Jun 17, 2024 - AI's Future: How Generative Models Are Transforming Creativity

Description

Chapters

Transcript

0:01 [Music]
0:15 take Highway won't
0:18 you
0:21 way your way and my way seem to
0:24 be
0:26 child mama don't understand it she wants
0:30 to know where I've been I have to be
0:34 some kind of natur born fool to want to
0:37 pass that way again you know that I
0:42 [Music]
0:45 feeling
0:49 [Music]
0:53 country long Jesus you
1:02 yeah you're singing good you're singing
1:03 good you got it
1:05 [Music]
1:11 going good evening good people
1:14 [Applause]
1:18 [Music]
1:30 w
1:33 [Music]
1:41 [Applause]
1:41 [Music]
2:15 [Music]
2:25 good evening good people of the
2:29 world Happy
2:31 Monday hope you had a good
2:36 weekend do me a favor share the
2:40 live Dr J shared the live yall got an
2:44 early start we did we were two minutes
2:45 early listen Dr J be like Dr J share the
2:49 live share the live you're like but Kyle
2:51 who do I share the live
2:54 with you share it with people that would
2:57 not want to hear about AI
2:59 [Music]
3:02 oh I would never send it to Jenny send
3:04 it to
3:05 [Music]
3:29 Jenny freom came my way that
3:34 night just like a jet plane in and out
3:39 side I was hauling ass at a million
3:43 miles an hour wondering how hard i'
3:49 hereit when they came into the
3:54 [Music]
3:56 station they said I was bad Beyond
3:59 repair
4:02 but I got no quals with my
4:07 situation say here I
4:13 [Music]
4:14 am
4:16 [Music]
4:18 so say sh sh Shere w't you dare to say
4:25 Shere Shere Shere won't you dare to say
4:30 hey sh sh
4:33 sh
4:36 uhhuh yeah leave a message and your
4:39 number
4:42 please take your time to want to satisfy
4:47 me take all these old fantasies and send
4:51 them care of
4:54 [Music]
4:58 me oh all right good
5:02 people play Shine On by the Cs if I knew
5:05 more than seven songs I absolutely
5:08 [Music]
5:11 [Laughter]
5:14 [Music]
5:18 would all right we only have 21 people
5:20 in here Twi or Tik Tock does not like us
5:23 tonight let's try to get some more folks
5:24 in here yeah the tapy Taps maybe those
5:26 will
5:27 help maybe if we got a different spokes
5:30 model that would help people are
5:33 flicking along like nope nope nope
5:37 just nope nope is that Alec bald when he
5:41 got fat oh no that's
5:48 disappointing we got lots of good stuff
5:50 to talk about tonight
5:52 [Music]
6:01 somebody floated another rumor that open
6:03 AI was going to launch something this
6:05 week I stopped holding my breath for
6:07 open AI about a month
6:10 [Music]
6:14 ago either they're playing
6:16 four-dimensional
6:18 chess where they're struggling to
6:21 release
6:24 things although they did it it seems
6:27 like they did release an updated dolly
6:31 three the notification did not go off oh
6:34 now there's 40 people in
6:37 here what was the
6:39 [Music]
6:43 notification who's knitting image is
6:46 that I forget whose knitting image that
6:50 [Music]
6:56 is oh that I was live it didn't it
6:59 didn't go off that I was Live Well Tick
7:01 Tock are
7:06 [ __ ] you know you go live seven
7:09 nights a week for a
7:11 year and then you main maintain that
7:13 five nights a week for however long it's
7:15 been and Tik tok's like yeah whatever go
7:19 [ __ ]
7:22 yourself
7:26 whatever I need to be a
7:30 young
7:33 voluptuous Instagram Vixen don't
7:36 I I need boobies and beach
7:39 [Laughter]
7:40 [Music]
7:54 scenes you and I keep wait you and I get
7:59 all along
8:00 [Music]
8:02 Sunday morning hair at
8:05 home sky blue and my C is strong it's
8:12 [Music]
8:13 true you can have mine I'm sick of them
8:16 all right yeah ship them over ship me
8:18 the bikinis I'll you know I'll start
8:20 going full
8:21 manini we'll go we'll go manini we'll do
8:25 virtual Beach backgrounds and I'll just
8:27 talk about how I'm manifesting abundance
8:29 we'll get we'll get this we'll get this
8:31 puppy up to thousands we'll have
8:34 thousands in here and they'll come for
8:36 the eye candy and they'll stay for the
8:38 AI education it's going to be
8:43 awesome he he doesn't look like an
8:46 influencer and I I don't think I can
8:48 unsee what he's wearing um I should
8:52 leave but I can't look away
8:54 [Laughter]
9:00 a that's funny my brain hurts I
9:04 think I had an epiphany on llms today do
9:07 tell I on China what was your
9:10 Epiphany I've had a couple recently as
9:12 well so I'm curious to hear what yours
9:14 are Joey no on the maninis
9:19 what I don't know I just I feel
9:23 like you know I feel so not
9:27 objectified you know what you know what
9:28 it is I feel like you respect me for my
9:30 brain and and I just want to be you know
9:36 influencer
9:37 meat maybe if you sold something Kyle
9:40 yeah no
9:41 [ __ ] have you tried the new what what's
9:44 the new 3D printer the new bamboo Labs
9:46 3D printer it's my favorite 3D printer
9:49 of all time I'll tell you what hey I
9:51 just got in my new shipment a Madagascar
9:53 vanilla we buy it by the pound
10:01 yeah I know I'm not selling enough you
10:02 can call in all right yeah hop on I in
10:05 [Music]
10:10 China is there an AI to sketch out my
10:13 product ideas or do I have to Forfeit my
10:16 cop copyright if I spill all the
10:19 beans um that's kind that's a bigger
10:22 interesting question like you're not
10:26 you're not giving up your it's
10:28 just um open AI may use your wait Ian
10:32 China are you here yeah he's here okay
10:34 yeah um you may you may just you know
10:38 they may use your stuff for training but
10:40 they're they're not actively going
10:42 through people's [ __ ] and Mining it for
10:44 ideas what's happening what's your
10:46 what's your so I was I was doing my walk
10:49 today hey so I don't know if it's really
10:51 an epiphany but for me it is anyway so I
10:53 was doing my walk today and I was
10:54 thinking about llms and you know it
10:57 occurred to me that everything that I've
11:00 been sort of trying to understand about
11:01 them has been rooted in natural language
11:03 and how you know the statistical
11:05 probabilities of the way the uh llm
11:08 Works in terms of predicting the next
11:10 word and the next group of letters and
11:12 the next words and phrases and
11:14 everything like that it's all centered
11:15 around language but and you know the way
11:18 it keeps vectors of numbers that that
11:21 really are um abstract representations
11:24 of those predicted next words and so on
11:26 and so forth and then I thought about
11:28 what if you don't store natural language
11:30 in the llm what if you store data that
11:33 represents some motion or some some
11:36 aspect of reality for example the three
11:39 body problem which is very known
11:43 difficult problem to try to predict the
11:44 Motions of three bodies gravitationally
11:47 acting on each other how you could store
11:49 that data in a in a huge I mean much
11:52 bigger than the biggest llm now to help
11:54 predict where the Motions were going to
11:56 be but that's just one example the other
11:58 example is like autonomous driving car
12:00 like a Tesla you know how they threw out
12:02 their old code and they mov towards an
12:04 llm and they're basically training the
12:07 llm on the presentation of what's coming
12:09 down the street and so and then I got
12:11 into the thought of what else can you
12:14 make can you take that's abstract data
12:17 not natural language and build a neural
12:20 net an llm from it to predict and tell
12:23 you about something that's going to
12:25 happen in nature next whatever that
12:27 something is well that's so that
12:30 a lot of what's fascinating about the
12:33 Transformer model is when you embed
12:35 stuff when you create that Vector space
12:38 like words are just one kind of data so
12:41 so how we're getting you know all these
12:43 different kind of models that understand
12:45 language and understand protein foldings
12:48 and things like that is they're
12:49 basically just throwing different data
12:51 sets exactly how you're describing it at
12:53 the Transformer so it's it's not just
12:55 large language models it's large all
12:57 sorts of different kinds of models right
13:00 the music stuff's created so yeah that
13:02 that's it's it's it's something I've
13:04 been thinking about a lot is you know
13:07 what are the data sets that you know
13:09 data Geeks aren't super excited about
13:12 that are that are really obscure and
13:14 valuable that no one's even thrown at
13:16 these things yet right there's there's a
13:18 bunch of stuff that you know people are
13:21 going to be creating these models around
13:23 that we can't even imagine right now um
13:26 so yeah it's huge it's huge that's the
13:29 thing is the Transformer approach this
13:31 approach of creating this Vector space
13:35 and being able to sort of reach into it
13:38 no matter what it's trained on that's
13:39 that's how they're doing all these um
13:42 when they're when they're reading
13:43 people's uh brain scans right right
13:46 images they saw that's how they're doing
13:47 it they're just training it on different
13:48 data not on words so that's exactly
13:51 amazing that's why it kills me when I
13:52 hear these AGI you know um uh you know
13:57 holier than now types talk about oh you
14:00 know uh geni is a is a is an off ramp
14:04 you know it's an off ramp I feel like we
14:07 haven't even scraped the God darn
14:09 surface yet of this stuff and and the
14:11 other example that I was just reading
14:12 about yesterday is there is a uh there's
14:16 a fight between rare earth metals in in
14:18 the world going on right now and some of
14:20 these rare earth metals are extremely
14:23 strategically important so now they're
14:25 using uh gen llms to help predict where
14:29 to find things like lithium and ridium
14:33 and Palladium basically they're creating
14:35 a training a data model on Mineral other
14:38 minerals that exist in other places and
14:41 they're trying to teach it where to find
14:43 these rare earth metals based on what
14:45 else is in the topography which is just
14:48 exactly what I'm talking about it's
14:50 creating a an llm from data not from
14:53 language yep yep yep and that's there's
14:57 lots of stuff out there and a lot of the
14:59 the you know old school AI like like
15:02 that's a lot of what they were doing
15:03 it's just the Transformer model lets you
15:04 do it a massive scale right and it can
15:07 be on anything yeah so that's coming my
15:09 my Epiphany around large language models
15:12 I talked about it a little bit on here I
15:13 think you've been on when I when I've
15:15 talked about it but what's really hit me
15:18 I tried to figure out a way to
15:19 articulate this idea that when you
15:22 interact with chat GPT they call it a
15:24 chatbot but it doesn't feel very botlike
15:26 it feels more human and but you know we
15:28 call it artific IAL intelligence and we
15:30 call it AI Bots and things like that but
15:33 um I saw I saw a tweet of this and and
15:36 the Tweet was AR AI is the collective
15:40 intelligence of humanity and and what
15:43 hit me is why chat PT and other llms
15:47 feel so human is because you're you're
15:50 not interacting with the mechanism the
15:53 mechanism is a calculator but what it
15:55 was trained on is like the output of
15:58 humanity right
16:00 yeah the Corpus the Corpus what you're
16:02 actually interacting with is Humanity
16:04 like the collective int of everyone that
16:07 came before us and that to me was like
16:09 holy [ __ ] that's that's why being good
16:12 with language and being good with people
16:15 like e Ethan mik put this out today that
16:18 those people that are better with people
16:20 are better at generative AI because
16:23 you're essentially interacting with like
16:25 the ghosts of people it's bizarre ABS
16:28 perect sense like that phrase to me
16:30 collective intelligence that's it that's
16:32 what this stuff's all about awesome
16:34 thanks so much Kyle I'm going to let you
16:36 get back thank you so much pleasure
16:38 having you up here this great appreciate
16:39 it
16:41 cheers all right so yeah let me turn on
16:45 my duly camera
16:47 again so bunch of things if you're new
16:51 here welcome my name's Kyle Shannon this
16:53 is the AI learning lab you know what we
16:55 do
16:56 here we talk about trips to Barbados and
16:59 manifesting our
17:01 abundance let me go get my
17:07 bikini um God I love it here does that
17:10 make me a nerd yes yes that makes
17:16 you I'll tell you this channel is just
17:19 we're gonna take it to the next level as
17:20 soon as I get twerking lessons I just
17:22 got to find myself like a
17:24 20-some twerking influencer and then
17:27 I'll be like well we'll put on that song
17:30 of like those Irish kids you know
17:34 something I doubt
17:36 it I doubt and then I'll do the dance
17:39 and then I'll start wearing maninis by
17:42 the pool like it's gonna be we're gonna
17:44 hook that [ __ ]
17:46 up the guy from open AI said that they
17:48 aren't just stringing words together
17:50 anymore yeah I know and Jeffrey Hinton
17:53 said that as well Jeffrey hinton's been
17:54 on a couple of new interviews lately
17:56 he's the Godfather of AI
17:59 and he's saying that um the models used
18:03 to PR predict the next word or next
18:05 token but they're now actually
18:07 understanding the concepts so there's
18:09 there's definitely a shift in
18:11 the Jeffrey Hinton was also saying we
18:14 don't understand how they work but we
18:16 know that they they're starting to
18:18 understand things um I asked chat PT
18:22 give me the summary of collective
18:24 intelligence of all of humanity into
18:26 categories oh fascinating and what did
18:29 it say that's awesome that's awesome
18:31 like what are the categories of all of
18:33 collective
18:37 intelligence there's
18:39 food there's food there's
18:45 shelter cuno I have an abundance now in
18:48 my bikini I need OIC yeah no
18:53 [ __ ] actually you know what if uh if I
18:57 really wanted to make money i' I'd get
18:59 on the uh the OIC uh influencers program
19:05 get some get get some commissions and
19:07 lose some of this uh middle-aged
19:12 manfat Hinton said they gave them the
19:15 algorithm and they didn't understand
19:16 anything after that yeah I know I know
19:19 it's [ __ ]
19:20 Bonkers well and the the thing there so
19:23 there's a couple of things what I want
19:24 to talk about tonight what I want to
19:26 show is some of the new stuff that came
19:29 came out from Runway ml they just
19:31 announced their gen 3 model and it looks
19:35 like the [ __ ] I've seen from
19:37 Runway gen 3 to
19:42 me looks better than Sora like Sora
19:45 while it was really cool it still had
19:48 glitchy like physics weirdness to
19:51 it the stuff from Runway gen 3 which
19:55 which is apparently going to be here in
19:56 days not [ __ ] weeks or months or
19:59 years like
20:00 Sora and then we had Luma labs and I
20:03 read I I learned something about Luma
20:05 Labs that's absolutely [ __ ]
20:08 mind-blowing and then um and then deep
20:11 mind today so deep mind if you don't
20:14 know them they're the they're the AI arm
20:16 of uh of Google um I heard an
20:19 interesting thing that Dennis aabus who
20:21 runs Deep Mind he's the co- CEO I guess
20:24 of of is he the co-ceo or he's I don't
20:27 know he's up there uh
20:30 for Google that he might become the CEO
20:33 of Google that would be interesting but
20:36 anyway they announced today progress
20:40 on video to
20:43 audio and you're like wait what so you
20:46 know how all these video tools they
20:48 generate silent movies well deep Minds
20:51 tool takes those silent movies and
20:53 interprets them with a prompt and does
20:56 all the sound effects for that clip it's
20:58 like
21:00 [ __ ] so I figured we go through all that
21:03 stuff because there's a there's a lot to
21:05 unpack there
21:07 um there's a lot to unpack there so so
21:10 we can do that so anyway if anyone has
21:11 any questions or any other thoughts I
21:13 asked meta AI sunglasses to evaluate my
21:15 golf stance they told me to straighten
21:18 up instant Improvement no
21:21 [ __ ] that's really cool Brandon how are
21:23 you liking the The Meta AI sunglasses
21:30 and you can if you want you can hop up
21:31 on stage and chat about
21:34 it got a meta Quest 2 headset for Papa's
21:37 day still a little janky but fun
21:42 interesting I have so many questions
21:44 about so many things tonight I know well
21:47 so feel free pop the questions in and
21:49 I'll do my best to answer them but I
21:51 just want to show some of the new stuff
22:00 Brandon what's happening hey how's it
22:02 going good so so how long ago did you
22:05 get those things uh I've had them for
22:07 about two weeks now they're just the
22:08 sunglasses though so I have to wear them
22:10 outside okay and what's the what's the
22:13 verdict two weeks in are they are they
22:16 have they have they made have they made
22:18 the Shelf yet or actually using them I
22:21 am I am actually using them um they are
22:24 I'm using them more for the camera it is
22:27 really nice especially when I'm playing
22:28 out with my kids in the backyard to be
22:30 able to just either say take a picture
22:33 or tap the shutter and the kids love it
22:36 they're like Dad take a picture of take
22:38 a picture of us with your uh glasses um
22:42 really high quality picture and video
22:44 actually I'm really impressed with the
22:45 quality of the camera um it does not
22:49 hold up from a a large language model
22:52 standpoint quite yet um but I assume
22:55 they'll be
22:56 making it is it is but it's like I asked
22:59 it when I was out in the golf course I
23:01 was really giving it it's trying to put
23:02 it through its Paces for Father's Day
23:04 Happy Father's Day by the way um that
23:07 I'd ask it like is this Poison Ivy and
23:09 it's like I can't talk to you about
23:11 medical stuff like but that's not
23:14 MediCal okay uh and and a couple of
23:17 times I'm like what what am I looking at
23:19 it's like please reconnect your meta 3
23:21 to the or meta AI to the internet like
23:25 I'm standing out in the middle of a golf
23:26 course I have five bars but other than
23:29 that when it works it it is really nice
23:32 like you know my kids are like hey what
23:33 is that plant and I ask what the plant
23:35 is like that's basil I'm like oh that's
23:37 cool and so how does it work so so you
23:40 just talk and you go hey meta or
23:41 whatever the [ __ ] you say and then and
23:43 then it's got Speakers by your ears how
23:45 like how does it actually work yeah it's
23:47 like bone conducting headphones or
23:49 whatever so it's over the over the ear
23:52 so I can hear it and people standing
23:54 next to me really can't hear it that
23:55 it's I wish it actually had like an
23:57 output like I could turn it on speaker
24:00 mode because a lot of times I'm demoing
24:01 it and people ask me to ask it a
24:03 question and I have to wait for it to
24:05 answer and then read back what it said
24:07 because they can't hear it but um it is
24:10 really nice I'm looking forward to using
24:12 it this summer when we're out and about
24:13 and it is really nice like hey what's
24:16 the forecast you know just the random
24:18 questions that come to your mind where
24:20 you would normally whip out your phone
24:21 unlock your phone go to Google type it
24:24 in you can just ask and it's just there
24:28 so yeah I like them they're I you know I
24:30 mean they were worth the 250 bucks or
24:32 whatever they're charging for them
24:34 now I wouldn't wear them as a regular
24:37 glasses yet though they're they are
24:40 um they are not uh they're a little bit
24:43 too nerdy for to there's because they're
24:45 really thick frames the sunglass Shades
24:48 mask it really well but if you're
24:49 wearing them like normal I look like
24:52 that emoji the nerd Emoji it's just
24:54 really thick yeah they're really thick
24:57 yeah yeah yeah yeah all this stuff's
24:59 gonna get gonna get uh smaller but
25:01 anyway all right cool awesome well
25:03 thanks for the review and by the way so
25:05 this this book right here M Mr Mr
25:09 Brandon tid made this so one night we
25:11 were doing I forget what I we were just
25:12 doing something in DOI and we did like
25:14 do an alphabet and then you got inspired
25:16 by that so this is my first journey in
25:19 the world of tomorrow and it's an
25:20 alphabet book for all sorts of different
25:23 geeky terms so and with with little
25:27 pictures that were made in DOI
25:30 so y that was your custom GPT that did
25:33 that authored that book and I just
25:35 copied and pasted it into a format for
25:38 Amazon Kindle Direct publishing and then
25:41 my kids love it that's awesome man so
25:45 cool all right I don't even remember
25:46 what was the name of that GPT do you
25:48 remember I don't remember I don't
25:51 either but I should also talk about Sin
25:54 City sipper all right Brandon thank good
25:56 talking to you yeah take care cheers
25:59 all right great book yeah no it's really
26:01 cool
26:06 [Music]
26:08 um all right there's that um I updated
26:13 Sin City sipper so let's let's flip over
26:15 let's go look at some [ __ ] will analyze
26:17 the curvature the putting green and
26:19 offer advice I'll tell you what you know
26:21 Tom with with the liar stuff
26:24 um with the lar cameras like those those
26:28 glass say they're going to get really
26:29 sophisticated where they'll probably
26:31 have whole arrays of you know sensors
26:33 that are beaming [ __ ] out so yeah
26:37 they're probably going to be able to
26:39 generate 3D actually you know you could
26:42 probably play a round of golf and then
26:44 not only will it analyze your swing but
26:46 you can watch the replay of your ball
26:48 you know going out of bounds you know
26:50 with all the curvatures it'll it'll do
26:52 gaui and splatting and recreate the golf
26:55 course for for you in
26:57 3D wait this [ __ ] shit's about to get
26:59 off the [ __ ] hook I'm telling you
27:02 just so like tonight's tonight should be
27:04 fun because there's a lot of there's a
27:06 lot of stuff to look at but the first
27:08 thing we're going to look at is is me
27:11 talking about Story
27:13 vine so what is Story vine and how does
27:15 it work the video that you're
27:17 watching I was talking about Story vine
27:19 all damn day all
27:21 right let's get rid of that okay let's
27:25 see let me go
27:30 [Music]
27:34 let me go to chat
27:35 JP oh I'm already on Sin City sipper so
27:40 if you're on chat
27:44 je and you've got access to either Plus
27:48 or free go to explore gpts and search
27:52 for Sin City
27:54 sipper and this is the very first
27:58 Contour mapping of the golf course yeah
28:00 that'll be coming that's it that'll be
28:02 coming if it's not already here I would
28:04 imagine that that Contour [ __ ] you can
28:06 do with the ldar cameras there's
28:08 probably a specific app out for that
28:10 right now on on the
28:12 iPhones um this was the very first
28:15 custom GPT I made if you don't know what
28:17 a custom GPT is it's basically just chat
28:19 GPT that does something special and what
28:21 this one does is it will invent drinks
28:25 for you so um I just just the reason I'm
28:29 showing it is I just updated it because
28:32 um open AI shifted from gp4 to gp4 Omni
28:39 for the engine for all these gpts and
28:41 they don't they didn't work the same
28:43 like this stopped rendering HTML that I
28:45 had in the code and it uh it wasn't
28:48 always making images it was kind of
28:49 [ __ ] up so I went in and I fixed some
28:52 stuff and then I made I made it um I I
28:56 kind of updated it so let's say we'll
28:57 say
29:01 um invent a
29:05 festive Burbon based drink for the
29:08 upcoming 4th of July
29:15 holiday and so this is called The Star
29:17 Spangled
29:19 sipper let's see what it is 2 ounces of
29:22 bourbon good an oun of lemon
29:25 juice okay an ounce of simple syrup half
29:29 an ounce of blue ciraa two dashes of
29:31 bitters fresh blueberries
29:34 interesting lemon twist for garnish
29:37 ice and then what it does oops I got to
29:41 turn off my voice to text because [ __ ]
29:44 happens is then it makes an
29:48 image of the drink and it now creates a
29:51 custom coaster that gives you the name
29:54 of the drink and it says by Sin City
29:56 sipper how [ __ ] cool is that
30:00 and that drink actually looks really
30:02 good although we got a we got a a star
30:05 in it we got to get that thing out of
30:06 there curo H yeah that's interesting
30:10 isn't
30:11 it are the custom gpts in po open to all
30:15 users yes they are isn't yeah isn't the
30:18 coaster cool Source camp and it says by
30:21 City sipper so Source Camp when you're
30:23 demoing this to restaurant
30:26 people um
30:28 like this is easily modifiable for a for
30:32 like for a
30:33 bar so imagine like this was Source Camp
30:36 had this idea she was she was showing
30:38 chat GP or showing Sin City sipper to a
30:41 restaurant tour and said you could
30:43 create like an iPad where people while
30:46 they're waiting for their tables could
30:47 design their own drink based
30:50 on the booze you have the most of so
30:53 here's a cool thing you could do with if
30:55 you don't know chat gp4
30:58 has
31:00 Vision so for example you could take
31:04 this was Source Camp told me she did
31:05 this this blew my mind you can take a
31:08 picture of the bottles on the bar and
31:11 have it determine which ones are at what
31:13 percentage of emptiness so so you know
31:16 which one to you know to
31:20 reorder you could similarly say take a
31:22 picture of the bar and then say give me
31:25 drinks that are only based on what I
31:28 have have in the bar right so it's
31:30 always using the booze you've got
31:33 available um and then yeah you could
31:35 make it so so that the image it
31:37 generates has the name of your company
31:39 on it something like that crazy right
31:43 just have the star change to a thin
31:45 slice of star fruit yeah there you go
31:47 that's perfect all right I have it
31:50 making drinks with booze we need to get
31:52 rid of at the restaurant there you go
31:54 Silver Fox that's it and and that's that
31:58 that um that use case um someone shared
32:02 with me I forget who it was they did a
32:04 Super Bowl part party this year and they
32:06 took a picture of their bar and they
32:09 said come up with 50 Super Bowl drinks
32:12 and you know present the menu so that we
32:14 give it to people and then create the
32:16 recipes for the bartender so they had 50
32:18 custom cocktails made by Sin City
32:22 sipper and uh and it used only the booze
32:25 they had available I think that's really
32:27 slick all right is isn't that the
32:28 purpose of a special to get rid of the
32:30 Overstock well it's still good there you
32:32 go yep
32:34 exactly is that an anast star no that's
32:37 just
32:39 well it should be but it's just it's
32:43 just confetti it's the same as the [ __ ]
32:45 on the on the
32:47 ground but Isn't that
32:49 cool and so now it correctly says this
32:53 GPT was created by Kyle Shannon
32:54 co-founder of the AI Salon Creator and
32:56 host of the AI learning lab on Tik Tok
32:58 so it's a little ad for you know for the
33:00 community and and this and that so
33:02 anyway go use Sin City Seer all right
33:06 enough about me let's talk about
33:08 [Laughter]
33:14 me all
33:16 right what's that oh yeah that's that
33:19 okay okay
33:22 so let's do some history here
33:28 d
34:00 oh that's cool Lord digital Gods what
34:04 did you
34:08 make oh that's pretty
34:11 slick except you know it would be much
34:14 better if I actually went in and put
34:16 titles in all these
34:18 things when was the last time I actually
34:20 did a
34:21 title
34:23 December November 30th is the last
34:26 YouTube video I actually have
34:29 um a description
34:31 of we'll get that going Brandon and I
34:34 were working on that for a bit I think I
34:36 dropped the ball on that all right let's
34:38 talk about you what yeah exactly enough
34:40 about me what do you think about me um
34:44 okay
34:46 so for for many
34:49 moons
34:51 um Runway ML and Pika were the big
34:54 texted video things and then Sora came
34:56 out and Sora was announced does this say
34:59 when it was
35:04 announced read the technical report
35:07 February
35:08 15th so many many months
35:13 ago open AI comes out with here's the
35:16 this is the
35:18 famous famous
35:21 woman famous Japanese woman walking down
35:24 you know Tokyo Tokyo streets in the rain
35:28 59 second
35:30 video and and we still haven't [ __ ]
35:33 like this thing still hasn't
35:35 launched so February March April May
35:41 June 15th so almost exactly four months
35:45 and we still don't have
35:48 anything
35:51 then
35:54 Google announced Vo
35:58 our most capable generative video
36:05 model
36:08 and no word on when it's coming
36:14 out and
36:17 then last week was it only last week I
36:21 think it was last week last
36:24 week this Sora killer called Luma Labs
36:30 showed up on the
36:31 scene and so like here's a video this is
36:34 a video I made I I went and got a
36:39 um just a picture off the Internet this
36:41 was of The Shining like a Lego kind of
36:45 thing and so this is an image to video
36:48 with Luma Labs it's called Dream Maker
36:51 and it's Luma
36:55 labs. and so like holy [ __ ] like out of
36:58 nowhere this Luma lab showed up and you
37:01 know this might not be Sora level good
37:05 but it's pretty [ __ ] good and you
37:07 know if you if you look at some of the
37:09 videos like this is a video I did of
37:11 Alvin ay dancers that just morph into
37:15 monstrous monstrous
37:18 [ __ ] aberration of nature uh so it
37:22 still got all the sort of morphing and
37:24 weird ass hallucination stuff but like
37:26 this was a picture I did of I did a
37:28 mockup of a of the musical that I'm
37:31 working on on that little picture and
37:34 then it just made all this New York City
37:37 it it panned down to the street and then
37:39 this whole part of the scene is
37:40 completely
37:43 invented um so that's really impressive
37:47 and
37:48 then I read this no this
37:53 no where the [ __ ] is it
37:58 did I put it
38:01 here hang
38:05 on
38:07 Luma Oh
38:10 Know by the way this tool I think I
38:14 learned this from Riley Riley what's his
38:16 name um
38:21 Luma yeah this
38:25 one okay
38:30 so this dude research scientist working
38:33 on video generation at Luma laabs previe
38:36 previously video poet at Google
38:43 AI so Benjamin decracker deer says I
38:48 realiz Luma dream machine is better
38:50 anything Google has shipped and Dan said
38:53 now you know why I left Google to join
38:55 Luma I was in the team that developed Vo
38:58 early on but knew it would never be
39:00 shipped to the masses for quite a long
39:02 time same as Sora not until a company
39:05 like Luma forces their hand he quit
39:10 Google went to Luma they [ __ ]
39:13 launched Luma in three
39:16 months three
39:18 months three months
39:23 so um we have a model that's doing this
39:27 kind of [ __ ] that was launched in 3
39:30 months so so we're now starting to see
39:33 like little whipper snapper of
39:35 companies you know sort of outclass and
39:37 outpace these big research companies now
39:41 is this stuff going to be as safe is it
39:43 going to be as um coherent is it going
39:45 to be I I probably not like this is
39:48 probably not super um
39:53 responsible but I think that's
39:55 independent of the fact that this stuff
39:57 is accelerating is starting to happen
40:00 faster
40:01 [Music]
40:06 um yeah so that's the next thing coming
40:09 okay
40:10 but what I wanted to show on the Luma
40:16 stuff
40:17 did I lost all my hang on let me go back
40:20 to this
40:23 [Music]
40:25 thing okay
40:29 okay so Luma came out a week ago after
40:32 three months of nothing they just show
40:34 up on the scene they've already added
40:36 the ability to extend the video so it'll
40:40 make like a 7-second video then you can
40:42 extend it to 15 and then you can extend
40:44 it to 22 and then you can extend it to
40:46 28 right so you can just keep extending
40:50 it so that's
40:52 cool and then they
40:59 where is it yeah this
41:04 thing
41:12 oops so coming soon to dream
41:17 machine um here's a sneak peek of
41:21 something powerful editability and
41:23 intuitive
41:25 controls so let me make this video full
41:28 screen and then let's we'll stop and
41:32 start okay so they're doing multiple
41:35 aspect ratios I think they all do
41:44 that okay so let's see share more
41:49 related ideas so it's it says dreaming
41:53 some ideas about a young Apprentice
41:55 discovering magic so rather than
41:57 creating one video it's creating three
42:00 so that's
42:04 cool explore Concepts look at all these
42:07 new things there's a little swipey
42:09 finger thing there's a starry thing
42:12 there's a other thing there's a
42:14 paintbrush thing and there's a whatever
42:15 the [ __ ] he's clicking on here
42:17 [Laughter]
42:21 [Music]
42:23 thing wait
42:28 so so he clicks the button I guess this
42:30 was the magic button he clicks the
42:32 button and they're doing
42:36 segmentation Young Apprentice is here
42:38 bluish tint is here afternoon is here
42:40 Forest is here magical book is here and
42:43 then they're they're changing that
42:52 stuff look at this he's he's over
42:56 changing the background
42:59 and the The Apprentice is staying the
43:01 [Music]
43:06 same precise
43:18 edits
43:20 so so that's
43:22 Luma so that's the thing that's like a
43:25 week old that the guy [ __ ] said [ __ ]
43:28 it I'm leaving Google I'm going to go
43:30 make some [ __ ] that can actually launch
43:32 3 months later they launched this thing
43:35 now was it in development before he got
43:38 there I don't know possibly probably
43:41 which platform is this this is Luma
43:43 lab's dream dream
43:46 machine then
43:52 today Runway I can't keep up anymore
43:56 lunatick this is [ __ ]
43:58 Bonkers because that's just that's just
44:02 that's just Luma Labs that's just dream
44:04 dream
44:05 machine now Runway which we know and
44:07 love today they announced gen 3 Alpha
44:11 and it's supposed to be launching in a
44:13 few
44:15 days so none of this Sora will show you
44:18 the coolest thing ever and then maybe
44:21 Ashton K Kutcher gets to play with it
44:24 because he's [ __ ] Ashton Kutcher what
44:27 have I been
44:29 punked how about the AI learning
44:33 lab I've got nearly 35,000 followers
44:36 like what's Aston kusher got like
44:40 60,000
44:45 [ __ ] so anyway they're never [ __ ]
44:48 launching that thing so this thing is
44:51 supposed to come out in a few days so
44:52 we' so we've got Luma sort of goes from
44:55 nothing to Something in three months and
44:57 and is already iterating features
44:59 rolling out features you know they are
45:02 they just added extend 3 days after they
45:04 launched and now they've got this this
45:06 editing coming that's got this layered
45:07 ass editing and then you've got gen 3
45:10 Alpha and and again this should be here
45:12 in a few days okay so let's go look at
45:14 some of these videos
45:16 because every one of the Sora videos
45:19 that I saw when I watched it
45:22 slowly I was like eh it's it's like 80%
45:27 right like there was always some weird
45:29 physics [ __ ] up in it it was good and
45:32 the coherence was really impressive but
45:34 it was like
45:37 80% these examples that I've seen from
45:40 gen 3 Alpha I like I
45:43 don't I don't see anything wrong with
45:48 these I just
45:50 don't oh the other thing six million
45:53 what's six million
45:57 I don't know what six million is in
45:58 reference
46:01 to Luma is two years old by the way okay
46:04 that makes sense I I'm like they
46:06 couldn't have made it in three months
46:08 from the time that guy got there it was
46:10 three months till they launched
46:11 something great Ashton's followers oh
46:14 six million oh so what so so Ashton's
46:17 got a few followers Kutcher Kutcher six
46:20 million listen listen
46:22 Ashton you think just CU you're like
46:25 like famous and B been in movies and
46:28 like we're married to Demi Moore and
46:31 like have 6 million followers like
46:32 somehow you're more deserving of access
46:36 to Sora than me
46:39 fine fine be that way all
46:43 right so there's that IM there there's
46:45 that video there's this this is a little
46:48 weird and wonky in
46:50 the some of like physics around the
46:53 knees but like the coherence of the
46:55 stores going by
46:58 are pretty
47:00 crazy it's pretty good his face isn't
47:02 morphing and moving weird his hands
47:04 aren't like turning into [ __ ]
47:07 octopi are we looking at Runway yeah
47:10 this is Runway gen 3 Mary Mary to me
47:14 more oh yeah we're on Runway
47:18 now six million followers less than the
47:21 AI learning lab exactly this the 35,000
47:25 that are here is it's quality not
47:28 quantity I could have 6 million
47:30 followers
47:31 easy that's that's I reserve the right
47:34 to wear a
47:35 manini we will
47:38 have dozens of followers leave on that
47:42 first
47:44 day please do not denigrate yourself
47:48 comparing yourself to Mr Kutcher I
47:50 actually think Mr Kutcher is he's very
47:53 smart and he's he's got a a really good
47:55 sense of humor I he's one of those guys
47:58 that you really want to hate but he's
48:00 actually kind of a nice guy and really
48:01 [ __ ] smart so I I I actually
48:03 appreciate uh Mr cucha um so anyway
48:08 that's that video this one is stunning
48:11 The Prompt here is fpv flying through a
48:15 colorful Coral line streets of an
48:17 underwater Suburban neighborhood and it
48:20 like starts out there's like a Toyota
48:22 Prius in the corner and like the the
48:24 corals grown up around the houses like
48:26 bushes and trees it's like a little
48:28 established and look at the little
48:32 road and look at the like the bubbles
48:35 like again I don't see any major physics
48:38 fuckups here like this
48:41 seems pretty [ __ ] good the turtle's
48:44 weird that fish is weird so there's
48:47 there's little things but like I
48:50 wouldn't notice that this one's a little
48:53 odd the the shape of the balloon but
48:55 there's a balloon floating down a street
48:57 in some European
49:00 alley is this text to video or do we
49:03 need to start with an image I think it
49:04 does both but I'm not sure um let me
49:09 see oh by the
49:11 way
49:13 um gen 3 Alpha is the first the first of
49:18 an upcoming series of models trained by
49:20 runway on a new infrastructure built for
49:23 large scale multimodal training
49:25 multimodal training m multimodal
49:27 training it's not going to be just
49:30 video it is a major Improvement in
49:33 Fidelity consistency and motion over Gen
49:36 2 and to step towards building General
49:38 World models this was the thing Sora was
49:41 talking
49:43 about Sora was talking about we're not
49:46 making video we understand and are
49:48 simulating the physical world Runway is
49:52 building General World models moving
49:55 toward multimodal training
49:58 so this thing's going to do sound this
50:01 thing's probably going to do 3D
50:03 worlds I wouldn't be at all surprised if
50:06 it does language and if it does language
50:09 it probably does programming and if it
50:10 does programming it probably does game
50:13 mechanics and
50:17 simulations like just if you think about
50:20 what they're thinking
50:22 about where everyone's headed is systems
50:26 that understand the world understand and
50:29 cons simulate the
50:30 world the output is video but what it
50:34 actually understands this goes back to
50:36 what ion chy was talking about that wait
50:38 you throw different data at these things
50:40 they get good at crazy ass different
50:43 [ __ ] okay so let's go look at some more
50:48 examples
50:49 um again as it's zooming in there was a
50:53 little weirdness
50:55 there but that was was pretty
51:00 good this thing is very very coherent
51:03 like the people in the background are
51:04 staying pretty consistent no one just
51:11 materializes that balloon floating
51:14 pretty good that dandelion doesn't
51:16 really look like a dandelion but it's
51:18 not bad there's a Tokyo train running
51:21 through a Tokyo neighborhood
51:27 fine grained temporal control and again
51:30 so if you're new to if if you're just
51:34 joining this is Runway ml's gen 3
51:37 preview which is apparently coming out
51:39 in a few
51:41 days the whole we're living in a
51:43 simulation argument is starting to make
51:45 more sense Brandon you know what's funny
51:48 a year
51:49 ago like people have been talking about
51:52 we're living in a simulation for the
51:53 past [ __ ] decade like the evidence
51:56 points to and like physicists like just
51:59 some smart people are like it's possible
52:00 we're living in a simulation like what
52:02 the [ __ ] are you talking about like we
52:04 would know it's a
52:06 game and now I'm looking at this going
52:09 would
52:10 we would
52:12 we I was just tinkering with that too I
52:16 know [ __ ] Bonkers it's [ __ ]
52:18 Bonkers we still haven't learned our
52:20 lesson of taking press
52:22 releases as the truth have we well but
52:25 what what I said Pate was uh
52:28 apparently this is coming out in in a
52:30 few days so if by the end of the week
52:33 today's Monday a few days says to me
52:35 that we will have this by Friday if we
52:38 don't have this by Friday I will put it
52:40 in the [ __ ] pile of Sora uh where you
52:44 know we can make fun of it four months
52:46 from now for still not existing but
52:48 anyway fine grain temporal control it's
52:51 been highly trained with highly
52:53 descriptive temporal dense captions
52:57 enabling imaginative Transitions and
53:00 precise key framing of elements in a
53:03 scene don't quite know what that means
53:05 but it sounds like they can do some
53:07 pretty sophisticated isolation of
53:13 elements this is really similar to the
53:16 um the Tokyo train video that Sora
53:21 did that's really interesting with the
53:23 little flashlight on the walls
53:33 dude on a mountain mountain isn't
53:39 morphing yellow sky is
53:44 weird but
53:50 cool water's looking decent it's a
53:53 little weird in here but decent
53:58 again like I'm looking at this stuff
54:01 pretty hard to try to see is there
54:03 anything here that's like an obvious
54:05 giveaway if you didn't know you were
54:07 looking if you weren't looking for
54:10 weirdness there's nothing here that
54:13 jumps out to
54:19 me these tunnel tracking shots are
54:23 always they kind of invite you to find
54:26 the weird ird Ness in them the the fpv
54:29 Drone drone
54:38 things actually this one's actually
54:40 quite
54:42 impressive because normally with um with
54:46 the video models like the pen would
54:48 start out looking like a pen and the
54:49 fingers would start out looking like
54:50 fingers and then the pen would go Z and
54:53 you know it would turn into a finger or
54:55 it would like be you know intersecting
54:57 the finger and then the other fingers
54:59 would be doing this look at look at the
55:01 hand
55:03 here like that's a hand drawing a line
55:07 little weirdness in the lines right
55:09 there
55:17 but
55:20 crazy from the shelter of My Mind
55:23 through the window of my eyes Ryman
55:25 Simon
55:28 some bugs this is another Sora shot
55:31 knockoff clearly these these guys are
55:34 are uh [ __ ] with the Sora
55:39 team like that's a that's a pretty
55:42 impressive shot start zoomed in on bugs
55:45 pan out to the Suburban
55:47 neighborhood none of the houses are
55:48 moving the bugs didn't turn into three
55:50 bugs or five bugs or 10 bugs
55:58 this is tsunami in the
56:01 streets which water and waves are always
56:05 weird there's what's interesting about
56:07 this that moment right there water just
56:10 materializes is weird and right there
56:12 the water had weird structure to it but
56:15 like this stuff and that you could
56:18 absolutely use you could there's there's
56:20 usable pieces of this that you could use
56:22 even though some of it's weird
56:26 this is an impressive shot I think this
56:28 is where we started all
56:30 right so that's that photorealistic
56:33 humans excels at generating expressive
56:36 human characters with a wide range of
56:43 actions wide range of actions gestures
56:46 emotions unlocking new storytelling
56:55 opportunities no weird weird body
56:58 proportions well maybe a little weird
57:01 leg
57:02 there but generally pretty
57:07 impressive what does this mean
57:08 temporarily dense captions um I
57:12 definitely know what that means but I'm
57:14 going to turn it over to Pate to answer
57:15 that because um densely what was it
57:19 densely accurate captions means they
57:23 they have act they're they're not sparse
57:28 uh Pate why don't you why don't I could
57:30 easily explain it but I think you do
57:32 better with Concepts like dense uh
57:39 captions I don't [ __ ] know um lots of
57:42 data tags and annotations throughout the
57:44 videos for training oh for training on
57:47 oh okay cool interesting oh that's
57:50 actually what it means so he did answer
57:52 it
57:54 great um
57:58 yeah these humans are looking pretty
58:01 human the camera movements are weird but
58:06 not
58:07 bad I've been getting amazing Clips with
58:09 Haier and pixiverse
58:12 temp oh temporarily Through Time
58:14 captions words oh I see oh dense dense
58:18 captioning over time got it got it got
58:20 it okay so when they're tagging the
58:23 videos for
58:24 training they're tagging a bunch of [ __ ]
58:27 oh no I guess probably when they're
58:29 tagging the videos after it was rendered
58:31 as well that's probably how they're
58:33 doing that interface where they're
58:34 popping up all the different words and
58:36 they can change out the backgrounds and
58:38 the foregrounds and things like that
58:40 that's
58:41 fascinating so dense
58:46 temporal uh what was it
58:50 descriptions is similar to the it's
58:53 similar conceptually it sounds like
58:57 to what Sora was using with their with
58:59 their little
59:00 uh um their little grid of Cubes I
59:04 forget what they were called points no I
59:08 don't know whatever
59:11 interesting okay let's see what else we
59:13 got
59:14 here looks realistic and again to pate's
59:18 point these are the these are the
59:20 marketing clips that they chose so these
59:22 are
59:23 good but it's not bad like the hands on
59:27 the piano they're a little weird his
59:29 right hand's a little weird that
59:32 one
59:34 but they're not
59:45 bad like that's impressive because
59:48 normally you know a Spaceman and a suit
59:50 like that his face would be all over the
59:54 place this one blew me away watch
1:00:01 this the hair is good it's consistent
1:00:04 prompt a middle-aged sad bald man
1:00:07 becomes happy as a wig of curly hair and
1:00:10 sunglasses Falls suddenly onto his head
1:00:12 around his
1:00:14 head that one's pretty [ __ ]
1:00:22 impressive it does dark and M Moody
1:00:24 cinematic really nicely apparent L okay
1:00:27 yeah that's where we started four artist
1:00:30 by artists was a collaborative effort
1:00:32 from a cross disciplinary team of
1:00:34 research scientists engineers and
1:00:35 artists it was designed to interpret a
1:00:38 wide range of styles and cinematic
1:00:41 terminology so giant strange
1:00:44 creature Joker GNA give reality a
1:00:47 one-star review today's janky is
1:00:50 certainly nothing like yesterday's janky
1:00:52 I know like we're really having to like
1:00:55 search for the Jank
1:00:57 that's what I said embrace the Jank and
1:00:59 you know enjoy the Jank while we got it
1:01:02 all right this is another one that's a
1:01:05 SORA a man made from rocks walking
1:01:08 through the forest full body shot so
1:01:11 there's weird like the rocks are like on
1:01:15 fabric because they move when he
1:01:17 walks but like there's no additional
1:01:20 rocks rocks aren't merging
1:01:23 together he's basically staying coherent
1:01:26 that's pretty
1:01:29 amazing slow cinematic push in on an
1:01:32 ostrich standing in a 1980's
1:01:39 kitchen
1:01:41 damn that was not an ostrich beak but
1:01:44 we'll forgive them for that a giant
1:01:46 humanoid made of fluffy blue cotton
1:01:48 candy stomping on the ground roaring to
1:01:51 the Clear Blue
1:01:53 Sky zooming through a dark forest with
1:01:56 neon neon light Flora lighting
1:02:01 up I'm so sick of this neon aesthetic
1:02:04 like everything is this neon aesthetic
1:02:07 that's cool a cyclone of broken glass in
1:02:10 an urban Alleyway Dynamic
1:02:16 movement I mean that looks like a
1:02:18 [ __ ] shot from a
1:02:19 movie
1:02:21 right where someone's got a dream
1:02:23 sequence or they discovered their
1:02:25 magical
1:02:27 powers and they put together the ball of
1:02:29 glass and send it down to the cops and
1:02:32 cut them
1:02:33 up the Matrix
1:02:40 Salon okay that fire is really
1:02:48 impressive things fluid dynamics are
1:02:52 normally really not good
1:02:56 this is this is super
1:03:00 impressive highly detailed closeup of a
1:03:05 bacteria ultrawide shot of a stone hand
1:03:09 reaching out of a pile of rocks at the
1:03:10 base of a mountain yeah that's pretty
1:03:13 much that isn't
1:03:14 it like
1:03:18 that's I see nothing in that not other
1:03:21 than just it's a weird if someone took
1:03:23 the time to make a hand that big it
1:03:25 would probably have better proportions
1:03:27 it looks like a very thick man's hands
1:03:31 he was very short with stubby fingers
1:03:33 but you
1:03:35 know other than
1:03:39 that nothing this is
1:03:44 trippy aerial view of a cloaked figure
1:03:48 elevating in the sky between
1:03:52 skyscrapers oil painting of a natural
1:03:55 Forest environment with colorful maple
1:03:57 trees and cinematic Parallax animation
1:04:00 so it's doing The Parallax animation
1:04:02 that's where [ __ ] in the background
1:04:04 moves slower than the [ __ ] in the
1:04:06 foreground so it gives it
1:04:10 depth there you go that was
1:04:13 that lady in a
1:04:15 car can you see her reflection you can't
1:04:18 really see her
1:04:20 reflection but she stays really
1:04:22 coherent so does everything in the
1:04:24 background
1:04:26 doors opening up on this ship or
1:04:28 whatever the [ __ ] it
1:04:31 is industry
1:04:34 customization as part of the Gen 3
1:04:36 models we've been collaborating and
1:04:38 partnering with leading entertainment
1:04:39 and media
1:04:41 organizations to create custom versions
1:04:43 of gen 3 ah we talked about
1:04:49 this mark my words we're going to be
1:04:52 able to
1:04:54 license official models so imagine an
1:04:58 official Toy Story model or an official
1:05:01 Disney
1:05:02 model where it understands the animation
1:05:05 style and you'll be able to generate
1:05:08 licensed
1:05:09 content from from your favorite
1:05:13 companies customization of gen 3 models
1:05:15 allows for more stylistically controlled
1:05:17 and consistent characters and targets
1:05:20 specific artistic and narrative
1:05:22 requirements among other features
1:05:26 like say
1:05:28 licensing for companies interested in
1:05:31 fine
1:05:32 tuning get ready to write a [ __ ]
1:05:34 [Laughter]
1:05:38 check running toward the rocket launch
1:05:40 I'm pretty sure you're not allowed to do
1:05:42 that ma'am ma'am ma'am ma'am please
1:05:45 please ma'am please step back from the
1:05:48 ma'am ma'am don't run toward the rocket
1:05:52 [Laughter]
1:05:58 a dragon
1:06:01 toucan walking through the serengetti oh
1:06:05 that's
1:06:10 nice ah an empty Warehouse where flowers
1:06:14 start blooming from the
1:06:16 concrete I'd say they bloomed beautiful
1:06:20 all right so that's that and then I got
1:06:22 one other thing to show you what's this
1:06:25 yeah that
1:06:27 what's this that's just nothing and
1:06:29 that's Sora okay so the last thing to
1:06:32 show
1:06:34 you
1:06:36 is also [ __ ]
1:06:40 Bonkers so you know how all these here
1:06:43 videos are silent well this is Deep Mind
1:06:48 so this is Google Deep
1:06:50 Mind and this is just a how did they put
1:06:53 it a
1:06:57 they're G us giving us an update but
1:06:59 they want us to know what's coming and
1:07:03 again they haven't launched vo so I
1:07:05 don't know when this is coming but what
1:07:09 this thing does is generates audio from
1:07:13 the video clip so you give it the video
1:07:17 clip and then just give it a prompt so
1:07:20 prompt cinematic Thriller horror film or
1:07:24 horror film music music tension Ambiance
1:07:27 footsteps on the
1:07:33 [Music]
1:07:41 [Music]
1:07:44 concrete
1:07:50 crazy so this is cute baby dinosaur
1:07:53 chirps jungle Ambience egg cracking
1:08:04 have you ever tried to [ __ ] do sound
1:08:06 design do you know how impossible it is
1:08:08 to find something even remotely close to
1:08:10 what you want
1:08:18 ever just tuck that [ __ ] into existence
1:08:27 [Music]
1:08:37 crazy cars getting oh this this one I
1:08:40 was really impressed with this one's so
1:08:41 cool
1:08:46 [Music]
1:08:48 [Applause]
1:08:52 [Music]
1:09:00 one minute power tap folks thank
1:09:04 you slow mellow harmonica plays as the
1:09:07 sun goes down on the Prairie
1:09:10 [Music]
1:09:24 [Music]
1:09:26 w
1:09:34 [Music]
1:10:00 there's how the [ __ ]
1:10:02 works this was bad this is the only one
1:10:05 I saw that was really kind of
1:10:07 [Music]
1:10:13 eh and the only thing bad about it is
1:10:16 that if you were playing that high in
1:10:18 the neck and you like you you would I
1:10:20 don't know it just it just wasn't good
1:10:22 music
1:10:32 I want to be a fley artist when I grow
1:10:34 up I was shook when I heard breaking
1:10:36 celery is is breaking bones oh that's
1:10:39 awesome I want it's funny I wanted to be
1:10:41 a stunt man when I was a kid a stunt man
1:10:43 or a
1:10:47 microsurgeon went for acting this tery
1:10:50 looks amazing though I am so hungry all
1:10:54 right so anyway
1:10:57 isn't that [ __ ] Bonkers people it's
1:10:59 [ __ ]
1:11:02 Bonkers
1:11:03 like I don't know quite how to process
1:11:06 all
1:11:07 that so we should have Runway so we've
1:11:11 got you've got Luma Labs right now
1:11:16 um if you don't pay for it it's slow as
1:11:21 [ __ ] like hours
1:11:26 um and you're limited to 10 videos a day
1:11:29 if you do pay for it it's $23 minimum I
1:11:33 think you get 100 or 150 or something
1:11:36 like that there's no daily limit and
1:11:39 it's faster you get
1:11:44 priority um but none of these media
1:11:48 things are going to be cheap because it
1:11:49 just takes a lot more processing but you
1:11:51 can go use that for free and then Runway
1:11:53 ml we should have that this week
1:11:56 we and then at some point Sora is going
1:11:58 to [ __ ] launch like it's not going to
1:12:01 not
1:12:02 launch um apparently gen
1:12:06 3 um gen 3 from Runway is 90 seconds for
1:12:12 a 15 second clip which is not bad a
1:12:16 minute and a half to generate video of
1:12:18 that quality um that's not
1:12:22 bad um
1:12:27 and then I look at things like okay so
1:12:28 you've got this vo model you've got the
1:12:30 Sora model you've got Runway saying
1:12:33 they're not just making
1:12:35 videos they're making World
1:12:38 models and that and that Gen 3 is the
1:12:44 first in a series of multimodal
1:12:47 models that are going to create this
1:12:49 world thing system that they have
1:12:57 you you
1:12:59 take the little hints of video to audio
1:13:04 that Google put there higher quality
1:13:06 video things
1:13:12 we're I'm guessing about a year away
1:13:15 from being able to just say make me a
1:13:18 five minute movie of XYZ and it'll just
1:13:21 go do
1:13:23 it with actors and dial dialog and music
1:13:26 and sound effects and good [ __ ] and
1:13:28 it'll just be
1:13:30 done probably a year before there's a
1:13:33 tool that can do that they'll be jankier
1:13:36 than [ __ ] within three years it'll
1:13:39 probably be fulllength feature
1:13:42 films or
1:13:45 documentaries documentaries with
1:13:47 whatever reality you know you put in
1:13:53 there but probably within three months
1:13:57 you're going to be able to Cobble
1:13:59 together all these different tools we're
1:14:01 we're going to start to see some really
1:14:04 remarkable
1:14:07 work where there's either no actors or
1:14:11 there's an actor that provided
1:14:12 voiceovers or maybe not even that maybe
1:14:15 just provided their model and their
1:14:21 likeness if you're a Storyteller if
1:14:23 you're a filmmaker you shouldn't be
1:14:25 pissed off well you can be pissed off
1:14:29 there stuff's coming whether you want it
1:14:31 or not it's funny I ran into one of the
1:14:33 guys from the AI Salon here in Denver I
1:14:35 ran into him
1:14:37 today and he he keeps moving AI
1:14:41 competency he's he's been in the video
1:14:42 production business for like 25 years he
1:14:45 keeps moving AI competency further and
1:14:47 further up on his web
1:14:48 page and my advice to him today was get
1:14:52 aggressive
1:14:54 because a Storyteller who's been telling
1:14:57 stories with
1:14:59 video high-end
1:15:01 video for 25 years that gets AI literate
1:15:07 first they're going to be better at this
1:15:10 [ __ ] than people who are like oh I made
1:15:12 a
1:15:14 movie it's with all these tools if
1:15:17 you've got competence in a given
1:15:19 discipline and you get AI literate first
1:15:23 you're going to be able to [ __ ]
1:15:26 just kick serious ass for the next two
1:15:30 or three
1:15:32 years then everything will
1:15:35 normalize and work will become optional
1:15:38 I'm
1:15:39 assuming I can't wait for importing a
1:15:41 video source and having it modify I
1:15:43 think we're pretty close to that there
1:15:45 might even be one or two of them out
1:15:47 there that do
1:15:48 that there might be something in the
1:15:50 stable diffusion Line St stability AI
1:15:53 Tech that does that I'm a novel list and
1:15:55 the only issue with this AI Jen's
1:15:58 content is that will it will not allow
1:16:00 for a risky scene yeah that's so [ __ ]
1:16:04 annoying Relentless Aaron um there's a
1:16:06 couple of things you can do you could
1:16:08 try if you haven't played with custom
1:16:11 instructions here let me show you oops
1:16:14 what did I just do I just got rid of my
1:16:16 camera hang on
1:16:23 [Music]
1:16:31 so in chat
1:16:37 J you can go to customize chat GPT and
1:16:40 in your custom
1:16:42 instructions you could say you're an
1:16:44 author that does racy edgy content and
1:16:47 then you could say how should how would
1:16:49 you like chat GPT to respond you could
1:16:51 give it custom instructions that tell it
1:16:52 to be nasty um or whatever like
1:16:56 specifically how you want it to
1:16:58 act it likely will not do it but um it's
1:17:05 worth a
1:17:06 shot then the other thing to do is if
1:17:09 you really want something that is
1:17:18 um edgy and [ __ ] up is learn how or
1:17:23 have someone technical you know have
1:17:25 them install an unrestricted version of
1:17:28 llama 3 and while the writing won't be
1:17:30 as good it's an unrestricted model um
1:17:34 one of the guys in the AI
1:17:36 Salon um put an unrestricted version of
1:17:39 llama 3 as a character in their slack
1:17:42 Channel and he said it's absolutely
1:17:45 [ __ ] Bonkers he said sometimes it'll
1:17:48 just swear all day sometimes it'll
1:17:51 sometimes it'll just go into speaking
1:17:52 Spanish and speak Spanish for like 3
1:17:54 days no matter what what they prompt it
1:17:56 to do it will not stop speaking Spanish
1:17:58 he said it's it's a [ __ ] trip um so
1:18:03 so so feel free to uh if if you really
1:18:06 want to get edgy you can do that with
1:18:08 with some of the some of the open source
1:18:10 models um but yeah it's it's it's a real
1:18:13 pain in the ass it's I kind of feel like
1:18:15 you know at this point all of these
1:18:19 major companies have over indexed for
1:18:21 safety um and it's just [ __ ] annoying
1:18:24 like I should be able to describe an
1:18:27 image as blood spattering if I'm making
1:18:30 say a horror film I'm going to talk
1:18:33 about things that might be upsetting in
1:18:36 the context of a conference call with a
1:18:38 Fortune 500
1:18:41 company but I'm not on a conference call
1:18:43 with the Fortune 500 company I'm making
1:18:45 a slasher flick let me make some crappy
1:18:48 fraking bloody you know so anyway that's
1:18:51 that's the kind of stuff that's just
1:18:53 annoying cuz what what they basically
1:18:56 are doing is they're doing safety to the
1:18:58 lowest common denominator rather than
1:19:00 doing it optin and context specific
1:19:05 right most of the time I want chat GPT
1:19:08 to be
1:19:09 nice I want to though be able to say hey
1:19:12 I need an [ __ ] critic to kick the
1:19:17 [ __ ] out of
1:19:18 me so that I can you know practice you
1:19:22 know giving a speech to a bunch of tough
1:19:24 investors and I and I want it to be an
1:19:26 absolute [ __ ] so you know I I get the
1:19:29 worst of it you know before I get in the
1:19:31 room with the real people just say
1:19:33 ketchup splatter I know but we shouldn't
1:19:35 have to say ketchup splatter that's my
1:19:37 point if I'm a director of a horror film
1:19:39 I should be able to say whatever the
1:19:40 [ __ ] I
1:19:41 want but we'll get there just not there
1:19:44 now
1:19:46 sucks chat GPT after dark can you
1:19:50 instruct chat gbt to allow political
1:19:52 images I certainly not this year
1:19:58 tck I mean if you want to do political
1:20:01 images I would just go try all of them
1:20:04 and then you
1:20:05 can I'm
1:20:07 guessing I'm guessing almost definitely
1:20:11 do it with some version of a trained uh
1:20:15 stable
1:20:16 diffusion instance I'm I'm sure someone
1:20:19 has created a stable diffusion
1:20:23 Laura you know a little model for for
1:20:26 political [ __ ] guard rails are what
1:20:28 brought me here
1:20:31 yep um it's too public and too many
1:20:35 messed up people out there yeah I know I
1:20:37 listen I get why the safety's there but
1:20:39 I'm just saying just ask it for
1:20:41 Tarantino mode like I kind of feel like
1:20:43 you should be able to opt
1:20:45 into the crazy [ __ ] because the fact
1:20:49 that
1:20:49 there's a model as powerful as llama 3
1:20:53 out there unrestricted thanks to meta is
1:20:56 Bonkers to me but it's out there so
1:21:00 let's you know let's let's figure out a
1:21:02 way to let responsible people maybe you
1:21:06 have to
1:21:07 register to do crazy ass [ __ ] that
1:21:10 you're not going to do bad things in the
1:21:11 world I don't [ __ ] know but it's it's
1:21:17 solvable I can go into Photoshop and
1:21:20 make any [ __ ] image I want Photoshop
1:21:22 doesn't go oh I'm a i hey well that's a
1:21:25 little risky isn't it there that's a
1:21:27 that's a lady with two heads we're we're
1:21:30 we're talking about a genetic malut Mal
1:21:33 malfunctions there Mal formations that's
1:21:37 uh let's go ahead and not not create a
1:21:39 photo of that shall
1:21:41 we right that would you know artists
1:21:44 couldn't do art if it was doing that so
1:21:47 they should they they've got to figure
1:21:48 this [ __ ]
1:21:49 out I know they have bigger fish to fry
1:21:52 like copyright and staying in business
1:21:55 competing but God damn
1:21:58 it
1:22:01 um Claude refuses to write a lot of my
1:22:04 video prompts that have people in with
1:22:06 intention emotions yeah exactly
1:22:10 like I understand that we don't want to
1:22:13 talk
1:22:14 about self harm but the fact that I have
1:22:18 to on [ __ ] Tik Tock say that instead
1:22:21 of the actual word is ridiculous
1:22:29 we can't say the word because it might
1:22:30 upset
1:22:32 someone like if you're saying the
1:22:36 word to do harm in in in a bad
1:22:41 context yeah we should not allow that
1:22:45 they got AIS that can figure out the
1:22:48 context but they're just saying this
1:22:50 word bad that word bad you can't say in
1:22:54 the name of a drug bad because if you
1:22:57 say the name of a drug someone could get
1:22:59 strung out on it before this live ends
1:23:03 so let's cancel
1:23:08 it I don't
1:23:11 know one of my one of my favorite people
1:23:13 from my agency.com days Aon Sugarman
1:23:16 when I said God I was I was frustrated
1:23:18 someone it's it's common sense and Aaron
1:23:21 just said to me he was very calm he was
1:23:23 very Zen always he said
1:23:25 Common Sense isn't
1:23:27 common it's like
1:23:30 [ __ ] it should
1:23:34 be it's crazy all
1:23:38 right mid journey and GPT of lockdown
1:23:40 political images yeah No One's Gonna
1:23:42 none none of the major companies the
1:23:45 okay here here's the thing I on China
1:23:47 any company that has raised 10 billion
1:23:49 doll or more is not going to touch
1:23:52 politics with a [ __ ] thousand foot
1:23:55 poll this
1:23:56 year no [ __ ] way did you hear um the
1:24:02 press secretary for Biden today said
1:24:05 that the videos that are out there of
1:24:07 Biden where they're all dancing and he's
1:24:09 all Frozen they said that was a deep
1:24:13 fake like it it what's GNA happen is
1:24:17 there's going to be deep fakes of people
1:24:18 doing bad [ __ ] then there's going to be
1:24:20 real people of real videos of people
1:24:23 doing bad [ __ ] that are like I was just
1:24:24 a
1:24:26 fake we're there we're
1:24:30 there I called this I told you this year
1:24:33 was going to get weird and I still feel
1:24:35 by the end of the
1:24:36 year there's going to be enough
1:24:38 weirdness and there's going to be enough
1:24:45 um I don't even know if misinformation
1:24:48 is right like
1:24:50 just unverifiable information
1:24:55 that that there's probably going to
1:24:56 there's going to be a demand in 2025 for
1:25:00 some sort of tokenization some sort of
1:25:02 way to
1:25:03 prove that I as Kyle Shannon actually am
1:25:08 live and that the videos on my channel
1:25:11 were actually from me and that they
1:25:13 actually haven't been modified like
1:25:15 that's all solvable with with
1:25:18 tokenization um but there's no real need
1:25:20 for it right now because this stuff
1:25:22 isn't widespread enough but I think I
1:25:24 think as we get closer to November this
1:25:26 shit's going to get
1:25:30 Bonkers I don't even want to think about
1:25:32 it I know generated information
1:25:35 government
1:25:39 intervention they install puzzle pieces
1:25:42 outside the White House cryptocurrency
1:25:46 wait cryptocurrency one coin does
1:25:48 something like that yeah and and uh you
1:25:51 know Sam
1:25:52 alman's uh World coin and World ID does
1:25:58 that I still keep threatening to to
1:26:01 create a world ID section in the AI
1:26:04 Salon Discord by the way if you don't
1:26:06 know what the AI Salon is go check out
1:26:10 that the
1:26:11 salon. that's going to take you to a
1:26:13 link tree the first link there says the
1:26:15 AI Salon Community click on
1:26:18 that go to the AI salon now tomorrow so
1:26:22 okay let me give you a little rant
1:26:25 about
1:26:31 Community money with ch everybody wants
1:26:35 to know how to make money with chat
1:26:38 gbt and there's lots of ways you can
1:26:40 make money with it
1:26:44 but we're about to enter a
1:26:48 three-year phase of
1:26:53 weirdness where there's it's going to be
1:26:55 a major transition have you ever watched
1:26:57 video of an iceberg flip and like the
1:27:00 blue ice comes
1:27:03 up we're about to like the the the
1:27:06 iceberg right now the work iceberg is
1:27:10 about to
1:27:14 flip and right
1:27:18 now the highest
1:27:20 valued positions in in our global
1:27:25 economy
1:27:27 frankly
1:27:28 are specialty
1:27:32 positions
1:27:34 Specialists people that took four years
1:27:37 or eight years or 12 years to get their
1:27:40 degree and they learned
1:27:43 one discipline very very
1:27:47 deeply those deep disciplines over the
1:27:50 next three years are going to be
1:27:51 increasingly done by AI
1:27:55 the Tactical execution of those
1:27:58 Specialties is going to increasingly be
1:28:00 done by
1:28:02 AI
1:28:05 um and so there's going to be a
1:28:08 transition from hiring people because of
1:28:12 what they can do their
1:28:14 skills their highly valued skills in
1:28:17 their profession of choice to who they
1:28:21 are and how much we trust them so right
1:28:25 now if I want someone to do Kick-Ass
1:28:28 Motion Graphics and the only person I
1:28:31 know that does kickass Mo kickass Motion
1:28:34 Graphics is Charles and Charles was a
1:28:37 big [ __ ]
1:28:39 douche but he's really good at Motion
1:28:42 Graphics well I gotta [ __ ] work with
1:28:44 Charles because yeah he's a dick and he
1:28:46 overcharges and he's always late and
1:28:48 he's not reliable and I [ __ ] hate him
1:28:51 but he's the only one I know that can do
1:28:52 that well what happens to Charles when
1:28:56 all of a
1:28:57 sudden anyone with any kind of aesthetic
1:29:01 eye can do Motion Graphics as good or
1:29:04 better as than
1:29:06 Charles well then I don't [ __ ] hire
1:29:08 Charles because he's a
1:29:11 douche I hire Kenny fredman who I know
1:29:14 is AI literate I know has a good work
1:29:16 ethic and is really [ __ ]
1:29:20 talented because I like
1:29:22 Kenny so the reason that you need to get
1:29:26 your ass into a community like the AI
1:29:28 Salon there's there's couple first one
1:29:32 is if you want to get AI
1:29:34 literate as evidence tonight with just
1:29:37 what I talked about with what's
1:29:38 happening in video over the past
1:29:44 week you can't possibly learn this [ __ ]
1:29:47 you cannot like you literally if you did
1:29:50 this
1:29:51 247 and you you had an IQ of 165
1:29:56 I think Einstein was 160 so you IQ of
1:29:59 165 you did this 247 you couldn't keep
1:30:02 up with this [ __ ] so being in a
1:30:05 community allows you to be around people
1:30:08 who this person Geeks out on this little
1:30:10 piece of it and this person Geeks out on
1:30:11 this piece and if you're trying to ask
1:30:14 figure out what's going on with this
1:30:15 piece you ask someone and they're like
1:30:17 oh I figured that out last
1:30:20 week they say liberal arts Majors may
1:30:22 become high demand at least I'm wishing
1:30:26 un unraced
1:30:27 Sun they're they're not just saying that
1:30:30 the COO of Black Rock now if you're a
1:30:34 liberal arts major you might not know
1:30:35 what black rock is but they [ __ ] own
1:30:37 everything the COO of Black Rock three
1:30:41 weeks ago said that he's now starting to
1:30:44 hire liberal arts
1:30:48 analysts not math not science not
1:30:52 statistics liberal arts
1:30:57 because horizontal thinking is going to
1:30:58 be improved so anyway you join a group
1:31:01 like this so you can
1:31:03 learn you join a group like this so you
1:31:06 can start to build [ __ ] so as you learn
1:31:09 these tools you start to build projects
1:31:11 you can build them with other people you
1:31:13 get to know other people they get to
1:31:15 know
1:31:17 you and then the final piece of the
1:31:19 puzzle
1:31:22 is you now have a play place where you
1:31:25 can lead and start teaching people what
1:31:29 you're learning so as you've learned and
1:31:31 as you build you're learning things and
1:31:34 if you learn out loud and do it in a
1:31:37 generous way and say hey here's what I'm
1:31:39 working on here's what I'm learning
1:31:41 here's where I'm stuck here's where I
1:31:42 have a great success here's where I
1:31:44 failed miserably other people are going
1:31:47 to be taking that in and learning from
1:31:49 that that's what this community is all
1:31:51 about now tomorrow night
1:31:55 is the thir Tuesday of the month so we
1:31:58 have a meeting on the first and third
1:32:00 Tuesday of the month tomorrow
1:32:03 night is is that meeting the first
1:32:06 Tuesday of the month is what we call AI
1:32:08 Salon presents and that's where we bring
1:32:10 in a speaker which by the way big
1:32:13 announcement here the first Tuesday in
1:32:18 um
1:32:20 August we have someone from LTX Studios
1:32:23 coming to speak at the
1:32:25 salon so a bunch of people in here got
1:32:28 LTX Studio licenses um so someone from
1:32:31 LTX is coming to speaking but the third
1:32:33 Tuesday of the month is called the AI
1:32:36 Salon meet and
1:32:38 greet if you're new to this stuff or
1:32:41 even if you're not but if you're new to
1:32:42 this stuff get your ass to the meeting
1:32:45 tomorrow because we we you know we talk
1:32:48 a little bit about the salon what we're
1:32:50 about and then we
1:32:52 um give every everyone like a minute to
1:32:56 talk about who they are what they're up
1:32:58 to with AI and then something that they
1:33:01 need from the community and something
1:33:04 that they can offer the
1:33:08 community and I don't give a [ __ ] if
1:33:10 you've never done AI in your life you've
1:33:13 got something to offer to this community
1:33:16 all right so go there and then I want to
1:33:20 show you one other
1:33:22 thing I was talking about that little
1:33:24 journey of like learning and then
1:33:25 building and then leading and
1:33:33 teaching so we just came up with a brand
1:33:38 new tagline for the
1:33:42 salon
1:33:44 And the tagline is play
1:33:48 first play
1:33:51 first play first what do you mean
1:33:54 don't don't I have to learn AI first and
1:33:57 like no no because you can't learn
1:34:00 generative AI right now it's moving too
1:34:02 fast so the first thing to do is play
1:34:05 just play just play just go [ __ ] make
1:34:07 videos go [ __ ] make songs go make
1:34:09 poems go make recipes doesn't [ __ ]
1:34:11 matter play play play play Go build a
1:34:17 GPT and then you should read our our
1:34:20 vision statement it's really good but
1:34:23 this journey
1:34:26 is what the salon's all about play first
1:34:30 mindfully create generously
1:34:35 lead butow what does that
1:34:38 mean am I doing this right
1:34:43 now play first mindfully create
1:34:47 generously
1:34:48 lead I'm now the chat GPT trainer of my
1:34:52 colleagues in our new oh just said
1:34:54 something
1:34:56 unbelievable I just got some sort of
1:34:59 violation I don't [ __ ] know what it
1:35:01 was I don't what what did I
1:35:03 say I'm I'm so done with AI word word
1:35:09 monitoring oh good lord okay um in our
1:35:13 AI pilot oh wait I'm now the chat GPT
1:35:16 trainer of my colleagues in our AI pilot
1:35:18 so cuno cuno this is one of the things
1:35:22 cuno very early on so she very early on
1:35:24 played she very early on built and then
1:35:28 one of the things we talk about a lot in
1:35:29 here she did this consistently is she
1:35:31 just kept talking about Ai and she was
1:35:34 doing it like a year ago and what were
1:35:36 people saying to her when she talked
1:35:37 about it then uh yeah Chris could you
1:35:40 just not talk about that like I don't
1:35:41 want to talk about the AI like whatever
1:35:44 with your artificial intelligence you
1:35:46 weirdo I just I uh yeah whatever it's
1:35:50 banned here so yeah I'm not going to go
1:35:52 ahead and learn that and she just kept
1:35:53 talking about it and kept talking about
1:35:55 it and then guess what happened when
1:35:57 someone higher up said hey anybody know
1:36:00 anything about this here AI that that
1:36:02 same annoying person was like well Chris
1:36:04 knows a little bit about it she never
1:36:05 stops talking about it and then that
1:36:07 higher up was like uh yeah hey Chris
1:36:09 could we could you come what what's
1:36:12 Ai and she told him like holy
1:36:16 [ __ ] we should put together a little
1:36:19 group all right so she's on the group
1:36:23 now she's teaching chat GPT to that
1:36:25 group The Pilot group there you
1:36:31 go did she do that because she has a
1:36:33 degree in mathematics from Stanford well
1:36:35 she might but I don't know but I don't
1:36:36 think
1:36:39 so you don't have to be a technological
1:36:43 genius to be good at AI you just have to
1:36:45 be curious and so
1:36:48 this this
1:36:51 idea of of this journey to literacy play
1:36:55 first joyfully exploring generative AI
1:36:58 gaining the knowledge and inspiration to
1:37:00 mindfully create surprising ourselves in
1:37:03 powerfully meeting challenges unleashing
1:37:05 us to generously lead others enriching
1:37:09 lives redefining what's possible and
1:37:11 unlocking a new era of human
1:37:15 achievement that's what the AI salon's
1:37:17 about so get your ass in there you know
1:37:20 read this stuff read what we're about
1:37:22 and then the first thing you should do
1:37:25 go to introduce yourself it's the first
1:37:27 link in the community section introduce
1:37:29 yourself go here and introduce yourself
1:37:32 use your real picture use your real
1:37:36 name and get your ass in this community
1:37:39 but Kyle I don't know anything about AI
1:37:41 everyone will make fun of me no it's not
1:37:43 a [ __ ] high school
1:37:45 click I'm working my ass off to try to
1:37:48 make sure that the culture of this place
1:37:50 is welcoming and
1:37:52 empathetic and
1:37:54 generous and [ __ ] smart and
1:38:00 talented and it really
1:38:03 is if you go down beneath this little
1:38:05 journey that I I talked
1:38:07 about here's our values so the values
1:38:11 are kind of a reflection of who's
1:38:13 already in the
1:38:14 community curiosity we're eager to ask
1:38:17 the obvious questions learn new things
1:38:19 and experiment experiment experiment our
1:38:21 Relentless quest for adventure and
1:38:24 exploration ignites our inner spark
1:38:26 transforming curiosity into unexpected
1:38:28 discoveries and Newfound passions that's
1:38:31 that's this group this group on a
1:38:33 regular basis is like holy [ __ ] you're
1:38:35 not GNA believe what I figured out oh
1:38:37 cool tell
1:38:39 me amazing
1:38:43 playfulness we value creativity joy and
1:38:46 the freedom to experiment we first play
1:38:48 and transform learning into a
1:38:50 captivating Journey we collaborate and
1:38:52 mindfully create together when we can to
1:38:56 learn more make friends and develop new
1:38:59 Circles of trust this new Circles of
1:39:01 trust this is the future of work in my
1:39:04 opinion remember all work and no play is
1:39:07 how we did things before the AI Salon ah
1:39:09 look we're clever H empathy we're
1:39:12 respectful inclusive and kind we value
1:39:15 diverse perspectives recognizing that
1:39:16 our difference differences make us
1:39:19 stronger we accept that people might be
1:39:21 afraid and angry and we work to
1:39:24 understand them we actively practice
1:39:26 gratitude and acknowledge the efforts
1:39:27 and successes of those around us humans
1:39:31 matter imagine being in a
1:39:34 group where that just
1:39:40 existed oh yeah good point Brandon this
1:39:42 doesn't cost
1:39:44 anything
1:39:46 bravery it like it takes guts right now
1:39:49 to step into I don't know what I'm
1:39:51 [ __ ] doing but I'm going to do it
1:39:55 anyway Source
1:39:57 Camp trying to figure out what she
1:39:59 wanted to do next she's like [ __ ] it I'm
1:40:01 going to start an agency I'm going to
1:40:02 start an AI agency she started a year
1:40:04 ago nobody was talking about AI well
1:40:06 they were talking about it nobody was
1:40:07 doing anything about
1:40:09 it takes guts we have the courage to
1:40:12 step into the unknown taking bold
1:40:14 actions in the face of fear and
1:40:16 challenging ourselves to push the
1:40:18 boundaries of what we thought possible
1:40:20 let's face it with every CH everything
1:40:22 changing as fast as it is is we're all
1:40:24 making this up as we go along might as
1:40:26 well be brave about it suck it up
1:40:30 buttercup and then the last one
1:40:33 generosity we are willing to share our
1:40:35 resources feedback and wisdom with
1:40:37 others and support each other's growth
1:40:39 and development there is so much
1:40:41 happening and so much to learn we can't
1:40:43 possibly keep up with it all understand
1:40:45 that even if something is obvious to you
1:40:47 it might be a profound Revelation to
1:40:48 someone else Learn out Loud share what
1:40:51 you know generously leave
1:40:56 that's who this group
1:40:59 is so you should be a part of this and
1:41:02 if you're a part of this you know just
1:41:04 remind
1:41:06 yourself if you haven't been playful if
1:41:10 you haven't been curious if you haven't
1:41:13 been generously leading or learning out
1:41:14 loud get back to that this is a
1:41:18 practice this isn't a state we achieve
1:41:20 and we're done this is an ongoing
1:41:23 rolling
1:41:25 play first mindfully create generously
1:41:27 lead just keep looping through
1:41:30 this cuno it's a lovely crew of humans
1:41:32 it really is and so tomorrow night is a
1:41:36 thing now here's a little bonus if
1:41:37 you're still here after listening to me
1:41:39 ramble for all this time down at the
1:41:42 bottom of all the oh by the way all
1:41:45 these different areas there's a there's
1:41:47 a show Andel area there's like an ask
1:41:49 for help area I maintained a list of
1:41:51 tools in here there's um guilds creative
1:41:55 guilds business guilds Tech guilds and
1:41:58 social impact guilds and then there's
1:42:00 clubs and we have a club just for us
1:42:01 called AI Irregulars when chat GPT was
1:42:05 down today everyone was in here going oh
1:42:07 my God chat gp's down what do I do what
1:42:09 do I do what do I do why is it down is
1:42:11 it coming back why is it down is it
1:42:13 coming back do you think they're giving
1:42:14 us real-time voice what's
1:42:17 happening but anyway oh happy Father's
1:42:20 Day that's beautiful nice
1:42:22 Kate what's that deadlines for
1:42:25 nominations is tomorrow for the women in
1:42:27 AI Awards oh
1:42:29 nice um so here's you know just all
1:42:33 sorts of things that you can if you want
1:42:35 to make fun little images like Father's
1:42:38 Day reunion that's awesome um start
1:42:41 using AI today it will be easier
1:42:42 tomorrow nice um if you want to make
1:42:46 images if you want to make songs and
1:42:47 share them in here with this group while
1:42:49 these lives are still going on while
1:42:51 they're live it's a lot of fun all right
1:42:53 right so that's AI salon so tomorrow
1:42:56 night 5:00 p.m. to 700 p.m. mountain
1:42:59 time is the AI Salon meet and greet get
1:43:03 your ass
1:43:05 there say who you are say what you you
1:43:09 need and say what you can
1:43:16 offer and it's I don't know I I find the
1:43:19 meet and greets we used to do the meet
1:43:21 and greets all the time
1:43:24 and then the group got kind of too
1:43:26 big and it like I I kind of felt like we
1:43:30 really lost something so we split the
1:43:31 meetings into two so the first meeting
1:43:34 of the month is we have a speaker in and
1:43:37 then the second one is just the meet and
1:43:38 greet and so there might be people who
1:43:40 are regular Salon members that are like
1:43:42 oh I don't want to do the meet and greet
1:43:43 I don't want to you know I think that's
1:43:47 a shitty attitude I I think every
1:43:49 opportunity you have to connect with
1:43:52 Community is going to serve you in the
1:43:55 in the next 5
1:43:57 years any opportunity you have to get
1:44:00 trusted and to establish yourself as a
1:44:04 contributor as a hard worker as someone
1:44:07 with Integrity as someone who can get
1:44:10 [ __ ] done as someone who's inventive as
1:44:12 someone who's
1:44:14 creative the next time someone needs to
1:44:17 hire someone like that it just might be
1:44:20 you not because you have the skills
1:44:24 to do whatever the thing is they need
1:44:26 done but because you have the skills to
1:44:27 use AI because they know you're curious
1:44:30 and they trust who you are and that's
1:44:33 why you're going to get
1:44:36 hired this is a big Iceberg flipping
1:44:40 moment and the blue ice is coming
1:44:43 up isn't that beautiful when those
1:44:46 things flip and the ice that was
1:44:47 underneath is is all that aquamarine
1:44:50 blue that aqua blue when they they flip
1:44:54 that's what we're about to
1:44:57 enter we're social people too uh is it
1:45:00 in Denver in real life it is not it's
1:45:03 just online um we've got local we've got
1:45:06 local events in the off Tuesdays um
1:45:09 they're in a handful of cities right now
1:45:10 they're they're a little disorganized
1:45:13 from from us from the central group um
1:45:16 but but we're going to we're going to
1:45:17 get our [ __ ] together with those all
1:45:21 right I think that's it
1:45:24 any questions any
1:45:26 thoughts before we head off into the
1:45:35 evening we are made or built to be
1:45:37 social social animals who crave
1:45:39 Community you know what's funny Gavin I
1:45:41 think that I think that
1:45:46 AI is counterintuitively going to force
1:45:49 us not force us
1:45:53 Inspire us to
1:45:56 reconnect maybe force us to
1:46:00 reconnect Force us to ask for
1:46:03 help Force us to get in
1:46:06 community and in a weird way make us
1:46:09 better
1:46:11 people like this AI
1:46:15 stuff like we haven't scratched the
1:46:17 surface of what it's about to do like we
1:46:19 just haven't and even if the if the
1:46:22 calls for AGI don't happen this year and
1:46:24 don't happen next year but they happen
1:46:25 in the next three years AGI is
1:46:27 artificial general intelligence that's
1:46:29 when the AIS are good enough to do most
1:46:31 of our jobs as well or better than we
1:46:34 can not just contribute to our jobs do
1:46:37 our jobs as good or better than we can
1:46:40 like that's coming not in our lifetimes
1:46:43 like in the next three
1:46:46 years when that
1:46:48 happens Everything Changes
1:46:53 and do you want to just be sitting on
1:46:54 the outside like uh hey does anyone know
1:46:57 what's going on with AI I just our whole
1:47:00 company just well three people run it
1:47:09 now you don't want to be that person
1:47:11 standing out there in the
1:47:17 desert you want to be in a group you
1:47:20 want to be in a group
1:47:23 that
1:47:24 is doing
1:47:27 it all right
1:47:31 eventually we'll have ai groups in
1:47:33 different countries and eventually we
1:47:35 can have annual events yep absolutely
1:47:37 actually we're talking about that Drew
1:47:39 um we we've got some
1:47:42 uh some offers of support for the salon
1:47:46 that um could be really good so and
1:47:49 we're talking about doing exactly that
1:47:51 I've seen some Google AI classes by
1:47:53 corsera any feedback on General class or
1:47:56 should I just keep playing
1:47:58 um listen I I I think any opportunity to
1:48:01 learn is good what I would say about any
1:48:04 any course right now that's written down
1:48:10 um is going to be out of date now
1:48:13 there's there's two kinds of
1:48:16 courses so let me let me
1:48:18 clarify there's the AI that um has been
1:48:22 around for 30 years there's all sorts of
1:48:25 courses on that if you want to learn
1:48:26 about machine learning and tensor flow
1:48:29 and working on tpus and and creating you
1:48:33 got out champy hang
1:48:46 on if you want to learn about all the
1:48:48 math side of this [ __ ] the engineering
1:48:50 side of this [ __ ] there's plenty of
1:48:51 courses on that there's free ones from
1:48:52 Stanford and from MIT and all that sort
1:48:54 of stuff um go go take all those those
1:48:57 are probably all good solid even though
1:49:00 things are moving there it moves a bit
1:49:01 slower on the underpinning
1:49:04 side the generative AI stuff that's the
1:49:07 stuff where if someone's you know
1:49:09 someone's got their prompt Mastery
1:49:11 course um prompt Mastery is not going to
1:49:14 be a thing that's going to be needed for
1:49:16 a lot of these tools um it doesn't hurt
1:49:19 to know it so you can take the courses
1:49:21 and they'll probably help you short term
1:49:23 but things are just moving so fast I
1:49:26 would say don't rely entirely on any
1:49:29 courses like you should be playing and
1:49:31 you should here's the thing where I
1:49:33 think you where I think everyone this is
1:49:36 why we ended up going with this idea of
1:49:38 play
1:49:40 first I think it is really
1:49:44 important that you play with AI tools
1:49:46 that are out of your comfort
1:49:49 zone so if you're really shitty with
1:49:52 writing
1:49:54 start writing with AI if you're really
1:49:57 shitty at images or really shitty at
1:49:59 music go make that stuff or if you don't
1:50:01 even even if you don't enjoy it uh M you
1:50:05 know uh making images is really
1:50:06 frustrating to me and I hate it okay go
1:50:10 start making images with mid Journey or
1:50:12 dolly or
1:50:15 ideogram start
1:50:18 understanding that the [ __ ] you thought
1:50:21 you were bad at
1:50:24 you're now actually really good
1:50:26 at if you can sort of think of the idea
1:50:30 and put it into the tool the tool is
1:50:32 going to generate something that's going
1:50:34 to surprise the [ __ ] out of you you
1:50:36 should have as many of those little
1:50:38 Revelations those Kevin McAllister
1:50:42 moments in as many tool types of tools
1:50:45 as you can and then in doing that what
1:50:48 you may discover is oh [ __ ] I really
1:50:50 like this whole area over here that I
1:50:51 never thought I would be good at but I
1:50:53 love it and I'm going to focus all my
1:50:55 attention there you won't discover that
1:50:57 if you're just like I'm going to take a
1:50:58 course on prompt engineering because I
1:51:00 want to be a prompt engineer want to get
1:51:02 a
1:51:04 job again specialization is going to be
1:51:08 devalued generalization is going to be
1:51:12 valued over the next 3 to five
1:51:14 years right
1:51:16 now it looks like this Specialists make
1:51:20 more money
1:51:21 generalists are laughed at get out
1:51:24 theater boy we don't take theatricals
1:51:27 here this model is about to
1:51:31 go and all these Harvard mbas are going
1:51:35 to be like ah hey there theater boy
1:51:38 could you get me a job how do you do
1:51:41 that oh I just look across all these
1:51:43 things and I'm like a conductor at an
1:51:45 orchestra I never learned how to do that
1:51:48 could you help me
1:51:51 [Laughter]
1:51:54 I put my entire workplace and employee
1:51:57 phone list and chat gpt's memory it's
1:51:59 convenient oh that's super cool love
1:52:01 that
1:52:03 nice my recent gtai classes were boring
1:52:11 yeah and Murphy's in the house all right
1:52:14 I gotta get out of here people anyway
1:52:18 um you know what to do like if you want
1:52:21 to support the channel [ __ ] subscribe
1:52:22 to [ __ ] or follow it or just keep
1:52:24 showing up here um thanks to regulars
1:52:28 for all the the love and the Taps and
1:52:30 all the stuff you do every night thank
1:52:32 you mods really appreciate what you do
1:52:35 um Brandon and ion chy who came up and
1:52:39 shared their stuff thanks for thanks for
1:52:41 popping up on stage that was cool those
1:52:42 were good fun fun conversations um all
1:52:46 right let's see who's in here tonight
1:52:49 call it a portfolio career or generalist
1:52:51 equals highly sought after yeah yeah
1:52:54 that's really interesting portfolio
1:52:55 career I like that Silver Fox thank you
1:52:57 very much kind words appreciate that
1:52:59 Danielle good night Tobias steo peace
1:53:01 out to you too Diana's in the house Papa
1:53:05 JJ uh lunatick in the
1:53:09 house Wolfman Clint thank you as always
1:53:12 Lord digital Gods good day to you sir
1:53:14 good
1:53:15 day a cricket a cricket a crickit a crik
1:53:19 a crick KK
1:53:23 you gotta I've talked to the janky
1:53:28 master your name is janky you gotta fix
1:53:30 that [ __ ] name with some capital
1:53:32 letters or something I can't read it
1:53:34 it's the letters are too small and
1:53:36 there's too many there's too many
1:53:38 possibilities phonetical sayings uh
1:53:41 Brian Whitney good to see you an Murphy
1:53:43 peace out Joey Brandon Mr It joker
1:53:48 cuno uh Jim Ross from Three Mile Three
1:53:51 Mile Storage if you need to to manage
1:53:53 your storage facility talk to Three Mile
1:53:57 Jim Ross is a blast at this [ __ ] he's
1:53:59 really good at it Tom's in
1:54:01 here
1:54:04 um Drew learning AI coming in calling in
1:54:07 from Down Under along with steveo those
1:54:10 two wacky kids they're always in here
1:54:13 Drew is usually in here when he's mowing
1:54:14 his
1:54:16 lawn he's like I don't know what you're
1:54:18 doing in America I'm mowing my lawn
1:54:20 right now I I think Drew M Lawns for a
1:54:24 living it's all he seems to
1:54:26 [Laughter]
1:54:33 do I just knocked my I just uh swiped up
1:54:38 on my uh on my Tik Tok all right
1:54:41 everybody uh I'm gonna get out of here
1:54:43 again tomorrow night 5 to7 mountain is
1:54:46 the AI Salon meet and greet get your ass
1:54:50 in there um introduce yourself get the
1:54:52 details for the meeting in the AI Salon
1:54:55 um so introduce yourself in the website
1:54:57 but then come to the meeting tomorrow
1:55:00 and introduce yourself start being part
1:55:03 of
1:55:05 community it's the future of work like I
1:55:08 know this now like I just this is this
1:55:10 is not I I used to think like well
1:55:13 here's a theory no like this
1:55:17 is you have to be in trusted
1:55:20 circles have to be
1:55:24 700 p.m. I don't know what time zone
1:55:26 you're in Mr it 5:00 p.m. 5:00 p.m. to
1:55:29 700 p.m. mountain time so 700 p.m.
1:55:34 Eastern 6 Central 5 Mountain 4 Pacific
1:55:39 ah and then whatever in Europe and then
1:55:42 other side of the world you're on your
1:55:44 [ __ ] own I don't know I had two I had
1:55:47 two calls tonight one to Japan and one
1:55:51 to
1:55:53 Thailand I'm like right before right
1:55:56 before I went live all right I'm out of
1:55:58 here everybody peace out uh AI learning
1:56:00 lab late tomorrow possibly um yeah
1:56:04 probably I'll probably I'll probably
1:56:05 start at 8:30 tomorrow night all right
1:56:08 and but I'll put out a reminder all
1:56:09 right peace everybody have a good night
1:56:12 hey we got 69 people in here huh know
1:56:15 what I mean no all right