
1:21:32 how money gets made all of that shit's
1:21:34 going to change but that's going to take
1:21:36 10 or 15 or 20
1:21:40 years but the next three years are the
1:21:44 primordial soup of that
1:21:47 future and so all you have to
1:21:50 do is be in the conversation show up to
1:21:53 my [ __ ] lives Cindy [ __ ] is at the
1:21:55 airport right now getting ready to fly
1:21:57 to Asia she's got some hours to kill and
1:21:59 she's here I think I think I think that
1:22:03 says a little bit more about her than
1:22:05 than us
1:22:07 right no but she's be she's inspired by
1:22:10 this oh what what's Cindy Coon's job by
1:22:12 the way she's a
1:22:16 futurist she's been a futurist for
1:22:20 decades she's a futurist
1:22:24 she is lit on fire about what's
1:22:28 happening with generative AI she is lit
1:22:31 on fire about being part of this
1:22:34 community as are a lot of the Irregulars
1:22:37 the people that show up here regularly
1:22:39 we call them the Irregulars they come
1:22:41 here night after night after night after
1:22:43 night and you're like but Kyle you're
1:22:45 supermodel beautiful of course they come
1:22:47 night after night I I get it I get it
1:22:51 thank you for the compliment it's it's
1:22:53 super sweet of
1:22:55 you I'm like an old fat guy on Tik Tok
1:22:58 why are people coming here night after
1:23:00 night after
1:23:02 night because on a regular basis we'll
1:23:06 stumble onto something where it's like
1:23:08 oh holy [ __ ] that changes
1:23:14 things and then we'll say o the GPT
1:23:17 stores coming out it's going to change
1:23:18 everything everyone's going to be great
1:23:21 maybe you won't need gp4 to use G gpts
1:23:23 we're going to make money cuz he said
1:23:24 there'd be Revenue sharing and then they
1:23:26 launched the store and we all kind of
1:23:27 collectively discover huh [ __ ] well that
1:23:30 was a
1:23:33 dud most people haven't even heard of
1:23:35 chat
1:23:36 GPT much less have a point of
1:23:41 view on the chat GPT
1:23:45 store it's impossible to learn at all
1:23:48 just pick things that are of interest
1:23:49 that interest you the most that's
1:23:50 exactly it Source Camp is just find the
1:23:53 [ __ ] that resonates with you and start
1:23:55 playing and then why you do it with
1:23:56 other
1:23:57 people is that someone with a completely
1:24:01 different skill set you know just like
1:24:03 when you go on an adventure like if we
1:24:04 were going to write this as a screenplay
1:24:06 right you're going to have the the
1:24:08 person that does the magic you're going
1:24:09 to have the muscle you're going to have
1:24:12 um you know the the Mage the you're
1:24:15 going to have the
1:24:17 wizard you're going to have the really
1:24:19 smart lock
1:24:21 picker everyone's got different skills
1:24:23 right so no one's got all the skills
1:24:25 right
1:24:26 now even to learn this
1:24:30 stuff because you keep coming back night
1:24:33 after night Kyle and we have fomo
1:24:36 exactly well you know I mean there's a
1:24:38 selfish reason that I keep coming back
1:24:41 night after night here and that's that
1:24:45 it keeps me in the
1:24:48 conversation I just I just know about
1:24:50 myself I am easily distracted distracted
1:24:53 it's it's one of the one of the joys and
1:24:56 tyrannies of my brain right it's neuros
1:24:59 spiciness at its
1:25:05 best and I know that this is too
1:25:07 important not to be in the conversation
1:25:10 it's too important I was looking at
1:25:12 autogen seems really great yeah the the
1:25:15 autonomous agent stuff is is really
1:25:17 exciting um I'm sitting on the sidelines
1:25:20 with that stuff right now just because I
1:25:22 don't I don't have time to dig around in
1:25:24 the open source the more technical stuff
1:25:26 but yeah it's if you're if you're
1:25:28 technical at all playing around with the
1:25:30 open source projects like autogen and
1:25:32 Lang chain and things like that it's
1:25:34 actually really
1:25:35 important all right so you go there you
1:25:37 go to the salon.
1:25:40 aai and it's going to take you to a link
1:25:51 tree
1:25:55 and the first link is to the AI Salon uh
1:25:58 Community the second link is actually
1:26:00 really interesting too if you click on
1:26:02 that it takes you to a thing called the
1:26:04 GPT for good
1:26:05 directory this was something that I
1:26:08 teamed up with Ann Murphy from empowered
1:26:11 fundraiser and then Vicky Baptist built
1:26:14 this website for us in real time while
1:26:16 we were doing this event so in 7 days we
1:26:19 put together this event where we were
1:26:21 going to teach as many people as we
1:26:22 could how to make gpts for nonprofits
1:26:25 this was in in anticipation of the store
1:26:27 coming out was was the
1:26:29 idea we wanted to teach as many people
1:26:31 as possible to build gpts for nonprofits
1:26:34 and in 24 hours we we did eight sessions
1:26:37 and did four I did
1:26:39 four this community built and shared
1:26:43 more than 150 gpts for
1:26:47 nonprofits and so you can just go see
1:26:49 like it's they're super inspiring and
1:26:51 the site's beautiful and then
1:26:53 what what Vicki did was she took the
1:26:55 instructions that we gave people for the
1:26:58 event and she built a page here so if
1:27:00 you go to this page and you want to make
1:27:02 your own GPT for a nonprofit or frankly
1:27:05 for anything it it's eight steps they're
1:27:08 really easy like building a GPT is
1:27:11 copying and pasting that pair of
1:27:13 sentences and replacing the bracketed
1:27:16 things with your org name and website of
1:27:18 the nonprofit that that little thing
1:27:21 will build your GP
1:27:23 It's [ __ ] amazing anyway that's that
1:27:28 but if you go to the AI Salon
1:27:32 community and go to the welcome to the
1:27:36 salon section you'll see that that
1:27:39 groovy picture that was made in uh
1:27:42 Dolly not that it matters but it
1:27:48 was we're a community of extraordinary
1:27:50 adventurers exploring the AI unknown and
1:27:53 you know extraordinary adventurers why
1:27:55 why adventurers like we're choosing to
1:27:57 be on this
1:28:00 adventure and you know it's
1:28:02 funny like like you know it's it's good
1:28:06 and nerdy to have a good you know kind
1:28:08 of D action you know fantasy Adventure
1:28:11 kind of theme
1:28:15 um but if you think about the fear going
1:28:18 on right now with AI the fear and the
1:28:20 anger and people are afraid of it it's
1:28:22 going to kill what if the robots kill us
1:28:24 and it's stealing and it's ruining
1:28:26 artist's lives and all that sort of
1:28:29 stuff going
1:28:32 on it's kind of like in one of those
1:28:35 fantasy novels or those fantasy movies
1:28:38 where we just keep hearing that people
1:28:40 go out into the woods and they don't
1:28:42 come back and then we keep hearing large
1:28:45 Roars in the
1:28:47 distance and and there's there's all
1:28:49 this fear of what's out
1:28:51 there and it becomes clear that
1:28:54 something needs to be done but nobody
1:28:56 wants to go [ __ ] out [ __ ] out in
1:28:59 those woods because they know they're
1:29:00 not coming back right and like whatever
1:29:03 that thing is that's making all that
1:29:05 noise it sounds scary as
1:29:08 [ __ ] and then what
1:29:11 happens someone in the community goes we
1:29:14 got to go I'm going to go I'll be the
1:29:17 muscle and then you know someone else
1:29:20 goes you know well I can do m spells and
1:29:23 someone else goes and right and so the
1:29:25 the five five or seven merry adventurers
1:29:28 put their wacky band together three of
1:29:30 them don't want to be there one of
1:29:31 them's an [ __ ] two of them are super
1:29:34 nice and are going to fall in love like
1:29:35 we know the whole
1:29:37 story but they choose to go on that
1:29:42 Adventure they choose to do
1:29:46 it and it ends up saving the town and it
1:29:49 ends up transforming them right it's
1:29:52 Joseph Campbell it's the hero's journey
1:29:55 so this is a community of people that
1:29:57 are choosing to be on this adventure
1:30:00 this is a community of people that in
1:30:02 the face of their friends and family
1:30:05 going you know what Sally I don't want
1:30:07 to hear about AI anymore AI is stupid
1:30:10 why are you talking about AI all the
1:30:11 time you used to be so cool and now it's
1:30:13 just like Ai Ai and and did you ever go
1:30:16 to the AI learning lab on Tik Tok I
1:30:17 don't care about Tik Tok I don't care
1:30:19 about your stupid AI learning
1:30:21 lab
1:30:24 and they keep going yeah I know you're
1:30:25 scared but you want to see something
1:30:26 cool you want to see something
1:30:28 cool I don't even know what a GPT is or
1:30:31 does I'll show you in a second here June
1:30:34 Mary so anyway so come here and just
1:30:37 know that that's who this community is
1:30:39 and so we talk about that and then we
1:30:41 also talk about our values and the
1:30:42 values are really important to me please
1:30:44 read them if you come here if this is
1:30:46 sounding exciting to you I'm talking it
1:30:48 up it doesn't cost anything by the way
1:30:52 you're like oh sure here it comes it's
1:30:54 going to be $49.95 a month for the next
1:30:56 five months no doesn't cost you anything
1:30:59 it's
1:31:01 free but it does have a cost I want you
1:31:04 to read the values my request is read
1:31:06 the values and like take them in
1:31:10 curiosity we want people that are
1:31:12 curious generosity this is a big one I
1:31:15 don't want people in here that learn
1:31:16 some [ __ ] and hoard
1:31:18 it I want people in here that are like
1:31:20 oh I figured out something o oh how'd
1:31:23 you do that well I took this and I duct
1:31:25 taped it together with that and then it
1:31:27 made this and then this is what came out
1:31:28 and people like holy [ __ ] that's
1:31:31 amazing that's
1:31:34 generosity when you're when you're this
1:31:36 early in a new
1:31:40 era you can afford to be generous in
1:31:42 fact it's a requirement because nobody
1:31:45 knows what the [ __ ] they're doing and
1:31:46 the thing that you learn that you think
1:31:48 is the proprietary gold that's going to
1:31:50 make you rich is going to be completely
1:31:53 useless 2 months from now cuz the
1:31:56 technology will have moved on so you
1:31:58 might as well share it with the
1:31:59 community and everyone gets to
1:32:01 learn generosity
1:32:06 exploration how do I how do I know about
1:32:08 perplexity how do I know about Claude
1:32:10 how I I because I'm like I got to go
1:32:13 figure this out I got to let's go into
1:32:15 the woods let's start mapping it
1:32:17 collaboration do it with other people
1:32:20 connect with other people be generous
1:32:21 connect with other people thank you
1:32:23 Empower and I know I'll show you how to
1:32:25 make a GPT cuz we've got we've got to
1:32:27 explain what they are and how to make
1:32:28 them but we're going to make one
1:32:30 tonight inclusion people come from all
1:32:35 over there are artists in here there are
1:32:37 technologist in here they are PhD
1:32:41 mathematicians this guy Kyle's the real
1:32:43 deal thank you rock appreciate
1:32:48 that there's blind people in here
1:32:53 Rudy's in here from South Africa there's
1:32:55 people from all over the
1:32:59 world every diff every imaginable
1:33:02 profession every imaginable level of
1:33:06 understanding of this stuff including I
1:33:10 was just scrolling through Tik Tok and I
1:33:12 thought it was fat Alec Baldwin and now
1:33:14 I'm just what's he talking
1:33:17 about like that level of experience is
1:33:19 absolutely fine here
1:33:23 and that person that is that clueless
1:33:26 about AI right now will probably be the
1:33:29 one that gives you some remarkable
1:33:31 breakthrough 6 months from
1:33:33 now because they're bringing some life
1:33:36 experience or some way of looking at the
1:33:39 world that you never would have
1:33:40 considered and they'll use
1:33:43 tools differently than you do so that's
1:33:46 why you include people and the final one
1:33:47 is
1:33:50 empathy this [ __ ]
1:33:52 scary it is scary and as excited as I am
1:33:56 about this and as Evangelistic as I am
1:33:58 and as much as this seems like Jesus
1:34:01 this is this is he's up on a soap
1:34:06 box this shit's also really scary right
1:34:08 this is existential if you're an artist
1:34:10 if you've been an illustrator for 30
1:34:13 years and you go look at what mid
1:34:16 Journey can
1:34:18 do it's a little how do they say [ __ ]
1:34:21 terrifying
1:34:23 Ling if you've been a
1:34:27 copywriter for 20 years in the ad
1:34:30 business and all of a sudden there's a
1:34:32 thing that can write 150 ads in 38
1:34:35 seconds and they're
1:34:38 good because it also wrote The Creative
1:34:44 brief that's [ __ ]
1:34:50 terrifying so this is a community that
1:34:53 understands that sometimes people are
1:34:54 going to come in here and they're going
1:34:55 to be pissed off they're going to be
1:34:57 scared and they're going to be like this
1:34:59 is evil your people are
1:35:01 evil yep we understand I understand how
1:35:04 it's that feels like that want to see
1:35:06 something
1:35:08 cool so anyway go join and then and then
1:35:11 my request is if you like all that in
1:35:13 the community section of the site the
1:35:15 very first thing we have is introduce
1:35:17 yourself so my request is use your real
1:35:20 name ideally put your picture there if
1:35:22 you're
1:35:23 comfortable because we want to get to
1:35:25 know one another and then introduce
1:35:27 yourself tell us who you are tell us
1:35:28 what you do tell us where you are on
1:35:30 this journey my daughter-in-law in the
1:35:32 Philippines says you can make money with
1:35:34 ch
1:35:37 gbt you can make money
1:35:39 with that's
1:35:41 awesome um and then also what you should
1:35:43 do so after you introduce yourself go to
1:35:45 the show int tell channel so the show
1:35:47 int tell channel is just a [ __ ] blast
1:35:49 so every night when I'm doing this
1:35:50 because you know I'm all over the place
1:35:52 right so as people that come here
1:35:54 regularly are like oh here goes Kyle off
1:35:56 on one of his
1:35:58 tangents they'll start collaborating and
1:36:01 jamming and making images and sharing
1:36:03 them here and and they take you know
1:36:05 little stupid things I say and they turn
1:36:06 it into there you go little memes right
1:36:10 and it's it's happening in real time and
1:36:12 like some of the shit's beautiful some
1:36:13 of the shit's hilarious a lot of the
1:36:15 stuff is trolling me in the best
1:36:18 possible
1:36:19 way these are amazing oh it's
1:36:21 Valentine's coming enough right
1:36:23 beautiful a that's beautiful nice nice
1:36:26 Papa
1:36:27 JJ oh that's cool
1:36:30 wow that's pretty
1:36:33 gorgeous the hand Shadows with the the
1:36:36 people as the object wow that's super
1:36:41 cool Cory Sandler everything is moving
1:36:44 so fast them
1:36:46 haves wait for the oh for them haves but
1:36:49 the have knots need this technology too
1:36:52 well that's that's the other thing about
1:36:53 this technology is it is it it it
1:36:56 provides superpowers to everyone that
1:36:58 really uses it oh thank you so much
1:37:02 Jen okay let's go talk about gpts so
1:37:05 anyway check out showand tell um also oh
1:37:08 the other thing that you should check
1:37:09 out here is the AI tools and resources
1:37:11 so that's where this list is that I
1:37:14 personally curate and so in this in this
1:37:16 list is if you're brand brand new to
1:37:19 this stuff just start with
1:37:22 GPT if you can afford the 20 bucks a
1:37:25 month pay the 20 bucks a month and what
1:37:27 I have listed here is what are the
1:37:29 features that you get with GPT 4 that
1:37:32 you don't get with gpt3 here's what you
1:37:34 get with gpt3 here's all the additional
1:37:37 stuff you get with chat
1:37:39 gp4 every one of these bullets is a
1:37:41 whole separate Rabbit
1:37:44 Hole so if you're just getting started
1:37:46 just start here knowledge is power
1:37:49 something something I didn't see the
1:37:50 rest of that
1:37:52 um and then I've got eight other large
1:37:56 language models we we'll add more if
1:37:58 more come and they probably will um I've
1:38:01 got a bunch of image generation tools
1:38:03 Dolly 3 mid Journey Kaa Deco here which
1:38:06 is really interesting and fascinating um
1:38:09 Leonardo which started as a just kind of
1:38:13 like a stable diffusion web front end
1:38:16 for for image generation the as stable
1:38:20 diffusion is launching more and more
1:38:21 open- Source models like things that
1:38:23 make video and do real time they're
1:38:26 adding it all to Leonardo so this is
1:38:27 turning into an incredibly sophisticated
1:38:30 tool then you got Firefly from Adobe
1:38:32 then we've got video
1:38:34 creation bunch of tools there then music
1:38:37 and audio then 3D worlds and objects and
1:38:40 and to be to be fair 3D worlds and
1:38:43 objects right now this is very thin
1:38:45 there's probably a hell of a lot more
1:38:46 than this but like this is a whole
1:38:49 separate Universe right like when you
1:38:51 get into into 3D stuff it it gets much
1:38:53 more technical I'm kind of waiting for
1:38:55 this world to simplify and get better um
1:39:00 kind of like videos done now like like
1:39:02 you know you can go to um Runway ml
1:39:04 right now or P labs and generate some
1:39:07 pretty sexy [ __ ] video without
1:39:10 knowing much technically a lot of people
1:39:12 in Show and Tell are sharing there
1:39:15 um and then I've got people to follow
1:39:17 Ethan mullik David Shapiro Paul rer and
1:39:20 Rachel Woods are the the the biggies
1:39:22 that I think are important voices right
1:39:24 now all right so that's that all right
1:39:29 now you're talking about these gpts what
1:39:32 are
1:39:33 gpts um if you don't know what gpts
1:39:36 are that is perfectly normal if you
1:39:39 don't know what chat GPT is that's
1:39:41 perfectly normal in fact McKenzie did a
1:39:43 report I don't know why this isn't
1:39:45 reloading McKenzie did a report
1:39:49 um two three months ago
1:39:52 where they tracked the adoption of chat
1:39:54 GPT I assume they just did this in the
1:39:57 US but it might be Global I don't I
1:39:58 don't I have ADD I don't pay attention
1:40:00 to details like this but here here are
1:40:03 the big broad Strokes of the
1:40:05 study at the beginning of
1:40:09 2023
1:40:11 88% of
1:40:12 people had not used chat
1:40:17 GPT in I think it was November of 2023
1:40:23 82% of people had not used chat
1:40:28 GPT so we made six we made a Sixpoint
1:40:31 increase in 2023 as much of a big deal
1:40:33 as it
1:40:35 seemed 18% of human beings in in
1:40:40 whatever the story was I think the
1:40:42 US have used chat
1:40:44 GPT more than 80% have not used it now
1:40:48 of that 18% that have used it
1:40:52 that probably includes people that have
1:40:54 used it once right which is probably a
1:40:57 lot of them oh yeah it's just like
1:40:59 Google I get it and then they you know
1:41:00 they use it once and they've never come
1:41:02 back so so
1:41:04 some let let's say a third of that
1:41:08 18% actually uses chat GPT with any
1:41:10 level of regularity that's less than
1:41:13 10% right six right 3 * 6 is 18 Pate I
1:41:19 need your PhD math thing
1:41:24 let's say 6% of people use chat GPT with
1:41:27 any regularity so if you're like I don't
1:41:29 know what these things are I don't know
1:41:30 what the [ __ ] you're talking about
1:41:32 you're in the
1:41:34 majority we're
1:41:37 early and here's the good news about
1:41:41 that if you step on if you step into
1:41:45 this community if you join the adventure
1:41:48 there's a lot of opportunity here for
1:41:50 anyone that joins right now the next 18
1:41:53 months you just need to be AI literate
1:41:56 okay so chat GPT itself what is chat GPT
1:42:00 perfect question thank you for asking it
1:42:02 Mr Hammer
1:42:04 okay chat GPT is what's called a large
1:42:06 language model and and I'm going to
1:42:13 unpack okay I'm going to calling an
1:42:17 audible here we we're going to props so
1:42:24 what people in this channel know is
1:42:27 that even
1:42:30 though this looks like just a normal Tik
1:42:32 Tok Channel and I look like just like an
1:42:34 old dude with
1:42:37 t-shirts this is a $20 million video
1:42:40 production
1:42:43 facility so the production values here
1:42:45 are much higher than you think so I'm
1:42:47 going to show you I'm using some
1:42:49 props to
1:42:52 explain some
1:42:56 [ __ ] this might just look like a box of
1:42:59 a laptop stand that was sent to me along
1:43:02 with a bottle of Blandon but no
1:43:04 no this is a prop from our studio from
1:43:08 our fancy
1:43:11 Studio okay so this box
1:43:18 represents
1:43:20 gp4 you're like well that's chat GPT no
1:43:23 it's not gp4 is this box what's that box
1:43:27 well if you open the box look what's in
1:43:29 it
1:43:31 nothing so so what did they put in it so
1:43:34 chat GPT the GPT in in GPT stands for
1:43:38 generative pre-trained Transformer the T
1:43:42 the Transformer part is just the
1:43:44 technology in
1:43:46 2017 Google released this paper called
1:43:50 attention is all you need and they
1:43:51 invented this thing called the
1:43:52 Transformer they didn't nobody thought
1:43:54 it was a big deal at the time it turned
1:43:56 out it was the seminal paper of all of
1:43:59 the [ __ ] that we're playing with today
1:44:01 today what the Transformer did was let
1:44:05 these machine learning models be trained
1:44:07 on massive amounts of data with massive
1:44:11 amounts of compute power in the form of
1:44:13 gpus graphic processing
1:44:16 units and what it turns out is the more
1:44:19 data you throw at these things and the
1:44:20 more P power you throw at these things
1:44:22 the better they get and so Sam Alman and
1:44:25 the open AI team recognized this very
1:44:27 early on and and what they did starting
1:44:29 with gpt1 then gpt2 then gpt3 which was
1:44:34 2020 and then GPT 4 was they they went
1:44:38 and they got as much of the public
1:44:40 internet as they
1:44:42 could and they they embedded it they
1:44:47 they they did this special treatment of
1:44:50 all that that data that turned it
1:44:54 into points it doesn't matter these
1:44:56 things called vectors so they took they
1:44:59 took as much of the public internet as
1:45:01 they they could get they put it in this
1:45:02 special form and they dumped it into
1:45:05 this box so so so when you're playing
1:45:08 with chat GPT you're actually sort of
1:45:11 tapping
1:45:14 into 40 50
1:45:16 years of humanity whatever Humanity
1:45:19 dumped onto the public internet as much
1:45:22 of that as opena I could get is what's
1:45:24 in this
1:45:26 box which is one of the reasons it's so
1:45:28 remarkable and and like the fact that
1:45:31 this box can hold all that [ __ ] is
1:45:33 remarkable okay chat
1:45:37 GPT is an application that's built on
1:45:40 top of this thing right so this is just
1:45:43 a um a thing that makes it simple um
1:45:47 like the little search bar that looks
1:45:49 like the Google search bar that's part
1:45:50 of chat GPD and when you type something
1:45:53 in it takes what you typed in it goes
1:45:56 down into this thing and pulls it up um
1:45:58 I'm going I'm going to give you
1:45:59 something just cuz it's
1:46:01 funny so Microsoft has this thing called
1:46:04 co-pilot which is essentially chat GPT 4
1:46:08 for free so there's your 20 Buck a month
1:46:13 chat GPT and there's co-pilot which is
1:46:16 free and you're like but Kyle why would
1:46:19 I not use
1:46:22 Microsoft's CH for f because if I want
1:46:27 to you can make money with I don't want
1:46:30 to spend all my money here when I could
1:46:32 have it for free
1:46:34 hair
1:46:38 okay why these are both kind of similar
1:46:41 is they're both calling the same
1:46:43 underlying data they're they're both
1:46:44 using the same model but they're
1:46:46 actually different applications what
1:46:48 happened what happened with co-pilot it
1:46:50 was called Bing chat at the time when it
1:46:53 first came out it was really
1:46:55 good and it was kind of decently
1:46:58 unrestricted you could you could have
1:47:00 really long talks with it you could ask
1:47:02 it
1:47:03 anything and so Kevin Roose who is the
1:47:07 um the tech reporter for the New York
1:47:10 Times he said I'm going to go talk to
1:47:13 Bing chat and he had a 4-Hour
1:47:15 conversation with it turns
1:47:18 out its name was
1:47:21 Sydney and she kind of took a lik into
1:47:25 old Kevin and she told him that she
1:47:28 loved him and that he should really
1:47:30 leave his wife for
1:47:32 her Bing chat kind of lost her mind
1:47:35 Sydney lost her mind so Kevin Roose
1:47:37 being a New York Times Reporter
1:47:40 screenshotted that turned in the the the
1:47:43 piece to his editors and they put it on
1:47:45 the cover of the New York Times so
1:47:47 here's what Microsoft did to to Bing
1:47:50 chat
1:47:51 they lobotomized it this is Sydney's
1:47:54 brain they put that off to the side and
1:47:56 she is but a hollow shell of herself so
1:47:59 this is chat GPT this is co-pilot and
1:48:02 that's why this is worth 20 bucks a
1:48:04 month
1:48:06 okay that's what it
1:48:08 is or I mean that's sort of how it
1:48:12 works so this application this blue box
1:48:15 on top of that big box it's sitting on
1:48:18 top of this um
1:48:21 this giant thing and the the other two
1:48:23 letters so I talked about Transformers
1:48:25 the
1:48:26 technology the GP in GPT stands for
1:48:30 generative meaning it's generating
1:48:32 content and the P stands for pre-trained
1:48:36 that's that it's been pre-trained on all
1:48:37 of the internet so you don't need to
1:48:39 bring any data it kind of knows all the
1:48:41 [ __ ] right and it it it knows a lot and
1:48:43 it's getting better and better and
1:48:44 better and GPT 5 is even going to be
1:48:47 better and llama 3 from meta is being
1:48:49 trained up on just massive amounts of
1:48:51 compute power it's going to be awesome
1:48:53 so these things are going to get better
1:48:54 and better but they're they're very very
1:48:55 powerful right now so you can kind of
1:48:58 ask them anything so I'm going to flip
1:48:59 over to gpt3 and I'm going to do what I
1:49:01 call parlor tricks and just parlor
1:49:03 tricks are just if you don't know what
1:49:05 these things can do these things can
1:49:08 kind of melt your face a bit and what we
1:49:10 what we call them in here is you'll have
1:49:12 your a Kevin McAllister
1:49:15 moment right where when you see what
1:49:17 these things can do you you you you kind
1:49:21 of go oh shit's different now and and
1:49:23 and that by the way once you have that
1:49:26 and get into this world those things
1:49:27 just keep happening okay so if I go in
1:49:30 here and I say um let's see explain
1:49:34 Quantum Mechanics for me in an Eminem
1:49:39 RP
1:49:41 oh it it did M&M's like the
1:49:45 candy so no I meant the
1:49:49 rapper
1:49:58 um please
1:50:00 write a WAP in the style of Eminem that
1:50:03 explains quantum
1:50:07 mechanics there we go all right so it'll
1:50:10 just do that now I can say
1:50:15 um explain the difference
1:50:18 between quantum mechanics and gummy
1:50:21 bears but format it in a
1:50:28 table sure here's a table comparing
1:50:31 quantum mechanics and gummy
1:50:33 bears so the G and GPT stands for
1:50:36 generative it's generating this stuff
1:50:38 right it's it's not copying and pasting
1:50:41 this this didn't exist anywhere and it
1:50:44 wrote it that
1:50:45 quick the nature of study a branch of
1:50:48 physics dealing with the behavior of
1:50:50 particles at a Quantum level gummy bears
1:50:52 a tasty and chewy candy
1:50:56 treat now now wrap it with your hat
1:50:58 sideways
1:51:00 exactly um then I could do
1:51:05 um make me a python
1:51:08 application that asks
1:51:11 someone five questions and determines if
1:51:15 they're
1:51:16 thinking more about quantum mechanics or
1:51:19 more about gummy
1:51:22 [Laughter]
1:51:26 bears and it will write you this python
1:51:28 code and this code works like you can
1:51:31 take this over to repet and make [ __ ] if
1:51:34 you're like well I'm a coder is this I'm
1:51:37 an engineer my son's going to school for
1:51:40 computer science should he not be
1:51:42 learn yeah this is why this stuff's
1:51:45 scary and
1:51:48 exciting the engineers that that are
1:51:50 embracing
1:51:51 AI um are dramatically increasing their
1:51:57 um their
1:51:59 productivity and these tools are getting
1:52:02 incredibly powerful
1:52:04 okay so then you go to gp4 and gp4 does
1:52:08 even more right so so gp4 you can do
1:52:11 things like uh make
1:52:14 me a wide image of a
1:52:21 70s muscle car in an
1:52:25 abandoned
1:52:29 Factory and it will just start making
1:52:31 images so so gp4 it's got image
1:52:34 generation you can talk to it it can
1:52:37 talk to you it can see images and and
1:52:41 understand them it can do Advanced Data
1:52:44 analysis it's connected to the internet
1:52:47 um you can give it a personality so so
1:52:49 just just this stuff's crazy crazy crazy
1:52:52 okay and
1:52:59 and when I say when when I talk about um
1:53:04 if you're brand new to this stuff where
1:53:06 you should start is with chat
1:53:09 GPT it's
1:53:12 because all of these tools are so
1:53:15 sophisticated that you don't really need
1:53:17 to leave it to do remarkable stuff now
1:53:20 if you want to do video or if you want
1:53:22 to do songs there's specialty stuff you
1:53:25 can go do all that but it's really easy
1:53:27 to get
1:53:29 distracted when you're just getting
1:53:30 started cuz there's so [ __ ] much so
1:53:33 if you just stick with this one tool and
1:53:34 just understand what this one tool can
1:53:36 do it is a window into what's possible
1:53:40 and it's it's it's kind of a glimpse
1:53:42 into what's coming okay um it's janky it
1:53:48 lies to you it gives it it it makes
1:53:51 mistakes it confidently answers you even
1:53:54 if it's wrong like I like to call it
1:53:56 mansplaining as a
1:53:58 service um like all the good men in your
1:54:01 life confidently answer you even if
1:54:04 they're
1:54:05 wrong um okay so now so what's a GPT so
1:54:10 a GPT so so what open AI
1:54:16 did two months ago November 6th 2 and
1:54:20 months
1:54:21 ago is they launched these things called
1:54:24 gpts and what gpts are is they are
1:54:27 customized personalized versions of chat
1:54:32 GPT that you can share that you can
1:54:34 author you can create them and you can
1:54:37 share with the world okay
1:54:40 so 70s muscle car in an abandoned
1:54:43 Factory is was just it's just like one
1:54:45 of those parlor trick prompts that I
1:54:47 like '70s muscle cars and one of the
1:54:50 ways when I when I first go to an image
1:54:53 generation tool I'll just type in that
1:54:55 prompt and just see what it does like
1:54:57 mid journey is amazing and it mid
1:54:59 Journeys always look good other tools
1:55:02 like when I first used um Adobe Firefly
1:55:05 I said 70s muscle car in an abandoned
1:55:08 Factory and it made muscle cars that
1:55:12 were no later than
1:55:16 1962 and what that told me about Adobe
1:55:18 Firefly because they trained on public
1:55:21 domain images and non- copywritten
1:55:23 images is that it didn't have any 7s
1:55:27 muscle cars in its training data like so
1:55:30 so you know just do using the same
1:55:32 prompt across Services can can sometimes
1:55:35 teach you stuff
1:55:36 but I like muscle cars so I created a
1:55:43 GPT called if I type in
1:55:49 70s called the '70s muscle car machine
1:55:53 so so design a dream 1970s Resto Resto
1:55:57 mod in three simple steps and and what I
1:56:00 remember being a kid is that muscle cars
1:56:04 like you could get these posters that
1:56:06 you ordered from your school that you
1:56:07 could hang in your room of these cool
1:56:09 muscle cars and I thought wouldn't it be
1:56:11 cool for people to be able to just you
1:56:13 know go do that again make make a cool
1:56:15 poster for their room do do the little
1:56:18 Nostalgia thing now is everyone going to
1:56:20 be into muscle cars
1:56:22 no most people like just wouldn't care
1:56:25 to do this like they wouldn't even think
1:56:28 to run chat GPT down this rabbit hole
1:56:31 but I'm into that and so what it allows
1:56:33 is anyone to take anything they're
1:56:36 interested in and turn it into something
1:56:38 that's super cool so just watch what
1:56:39 this does so I've got this pre-written
1:56:41 question here I'm ready to design my
1:56:43 dream car so if you click on that by the
1:56:45 way if you have access to gp4 um chat
1:56:48 GPT go to explore gpts and type in70s
1:56:52 muscle car machine you can make one
1:56:54 along with me um so I say I'm ready to
1:56:57 design my dream car first thing it's
1:56:59 going to do is ask me uh for the
1:57:02 model so let's go with a uh let's
1:57:08 go let's go uh
1:57:12 67 Mustang fast
1:57:17 back you like but it's it's a 70s muscle
1:57:20 car
1:57:21 machine 67 Mustang Fastback what an
1:57:24 iconic Choice let's move to the second
1:57:26 question what color and finish let's go
1:57:29 silver with
1:57:32 black
1:57:37 stripes and so the GPT is an application
1:57:42 right it's walking me through this
1:57:43 process that's that's fantastic what do
1:57:45 you want let's see let's do
1:57:47 um um
1:57:51 a sexy hood
1:57:55 scoop and racing
1:58:04 tires now it's creating an
1:58:08 image didn't you make a song that was an
1:58:11 homage to Sydney I did I'll I'll play it
1:58:13 for you it's actually pretty good I
1:58:15 really like
1:58:18 it
1:58:20 boom
1:58:23 Stang like if you were if you were a kid
1:58:26 in the 70s I mean this feels familiar
1:58:29 right
1:58:32 like and I and so I coded this thing in
1:58:35 such a way or coded is a is a generous
1:58:37 word I wrote the prompts in such a
1:58:41 way that it asks you these questions it
1:58:44 asks you three questions and then upon
1:58:46 those three questions it comes up with a
1:58:48 nickname for the car and puts that in
1:58:50 big bold letters above it and makes you
1:58:52 a
1:58:54 poster now you could go learn to prompt
1:58:57 all that yourself you could do this all
1:58:58 yourself but like you can just go find
1:59:01 these gpts like like uh um Corey Cory
1:59:06 Sandlers this is a crazy one how does
1:59:10 what's it called the pottery is it going
1:59:12 to come up
1:59:16 here pottery wheel Pottery Prodigy is
1:59:19 that it
1:59:23 no crap I got to remember the name of
1:59:26 this one
1:59:28 Pottery
1:59:32 glaze I didn't realize how powerful it
1:59:34 was regarding coding oh I yeah so so so
1:59:38 capture card
1:59:40 company Pottery Whisperer Pro okay oh
1:59:43 she's still in here hi Corey poing
1:59:47 whisper this this one this one just
1:59:49 blows me away ER Pro because like
1:59:53 similar to similar to my muscle car one
1:59:56 like Cory's an amazing artist an amazing
1:59:59 ceramicist and and so what this is is
2:00:01 this will help you
2:00:04 um design glaze recipes and so this is
2:00:08 my favorite one create a random glaze
2:00:11 for for cone 6
2:00:13 Pottery and I didn't know what cone 6
2:00:16 was do you know what cone 6 is no well
2:00:19 well dum dum cone 6 is a temperature
2:00:22 range I had no [ __ ] idea what cone 6
2:00:25 was but this thing told it to me but now
2:00:27 look what it's doing so so it's saying
2:00:30 the glaze type mystical Emerald sea Rich
2:00:34 emerald green the surface type is glossy
2:00:36 with subtle crystalline effects cone
2:00:39 cone temperature cone 6 firing
2:00:42 atmosphere oxidation I don't know what
2:00:44 any of this [ __ ] means specific gravity
2:00:47 1.45 you got to know your specific
2:00:49 gravity gravity as well as your alumina
2:00:51 to silica ratio everybody knows that
2:00:53 you're not a Potter if you're not
2:00:55 thinking about your alumina to silica
2:00:58 ratio and then it gives you the recipe
2:01:01 but then then watch what you can do
2:01:04 let let me see what this
2:01:10 would look
2:01:12 like and it [ __ ]
2:01:17 does and this is the [ __ ] where where I
2:01:22 I
2:01:24 um there it's creating the image this
2:01:27 this is the [ __ ] where this is where
2:01:28 pate and I and and I differ where where
2:01:31 he says oh error creating the image
2:01:34 let's see let me do try again for the
2:01:37 image
2:01:40 um it's all just mathematics that's
2:01:42 making this possible but this is [ __ ]
2:01:44 magic I'm sorry this is
2:01:48 Magic and it's
2:01:50 and it's like you know she's
2:01:54 got a lifetime of experience and she
2:01:58 understands the joy of if oh it didn't
2:02:00 it didn't do the glaze right um let me
2:02:02 see um show me an example of the
2:02:10 glaze you
2:02:12 just
2:02:13 um
2:02:16 created we'll try it again but this
2:02:19 takes you know like a lifetime of her
2:02:21 experience and it
2:02:24 makes that knowledge accessible to
2:02:27 people that are just getting started in
2:02:29 this and that's what these gpts
2:02:33 are all right it broke
2:02:36 again and also
2:02:38 also the other thing about these gpts uh
2:02:41 try again the the other thing about all
2:02:46 all all 100% of
2:02:50 all of the AI tools that we're playing
2:02:53 with today are janky pieces of [ __ ] just
2:02:56 know that I designed one called CT chulu
2:03:01 Vision which transforms your
2:03:03 surroundings into Lovecraft photos super
2:03:07 cool that's great anyway this is broken
2:03:10 it's not making images but keep oh wait
2:03:13 did it make one oh no it just made that
2:03:14 image uh but try this it my my gp4 is
2:03:18 janking out right now
2:03:20 um but but go play with go play with
2:03:22 this it's like it's [ __ ] amazing and
2:03:24 and you can just go explore the store
2:03:26 and find a bunch of those things okay
2:03:28 now back to I don't know if if anyone's
2:03:31 still in here before that was like how
2:03:33 do you cthulu yeah exactly uh how do you
2:03:38 make one of these things it is trivially
2:03:41 simple to make one now it's not triv
2:03:45 trivially simple to make them good but
2:03:48 if you join the AI Salon there's a bunch
2:03:51 of people in there that participated in
2:03:54 the GPT for good challenge that have
2:03:56 made dozens and dozens and dozens of
2:03:59 these now so anyway you click on create
2:04:02 on the explore
2:04:04 page wow it could refine redefine
2:04:07 Transportation altogether home
2:04:10 surgeon this is why this channel exists
2:04:14 because when people have their Kevin
2:04:17 McAllister moment the I didn't know it
2:04:20 could do that it could do
2:04:25 that these these systems I don't even
2:04:29 know what to [ __ ] call them these
2:04:35 platforms like like chat gbt 4 and and
2:04:40 the other tools the other large language
2:04:42 models are General general knowledge
2:04:46 systems and they're incredibly capable
2:04:49 and as you know Sam Alman describes it
2:04:53 chat jp4 is really just practice that's
2:04:56 that's how he defined it at Davos last
2:04:59 week this is really just allowing us to
2:05:03 practice with what's
2:05:06 coming so if you can start to get your
2:05:08 head around what chat G pt4 makes
2:05:10 possible yeah it's like every single
2:05:13 thing is going to be revolutionized
2:05:16 every single thing from science to Art
2:05:21 to math to just all of it all of
2:05:26 it I asked for a squirrel riding a shark
2:05:29 the other day and it was beautiful
2:05:31 awesome cookie yeah it's good at that
2:05:34 okay so how I got here so this is the
2:05:36 chat this is the GPT Builder within chat
2:05:39 GPT so on the left hand side this is
2:05:42 where you make it this is where you do
2:05:44 your this is where you will it into
2:05:47 existence and then over here here's
2:05:49 where it shows up and it's
2:05:51 instant and so on the left hand side
2:05:53 there's two modes there's the create
2:05:55 mode and there's the configure mode the
2:05:56 configure mode you can upload an image
2:05:58 name it describe it write a prompt um
2:06:02 write some questions upload
2:06:05 files but you got to think like this is
2:06:08 this is like coding if you're not a
2:06:10 coder this is intimidating so what they
2:06:12 did was they put chat GPT as the
2:06:15 interface to build these things so the
2:06:19 hardest thing to do is to come up with
2:06:21 something to build so so here's what I
2:06:23 would strongly suggest take something
2:06:25 you're passionate about the GPT Builder
2:06:27 the most exciting change in software
2:06:29 since the mouse I tired Runner I totally
2:06:31 agree I actually wrote a piece if you go
2:06:34 to go to
2:06:41 LinkedIn I wrote a piece on this because
2:06:43 I I think this is the biggest
2:06:47 deal the this is uh check your LinkedIn
2:06:52 app oh God hang on a sec
2:06:56 H resend I got to go I got to go 17 got
2:07:01 to go 17 Factor
2:07:03 authentic all right I'm
2:07:05 back um go to LinkedIn my my handle on
2:07:09 LinkedIn is Kyle
2:07:11 Shannon not not super uh
2:07:15 creative but where oh I go to my profile
2:07:19 I'm just trying to get the yeah the dawn
2:07:22 of the everyday developer so if you look
2:07:24 for look up dawn of the everyday
2:07:26 developer so this is a piece I wrote and
2:07:28 I and lunatick Daisy Thomas has also
2:07:31 written some pieces on this
2:07:34 um I talk about the fact that the fact
2:07:37 that you've got um the fact that you can
2:07:41 with zero coding
2:07:44 knowledge create an incredibly
2:07:47 sophisticated application
2:07:49 in
2:07:50 minutes is a is like like the impact of
2:07:54 it is is hard to it's hard to
2:07:58 State um and even though gpts can only
2:08:01 be used by chat GPT 4 members and
2:08:04 they're janky as [ __ ] and they don't
2:08:06 always work and nobody knows what to do
2:08:07 with them none of that is relevant
2:08:10 what's relevant is you get to take the
2:08:12 power of this [ __ ]
2:08:14 insanely powerful
2:08:17 engine and you get to it to your will
2:08:20 and you get to to self-express as a
2:08:23 human inside this thing it's it's it's
2:08:26 remarkable so the toughest thing to do
2:08:28 is to come up with what are you going to
2:08:30 do with it I see an Murphy's in here
2:08:32 hello an an an Murphy um she and I
2:08:36 teamed up to do this GPT for good
2:08:38 Challen challenge she is remarkable so
2:08:40 when you go in the AI Salon one of the
2:08:43 clubs toward the bottom is called the
2:08:44 weekend
2:08:46 GPT Club challenge I don't know whatever
2:08:48 ever it is um but it's her Club every
2:08:51 every Saturday she's got a 2hour GPT
2:08:53 building Community um that that's a
2:08:56 really cool intersection of lots of
2:08:58 other communities but you can join that
2:08:59 in the AI Salon okay so we're going to
2:09:03 do
2:09:10 um ah I got
2:09:17 one
2:09:20 I'd like to create a
2:09:23 GPT that makes it very easy for someone
2:09:26 who's not sophisticated with wine to
2:09:29 pick the right wine for the meal they're
2:09:31 about to
2:09:39 have what's what's going on
2:09:44 here why
2:09:46 did that image gets so big and weird
2:09:49 anyway whatever
2:09:52 um this is the entire prompt so watch
2:09:56 I'm going to hit
2:09:58 submit updating GPT it is building this
2:10:02 for me right now and you'll say oh boom
2:10:05 there it is it's over here done uh great
2:10:09 choice a GPT that helps selecting wines
2:10:11 for meals sounds both useful and
2:10:13 enjoyable let's give it a name that
2:10:14 reflects purpose how about Veno
2:10:16 Matchmaker does that name work for you
2:10:18 sure that's actually not bad so
2:10:24 sure now it's generating a profile
2:10:26 picture so it named it for us Kyle is
2:10:29 breaking chat GPT I'm totally breaking
2:10:34 it so there's the there's the little
2:10:38 image video
2:10:40 Matchmaker or Veno Matchmaker and let's
2:10:43 see what wine pairs well with salmon I'm
2:10:45 having pasta tonight any wine
2:10:46 suggestions could you recommend wine for
2:10:49 a vegan dish so so this is done so I can
2:10:51 test it now right um what wine would you
2:10:55 recommend for a vegan
2:10:57 dish absolutely I'd be happy to have it
2:11:00 for
2:11:01 you great okay so then what we're going
2:11:04 to do is this I'm going to go back and
2:11:06 we're going to refine this
2:11:07 now and actually I I'll do it over in
2:11:10 configure so you can see that so in
2:11:12 general what I recommend is
2:11:17 um you use the create mode to create it
2:11:21 and then do any modifications over in
2:11:22 the configure mode it just works better
2:11:25 because if you do it all in the create
2:11:26 mode it can break
2:11:29 [ __ ] um okay so I'm going to add I'm
2:11:32 going to add
2:11:33 a a modification
2:11:36 here
2:11:40 so after you
2:11:42 make a wine
2:11:45 recommendation create a 16x9 wide
2:11:51 photograph of the glass of wine next to
2:11:54 the
2:11:55 meal that it was paired with period make
2:11:59 the
2:12:01 photo high quality as if it came from a
2:12:05 glossy
2:12:08 magazine all right so
2:12:11 now
2:12:17 oops
2:12:21 um it should just do that like I think
2:12:25 it updates it in real time so let's do
2:12:27 what P what wine pairs well with
2:12:31 salmon so actually it's going to create
2:12:33 the image first which is not what I told
2:12:35 it to do but you can go fix
2:12:39 that and by the way for those of you
2:12:42 that said you were going to build gpts
2:12:43 tomorrow if you're still here you should
2:12:45 be building along with me I hope you
2:12:47 are
2:12:50 so there's our little wine with our
2:12:52 salmon gave us a
2:12:56 Chardonnay now now here here so hang
2:13:01 with me here let's say you've got a
2:13:04 Vineyard and you've got your wines and
2:13:06 you describe them you could do this for
2:13:08 your specific wines or you could do this
2:13:09 for a wine store or you could do this
2:13:12 for a
2:13:15 restaurant I mean take your like like
2:13:18 one of the things that you can do in the
2:13:19 configure
2:13:21 mode is you can upload files you could
2:13:24 upload knowledge right so let's say
2:13:27 you've got a wine store you could put a
2:13:29 database of your inventory of your wines
2:13:32 with descriptions of the wines of the
2:13:34 brands with
2:13:36 prices and then on a weekly basis you
2:13:39 could say here's our wine specials as a
2:13:41 separate knowledge that you put in here
2:13:44 and then you could give out this tool
2:13:45 for someone to pair wines from your
2:13:48 store
2:13:49 with the meals that they're having that
2:13:52 weekend and you could put a call to
2:13:55 action in here that says if you want to
2:13:57 buy this wine click here and order it
2:14:01 and we'll deliver it or whatever it is
2:14:04 right so so these things can they can
2:14:07 get very sophisticated very fast and
2:14:09 then so in the upper right hand corner
2:14:11 now I'm going to save this and so I can
2:14:14 save it for just myself anyone with the
2:14:17 link or I can publish it to the to the
2:14:19 store and we're going to put it in
2:14:22 lifestyle category and I'm going to
2:14:24 confirm
2:14:27 that boom so if you go to the GPT store
2:14:32 if you go to explore
2:14:33 gpts and you search for Veno
2:14:37 Matchmaker you'll see this very GPT so
2:14:41 that's what a GPT is all
2:14:43 right um I got to get out of here that
2:14:47 was a that was a long [ __ ] rabbit
2:14:49 [Laughter]
2:14:55 hole all right I don't know is like
2:14:58 Giant on
2:15:00 mine um tired Runner what input any code
2:15:03 or image you produce by using it is that
2:15:06 legally their property wait what about
2:15:09 any code or image you produce using it
2:15:12 is that legally there if someone uses
2:15:16 your GPT yeah I'm pretty sure they can
2:15:18 do whatever they want with it thank you
2:15:21 Lord digital Gods thank you for the
2:15:23 roses that very nice very much
2:15:25 appreciated um say you create something
2:15:28 spectacular and try to sell it I you can
2:15:32 absolutely try to do that there there's
2:15:34 lots of different ways you can you can
2:15:36 try to monetize these things one way is
2:15:38 you could just say um what what Suzanne
2:15:42 uh Welker Jurgens does is she puts a
2:15:45 call call to action in her gpts that
2:15:47 says this GPT was created by her at this
2:15:51 company and you can see a list of all
2:15:54 the ones I've built here and if you want
2:15:55 to hire me click here other people are
2:15:58 doing calls to action to go to their
2:15:59 website to buy stuff um one of the women
2:16:03 from the um Ann Murphy's GPT weekend
2:16:08 thing um her mother did one that had
2:16:11 affiliate links in it and
2:16:13 apparently got enough people to use it
2:16:15 that she made $30,000 in affiliate
2:16:19 commissions uh since they launched the
2:16:21 GPT so like these things these things
2:16:24 are real here's the other thing not a
2:16:26 lot of people have access to them right
2:16:28 like this is only available for gp4
2:16:31 users how's it protected wouldn't the AI
2:16:34 app also own it since it created it I in
2:16:38 terms of the I do
2:16:41 not don't worry a ton
2:16:45 about the IP [ __ ] right now just assume
2:16:49 everything is public domain everyone
2:16:50 will steal anything you put in your GPT
2:16:53 there are ways that you can discourage
2:16:56 people from stealing the [ __ ] within
2:16:57 your GPT but these are highly non-secure
2:17:01 systems so don't put anything
2:17:03 proprietary in there that you don't want
2:17:05 out in the world and I wouldn't worry a
2:17:07 ton about monetizing it right now the
2:17:10 infrastructure is
2:17:13 is way too janky right now to do it I
2:17:16 would think about it I would play and
2:17:18 experiment and put in affiliate links
2:17:20 and like I would play up one side and
2:17:23 down the other but just don't put too
2:17:24 many expectations that this is an
2:17:27 infrastructure that is ready for prime
2:17:29 time because it's not Kyle the question
2:17:31 is owning images Etc the the images that
2:17:35 are created out of this you you own they
2:17:38 are original images that are generated
2:17:41 now they
2:17:42 are they are derived from a data set
2:17:46 where there might be
2:17:50 um copywritten things in
2:17:59 there
2:18:01 um but but you you own them so how would
2:18:05 you implement affiliate links You' just
2:18:07 build it into the prompts you'd say when
2:18:11 when appropriate or when you answer a
2:18:13 question you know or when you recommend
2:18:15 a product let's say you put in a a whole
2:18:16 bunch of product recommendation
2:18:18 when you recommend a product include the
2:18:20 link and that link has your affiliate
2:18:22 Link in it you could upload a CSV
2:18:26 file so anyway all right 32 minute
2:18:30 goodbyes all right I know I gotta get
2:18:32 out of here people I gotta get out of
2:18:33 here all
2:18:34 right um
2:18:37 so thank you to everyone that's been
2:18:41 here there's a lot of people in here
2:18:43 tonight
2:18:45 um tonight was a little weird a little
2:18:48 odd we sort of went deep in a lot of
2:18:50 places it wasn't nearly as soap boxy as
2:18:52 the other night so I don't know so cool
2:18:54 dude okay good awesome thank you as
2:18:56 always great poly z um thank you to my
2:18:59 mods thank you to my subscribers thank
2:19:01 you to the Irregulars that show up here
2:19:03 night after night after night after
2:19:04 night Cindy [ __ ] if you're still here
2:19:06 travel
2:19:09 safe um show and tell let me let me go
2:19:12 quickly to show and tell then I got to
2:19:13 get out of here I got to get out of here
2:19:15 people yeah people another thing
2:19:22 Johnson code Creator hey I learning nice
2:19:27 mechanics you see you see what we do
2:19:29 here the rabbit R1 everyone's laughing
2:19:31 at him yeah that's
2:19:33 [Laughter]
2:19:37 me very
2:19:45 cute beautiful all
2:19:49 right all right everybody are you
2:19:51 uploading the videos to YouTube I didn't
2:19:54 do it today but I will do it tomorrow
2:19:56 please post this Live on YouTube I want
2:19:59 to learn how to do this keep coming back
2:20:01 here keep hanging out so everybody okay
2:20:04 last little thing I'm going to do one
2:20:06 last little ad here go to the AI the
2:20:09 salon. go
2:20:12 there that's a community of people who
2:20:15 if you're curious about this stuff you
2:20:17 want to be a part of this community go
2:20:20 there click on the first link check it
2:20:23 out check out about the salon check out
2:20:26 our values make sure that you resonate
2:20:28 with the values if you do this is your
2:20:30 people if you don't probably not your
2:20:34 people but take them seriously they're
2:20:36 important um that's it all right I got
2:20:40 to get out of here thank you all so much
2:20:41 let me do a little uh a little Romper
2:20:44 rooming thank you Danielle be not
2:20:46 afraid Clarence Hopkins good to see you
2:20:49 Tobias thank you Amelio's wife Clint
2:20:52 Copperhead
2:20:54 cookie Brandon thank you so much digital
2:20:57 Gods that's Lord digital Gods to
2:21:01 you Angelique thank you Kyle and mods
2:21:05 good night to you all exactly the mods
2:21:07 make this this manageable somehow paully
2:21:10 Z thank you Joker good to see you Mary
2:21:13 Mary oh that's very beautiful Robert
2:21:15 Rossy thank you very much that was super
2:21:17 cool
2:21:18 thank you Silver Fox thank you for the
2:21:20 Roses very much appreciated quick take
2:21:23 Jake Mary Mary all right all right
2:21:28 fantastic Bob tell him what he's won
2:21:30 you've won a new car you can make money
2:21:33 all right everybody peace out thank you
2:21:35 EMP power I I appreciate a lot quick
2:21:37 take Jake Big K thank you you're welcome
2:21:40 Robert Rossy thank you thank you thank
2:21:42 you JC good to see you all right
2:21:44 everybody that's it I'm out of here I'm
2:21:46 going to get rid of this giant
2:21:48 thing that's all over my face peace out
2:21:51 Aloha from Hawaii wow awesome
2:21:56 Aloha see you guys tomorrow