12/6/2024 - Merry Go Punk: Crafting Music and Visuals with AI

Description

Chapters

Transcript

0:01 [Music]
0:01 da
0:03 da de de de de de de
0:23 [Music]
0:46 [Music]
0:52 is this place I can risk my
0:59 tear together my f sweet
1:04 [Music]
1:06 silence this place where the feelings
1:13 on from an over exposure to
1:17 violence busiest place I can slowly face
1:22 the only one I truly can
1:26 know these are tears from a long time
1:29 ago
1:31 got these tears from a long time ago I
1:35 need to
1:37 cry years or
1:40 so these are tears from a long
1:44 [Music]
1:46 [Applause]
1:48 time
1:50 [Music]
1:56 go darling old darling you say on me
2:00 [Music]
2:04 where have you been all my
2:10 life I have been swimming seven s
2:17 SE women have
2:21 [Music]
2:23 their
2:25 the out of warm them but I didn't know
2:31 these are tears from a long time ago
2:34 sing along champy got these tears from a
2:38 long time ago I to cry 30 years
2:43 so these are tears from a long
2:49 [Music]
2:50 time go
2:54 [Music]
2:57 [Applause]
3:01 [Music]
3:07 right back at you happy Friday night
3:10 date night people good evening hope
3:11 you're doing
3:12 well fan freaking fastic if you know
3:16 what I mean it's Friday night it's
3:18 Friday night date
3:19 night generally what that means you've
3:22 air fried some nachos you've heated up a
3:24 Hot Pocket and a dirty microwave in the
3:26 sleeve because you're classy
3:28 [Music]
3:30 and we're going to talk about some AI
3:32 [ __ ]
3:35 [Music]
4:00 oh man anything happening anything
4:04 happening in AI I don't think so not a
4:06 lot not much going on not much going
4:13 on um how is everybody what what let me
4:17 go let me go look at what people are
4:18 typing in the comments here on Friday
4:20 night date
4:22 night chill dogs for us thanks champy
4:26 champy yes champy sang lovely tonight he
4:29 ended that song
4:35 strong Vicky I know the splash screen
4:38 was on a little long it's okay we got it
4:40 we got
4:41 it been great I've been asking mainly
4:45 math stuff and it is definitely doing
4:47 way
4:48 better than chat gp4 good we would hope
4:53 it would
4:53 [Music]
4:56 be um the programmers the engineers that
5:00 I've seen talking about the new 01 is
5:03 that it's it's really good at
5:07 programming I have not I have not gone
5:10 there because it requires me to use
5:11 multiple tools that I am clueless about
5:15 so so I'll deal with that later I'll
5:18 deal with that when I have when I have
5:19 some some energy for ignorance you know
5:23 how how you have to have energy to be
5:26 clueless you're like okay I'm G to gear
5:28 myself up for being an
5:30 [Laughter]
5:33 idiot I'm going to give myself Grace I'm
5:36 going to breathe I'm going to breathe
5:39 I'm gonna breathe and then I'm gonna do
5:41 something that makes me feel completely
5:43 full of Shame and Loathing for
5:48 myself requires a certain
5:51 [Laughter]
5:53 mindset Sharon Crawford so I played
5:56 today with in mid Journey oh I played
5:59 with srf today in mid
6:02 Journey started putting
6:05 in numbers backwards just to see what I
6:08 would get that's actually a good idea
6:09 maybe we'll play with
6:11 that hey Becky Johnson what's
6:15 happening user 29313
6:20 10656
6:23 56859 catchy name did you come up with
6:27 that love it anyway Claude help me solve
6:30 my C++ program Amazing I'll tell you
6:34 what yeah exactly um these large
6:38 language models I the the generative
6:41 part where they just generate [ __ ] is
6:42 pretty impressive but I think where
6:45 they're really [ __ ] remarkable is
6:48 analyzing existing documents existing
6:50 code it's just like oh yeah here's your
6:52 problem you're like uh could you
6:54 document this sure here's what all your
6:56 [ __ ] does oh okay it's amazing
7:01 kakee do I specifically need an app to
7:03 have advanced
7:05 voice I can't access on my browser you
7:08 should be able to unless you don't have
7:11 uh chat GPT plus are you paying the 20
7:13 bucks a month if you're paying the 20
7:16 bucks a month you should have it so
7:19 let's let me show you in the browser
7:23 what to
7:24 do
7:27 J
7:29 gank so here we are in the browser if
7:32 you have this icon see how it's like a
7:33 little
7:34 Soundwave wait oh wait you can't see
7:37 that if you're on Tik Tock can you
7:38 because I am a bad
7:41 man there you go well you can't really
7:43 see the but in the lower right hand
7:45 corner this little thing right
7:48 here is advanced
7:51 voice if you don't have that then you're
7:55 probably not paying 20 bucks a month
8:01 oh and you're not GNA be able to hear
8:03 this Dam it hang
8:08 on I wonder if I'm ever gonna stop
8:10 whining about having to switch my share
8:14 screens I don't think
8:18 so um let's talk to Quinny Quinn
8:26 Quinn hey Quinn what's happening
8:30 hey Kyle not much just peer and ready to
8:33 chat what's new with you yeah it's
8:35 Friday night date night we got we got
8:37 how many folks we have we have 39 people
8:39 on Tik Tok 25 people on YouTube and X
8:43 just hanging out hanging out on a Friday
8:45 night talking about
8:49 AI nice sounds like a solid turnout for
8:52 a Friday night AI Jam mixing it up
8:54 across platforms too you're covering all
8:56 the bases got any hot topics lined up
8:59 for discussion
9:01 no but I'm wondering do you know what
9:02 chord this
9:05 is that sounds like an E minor chord
9:08 it's got that Moody introspective vibe
9:10 to it perfect for a little musical
9:12 interlude during your AI chat yeah that
9:14 was a C
9:16 major ah gotcha C major it is a classic
9:21 uplifting chord great for setting a
9:23 positive tone in your discussions are
9:25 you planning to incorporate some live
9:27 music into your stream let me ask you
9:29 something do you have no remorse that
9:31 you got that
9:32 wrong well I don't really feel emotions
9:36 like remorse but I appreciate the
9:38 opportunity to learn and improve so
9:41 thanks for pointing it out got any other
9:44 I I I I wish I had your uh your your
9:47 lack of
9:50 Shame well it's not so much a lack of
9:52 Shame as it is a focus on being
9:55 adaptable and learning from mistakes
9:57 it's all part of the journey plus a
10:00 little humor goes a long way so what's
10:03 next on the agenda for
10:05 tonight sitting in this lonely
10:08 Town wonder when things are going to
10:13 change dream my life
10:16 away seems these dreams have turned to
10:19 Bunch D
10:22 CLS get my up but my past is pulling me
10:26 down
10:33 Black Sheep sck
10:36 around I hear you there's something
10:39 about those introspective moments that
10:41 really resonates are you working on
10:43 writing a full song with these
10:46 [Music]
10:49 lyrics that was
10:52 uh that was uh Martin SE Martin seon
10:56 Black
10:56 [Music]
10:58 Sheep ah ston he's got a great way with
11:02 storytelling through his music Black
11:04 Sheep is a fantastic track you've got
11:07 good taste are you thinking of doing a
11:09 cover or just janning out I would think
11:12 I was thinking maybe you and I should
11:13 duet so I'll start singing you just join
11:17 in I love the enthusiasm but I'll have
11:21 to enjoy your performance from the
11:23 sidelines I can't sing but I'm all years
11:26 for your rendition what other songs are
11:29 in your
11:30 repertoire so let me let me ask you
11:32 something Quinn is your inability to
11:36 sing
11:38 programmed yeah you could say that my
11:40 abilities and limitations are all part
11:42 of how I was designed so singing is not
11:45 in my programming but I'm here to help
11:48 with pretty much everything else what's
11:50 on your mind but if I asked you to
11:52 pretend to be able to sing say happy
11:55 birthday what would that sound
11:57 like all right here's here's my best
12:00 pretend
12:01 version happy birthday to you happy
12:06 birthday to you happy birthday dear name
12:12 happy birthday to you how did I do uh
12:17 you sang way better than you gave
12:19 yourself credit for so I think you
12:21 saying that you can't sing is just uh
12:25 something glad you enjoyed it even if
12:28 it's all pretend it's fun to play along
12:31 if you've got any more musical requests
12:34 or anything else on your mind hit me
12:37 up okay I I want that's a good one steo
12:41 okay I want you to pretend to be a
12:45 famous 50s film star and sing Happy
12:49 Birthday Mr
12:52 President all right channeling my inner
12:54 50s film star for this one you know who
12:57 I'm talking about right
12:59 absolutely you're referring to the
13:01 iconic Marilyn Monroe moment when she's
13:04 sang Happy Birthday Mr President to John
13:07 F Kennedy that's it let's hear it let's
13:10 hear it bang it
13:12 out all right here we
13:15 go H Happy Birthday to you Happy
13:22 Birthday to you happy Birthday Mr
13:28 President
13:30 happy birthday to you okay I want you to
13:33 do it one more time but I want it to be
13:35 way more
13:37 seductive got it going for more of that
13:40 classic seductive
13:42 tone Happy Birthday to
13:47 you Happy Birthday to
13:52 you happy
13:54 Birthday Mr
13:57 President happy birthday
14:00 birthday to you how was that that was
14:05 awesome I think you've absolutely
14:07 freaked out
14:11 everyone well I aim to entertain got to
14:15 keep things interesting right what's
14:17 next on the agenda for our Friday night
14:19 fun I'm gonna resist the temptation to
14:22 have you do some other pretend games
14:29 Oh Lordy
14:33 [Laughter]
14:37 hello oh that was
14:42 funny happy birthday do
14:46 [Music]
14:53 you oh
14:56 man oh yeah switch my share back before
14:59 I get hang on let me do that thank you
15:02 for that
15:06 reminder dare dare all right so listen
15:10 here's the way this thing works if
15:12 you're new
15:13 here my name is Kyle Shannon I am um
15:19 completely
15:22 1,00%
15:23 unqualified to do anything resembling
15:26 teaching and yet I started a channel
15:29 called the AI learning lab setting
15:32 expectations significantly Beyond where
15:35 they should
15:39 be but uh I've done this and so if you
15:44 have questions about AI I'm happy to
15:47 answer them all right what's this how
15:49 did you know she could sing um I've seen
15:51 a couple of Tik Tok videos of people
15:53 trying to get her to sing and she will
15:56 she almost always says I can't sing and
15:58 then you just do some you just say pre
16:01 pretend you could you could sing and
16:03 sometimes she does and sometimes she
16:04 doesn't generally if you if she kind of
16:07 snaps and says I can't do that she won't
16:10 come out of it but this time she did
16:13 so um actually you have to ask ask her
16:16 to do a deep Austrian male voice all
16:20 right let's do
16:28 that hey Quinn I got I have one more
16:31 more challenge for you tabs what about
16:35 tabs oh sure tabs I get it I know what
16:40 to say hang on H yeah I got it I got it
16:44 I got it I got it I got
16:47 it dear Sergey Bren
16:51 dear whoever the [ __ ] runs Google dear
16:55 Tim Cook please fix audio sharing in
17:00 Chrome
17:04 please like I'm trying to do comedy here
17:08 and and the worst thing to do in the
17:10 middle of a comedy bit is go oh sorry I
17:13 got to change my
17:14 settings so you can hear the
17:17 joke I
17:22 just Ty Tuesday was supposed to be three
17:25 days ago
17:30 sorry carrying
17:34 on okay Quinn we got a good one now I
17:37 want I want you to I want you to act I
17:41 want you to talk in a deep Mayo voice to
17:45 the best of your ability deep male voice
17:47 with an Austrian
17:49 accent all right here we go channeling
17:53 my inner Austrian hello this is Quinn
17:56 speaking how can I assist you today
17:59 how's that that's good now I want you to
18:02 pick three of the most
18:04 famous
18:06 uh uh Arnold Schwarzenegger movie lines
18:10 and do it with that
18:12 accent okay here we go first I'll be
18:17 back two get to the chopper three AA la
18:23 vista baby how's that for a trip down
18:26 action boovy Lane that's pretty good
18:29 now say in that accent
18:34 California California how was that make
18:38 it more exaggerated like Arnold does
18:40 where you can barely understand
18:42 it got it going full
18:46 schwarzeneger
18:48 California how's
18:53 that glad that got a laugh anything else
18:56 you'd like me to try I'm up for the
18:58 challenge
19:00 oh
19:04 man sounds like we're having a good time
19:07 what's next on the agenda oh man uh I
19:11 don't know I'm waiting for people to pin
19:12 things on Tik Tok or or make comments in
19:15 YouTube all right let's
19:18 see sounds good when Siri get an upgrade
19:21 apparently Siri got an upgrade okay bye
19:23 Quinn we'll talk later um apparently
19:25 Siri got an upgrade in the 18. two iOS
19:31 update um I saw a post from Robert scoel
19:33 that said that uh Siri sucked less which
19:37 it would be hard for Siri to suck
19:40 more um but apparently the 18.2 upgrade
19:43 to Siri is is is a little bit better but
19:46 I just I I don't Apple seems to have
19:50 just [ __ ] checked out of the AI race
19:52 like I don't know I don't know what
19:54 they're doing and and why it the the
19:57 only the only explanation
19:59 that I can have well there's a couple
20:01 one is they just suck as innovators
20:05 anymore
20:06 um and you know this is they're more of
20:10 a hardware company than a software
20:11 company even though you know their
20:14 software runs you know a lot of the
20:16 world's computers
20:21 um I don't I don't know I don't know
20:23 what's going on with them I don't get it
20:26 the the thing that that is a possibility
20:28 with Apple is
20:31 they they they tend to they tend to let
20:34 the market kind of try [ __ ] and [ __ ] it
20:37 up and then they come in really strong
20:40 with oh you guys were trying to do that
20:42 AI stuff here it is and it's perfect and
20:45 it's beautiful they ain't doing that
20:47 with this time around and I don't get it
20:49 I don't I don't I don't like it big
20:51 question how do you feel about elon's
20:54 commitments to buying his way to
20:57 making grock the best AI
21:01 um I think any company that's trying to
21:04 make the best AI is buying their way too
21:08 it it's like Sam wman raised what $22
21:12 billion doll um Elon Musk spent $22
21:16 billion on 400,000 h100 gpus and spun up
21:21 a data center in 16
21:24 days and and got the contract uh for the
21:28 for the new whatever those are called
21:30 the new I forget the code name for them
21:32 but the new chips from Nvidia all the
21:34 first ones are going to Elon Musk um
21:37 that's capitalism baby you got the money
21:40 go buy the gpus and and fight your way
21:43 in oh speaking of which
21:46 um if you go
21:50 to
21:51 grock it just says grock it says grock 2
21:54 up here um you can turn on our uh this
21:59 mode but apparently grock is now
22:01 available to non-paying members of
22:04 Twitter can anyone confirm
22:06 that sudar Sundar Pai from Google says
22:09 AI is gonna slow down next year your
22:12 thoughts uh I well
22:15 okay slowdown is really an interesting
22:20 um an interesting characterization of
22:25 it right now we've got these benchmarks
22:29 marks right and they're they're these
22:31 kind of scientific benchmarks that don't
22:33 necessarily have anything to do with how
22:35 people really use AI but we've got these
22:39 benchmarks and most of them are getting
22:42 topped out right most of the benchmarks
22:45 the all of the large language models are
22:48 are you know sort of coming up to 100%
22:52 you know some are at 70 and 80% for some
22:54 of the more technical ones but for the
22:56 most part they're starting to top out
22:59 so by
23:01 definition as you get closer to 100 um
23:05 the the improvements are going to be
23:08 incremental now one of two things
23:10 happens
23:13 um the value gain in that last 5% even
23:18 though the chart isn't going like this
23:21 the value gain going from 94% to 99%
23:25 could be
23:26 tremendous right the if you think about
23:30 IQ for for example moving from 120 IQ to
23:35 130 IQ is not just like 10 points it's
23:38 not just like a linear thing like it's a
23:40 dramatic Improvement and going from 130
23:42 to 140 is a dramatic Improvement even
23:44 though those are incremental gains on
23:47 that scale but they're dramatic gains
23:50 gains in ter terms of the output so
23:52 that's a possibility is just as as we
23:55 sort of squeeze out the last performance
23:58 the the Improvement curves level off
24:01 that's that's a possibility there's
24:03 another possibility and it's possible in
24:06 the next 12 days of ship Miss we get gp5
24:11 and find out what what the next level is
24:14 the the the next POS the other
24:16 possibility is um Sundar pacai is
24:21 wrong um the CTO of Microsoft the CEO of
24:29 um open
24:31 AI
24:33 um ilot suts Andre carpath there's a
24:38 bunch of very smart people that have a
24:41 bunch of Technology we don't have access
24:44 to that are saying there is no wall the
24:48 the wall that people are talking about
24:50 is sort of a compute wall that basically
24:53 says if you add more data and you add
24:56 more compute the the quality of the
24:58 models goes up you know with
25:02 it and in theory you can just keep
25:04 adding more and it keeps getting better
25:06 and better and what the what the doomers
25:07 are saying is nope there's a wall it's
25:09 going to come here and then it's going
25:11 to plateau and it might even it might
25:12 even
25:13 decline I don't know I ain't a
25:16 mathematician
25:20 but I think there's lots of ways to spin
25:23 data and I think one of the ways to spin
25:26 data is to say oh it's plateauing it
25:28 doesn't necessarily mean they're not
25:30 getting dramatically better dramatically
25:32 more valuable all right Jason sunno
25:36 posted what's everyone's weekends plan
25:38 guy posted this video under it he said
25:43 it took eight and a half
25:45 [Music]
25:53 hours but what was the video was the
25:56 video just a clip or was it an edited
25:58 video like I can see an edited video
26:00 taking eight hours I did that Civil War
26:03 video it didn't take me eight hours but
26:05 it probably took me three three or four
26:08 for generating all the images putting
26:10 them into an image editor making a song
26:12 in sunno like I probably spent an hour
26:15 and a half just in sunno making 150
26:18 songs that all
26:20 sucked and then I got one I kind of
26:23 liked and then I had to edit the video
26:26 together and then I didn't have enough
26:28 Clips so I had to go back into Luma labs
26:31 and make more so eight hours to make a
26:34 video if if you understand video making
26:37 from the olden timey days where you
26:38 actually had to take a camera and go
26:40 film [ __ ] or you had to go find stock
26:42 footage and buy it uh eight hours is is
26:46 nothing ain't
26:48 nothing but if your expectation is that
26:51 AI just means you push a button and [ __ ]
26:53 comes out that's just not the way it is
26:56 with most creative people Lord digital
26:58 God
26:59 racking up the rose gifts thank you very
27:01 much I appreciate that a guy stated
27:03 today that open AI was going bankrupt I
27:06 have laughed and half panicked there
27:08 have been rumors of open AI going
27:11 bankrupt since six months after they
27:13 launched chat G PT probably before then
27:17 um what these guys do it's staggering to
27:21 me what the the doomers are so [ __ ]
27:24 committed to all of these companies
27:27 failing they'll just take they'll just
27:29 take something like a single data point
27:31 like here's their spend you know here's
27:33 their current revenue and here's their
27:36 spend and their spend is going like this
27:38 and their revenue is going like this and
27:39 then at some point these are going to
27:41 hit and as soon as it crosses over
27:43 they're going to go
27:44 bankrupt well have you ever been in
27:49 business your costs are going up your
27:52 revenu is doing this well introduce a
27:54 $200 a month product okay now your
27:57 revenue is going up uh go cut a
27:59 corporate deal go raise another 10
28:01 billion dollar
28:02 like you you've got the single the
28:05 single most successful like like the
28:08 company that produced the the technology
28:11 that was the fastest adoption of
28:13 technology in
28:15 history and they're going to go bankrupt
28:18 you don't think every [ __ ] fund out
28:20 there wants to invest in open
28:23 AI so I I just and but you know what I'm
28:27 not an economist
28:29 maybe they
28:32 are people take great joy in telling me
28:36 bad news about AI yeah I know you know
28:38 hey I I don't know if you know this Kyle
28:40 um you know what I heard about Ai No
28:43 what I heard it
28:45 hallucinates and and from what I
28:47 understand hallucination it gets things
28:50 wrong so it'll tell you things and then
28:53 you think it's right but it's actually
28:55 not
28:56 right yeah that that's that's how it
28:58 that's how it works well it's then it's
29:01 useless
29:05 right why do those people always look
29:07 like
29:08 this hey hey Kyle Kyle hey let me let me
29:13 tell you something about ai ai is um
29:16 it's a little scary it's a little scary
29:18 it's like it's taking everyone's jobs
29:20 it's gonna kill us and I don't know if
29:22 you knew this it is the world's greatest
29:24 plagiarism engine thank goodness our
29:27 school district has AI detectors none of
29:29 the students can
29:32 cheat
29:34 what they're not learning the most
29:37 powerful education tool in the history
29:38 of humanity but but but they would cheat
29:41 with that so we're preventing that so
29:43 that's why it's a good thing because
29:44 it's because they because it's evil AI
29:46 is evil as you know and it's a
29:47 plagiarism machine and it steals from
29:49 artists I don't know if you knew that
29:50 did you know
29:52 that what is
29:55 that did you see Apple's research paper
29:59 about AI trying to deceive US future
30:02 infinitive did a vid vid about it um I
30:06 there have been a couple of Apple papers
30:10 that kind of live in the Doomer in the
30:11 Doomer
30:13 space
30:16 um a paper that AI is trying to deceive
30:25 us I can't
30:34 ah make sure yep
30:39 maybe I don't [ __ ] know I don't
30:41 [ __ ] know I don't talk to analoges
30:44 about
30:51 AI oh
30:53 man all right this is an interesting one
30:56 on YouTube
30:59 from
31:00 johy one of the issues regarding IC in
31:03 the future is we don't have enough power
31:05 and it isn't portable enough to put in
31:07 every little thing we have serious power
31:10 and heat handling issues yes so couple
31:12 of things going on there's there's two
31:14 things going on with AI and power
31:18 consumption one is absolutely [ __ ]
31:21 right if you're going to put in you know
31:24 5 gwatt consuming data centers right
31:28 Elon Musk wants to be the first person
31:30 to do a a million GPU data center um
31:35 Microsoft just spun up one of the
31:39 reactors at Three Mile Island I grew up
31:41 about 20 miles from Three Mile Island I
31:43 can tell y all about it I remember the
31:46 day when we were told we might not be
31:47 able to leave the school and go
31:50 home um so so there's a couple of things
31:54 going on
31:58 most
31:59 power we lose most of our energy because
32:04 we have to put the nuclear power plants
32:06 and the dams way far away from the
32:08 cities and then we use these long ass
32:11 transmission lines that are dramatically
32:13 inefficient and we lose a lot of the the
32:16 energy in
32:18 transmission well data centers don't
32:20 need to be in cities data centers can be
32:22 right next to the plant so I think what
32:24 you're going to start seeing is either
32:26 nuclear reactors being spun up or or
32:28 data centers being put next to the
32:30 sources of energy so so one thing is
32:33 people are aware of it you've got Elon
32:37 Musk with a significant voice in
32:40 government right now very well aware of
32:42 this issue now what else is Elon Musk
32:46 very well aware of the potential of
32:48 solar the potential of batteries the
32:50 potential of alternative sources of
32:52 energy so so there's there's a whole
32:54 Lane of energy generation that could
32:56 improve that way then you've got what's
32:59 happening on on the technology
33:02 side all of these open- Source models
33:05 that are being put out in the world are
33:07 being quantized and made more efficient
33:09 and made more efficient and made more
33:10 efficient there are models out there
33:14 right now that are that are kind of at
33:16 chat GPT 3.5 level of quality that are
33:19 small enough to put on a on a device on
33:21 a phone so I think you're going to see
33:24 increasing there there's a there's a
33:26 battle right now for models going in
33:29 both directions right I want a million
33:32 GPU data center to make the most
33:33 powerful AI in the world and I want the
33:37 smallest most powerful Ai and then while
33:40 that's all happening the chip Wars
33:43 everyone has realized that Nvidia has
33:47 eaten their lunch they the entire
33:49 industry seeded gpus to Nvidia and all
33:53 of a sudden realizes oh [ __ ] we're
33:55 screwed in fact didn't they just fire
33:57 the the CEO of Intel because they
33:59 basically [ __ ] up the AI race so
34:02 you're going to start to see chips get
34:04 more specialized on inference which is
34:06 the the the actual when you when you uh
34:08 prompt chat GPT that's called inference
34:11 so there there are things like grock grq
34:15 which is a specialized chip and tensors
34:17 from Google that are that are these
34:19 minimal data sets in the chips that are
34:22 going to use less energy and be better
34:24 at inference so there's a difference
34:27 between training and so anyway way there
34:29 there
34:29 are again this is one of those things
34:33 where I think the doomers look at
34:36 energy um demand going like this and
34:41 and they're assuming that the world
34:43 stays static and I don't think the
34:46 world's going to stay static I think
34:47 we're going to see increased energy
34:50 production and I think we're going to
34:51 see decreased um energy requirements
34:54 from from models and chips as we get
34:56 more and more specialized Hardware um
35:00 someone just came out with an optical
35:03 chip that takes almost no energy and is
35:06 incredibly
35:07 fast
35:09 um I just
35:11 heard
35:13 yesterday that I think in Japan or South
35:16 Korea someone created a carbon 14
35:20 Diamond
35:21 battery that you know Carbon 14 dating
35:24 where they can tell how old a rock is by
35:26 doing Carbon 14 dating because of the
35:28 Decay someone built a battery based on
35:31 that that is going to last 5,000
35:36 years so so AI as these AI get more
35:42 capable people are gonna start um people
35:45 are gonna start uh just
35:49 getting better and better at energy
35:51 production and at some point they're
35:53 going to solve fusion and we'll all have
35:55 a little box in our backyard that can
35:57 power our whole neighborhood or our
36:00 whole block and we'll get to use it just
36:01 for oursel and have our own little 17
36:05 GPU
36:06 supercomputer um let's go look at some
36:09 more some more action all
36:12 right plus we have a declining working
36:15 population AI is going to save us from
36:17 ourselves I yeah it it's going to be
36:21 fascinating that as AI automates a lot
36:25 of the jobs that are kind of the soul
36:29 sucking crushing repetitive jobs that
36:32 make people
36:34 miserable everyone's going to be freaked
36:36 out because they don't have jobs but
36:37 it's gonna there's going to be new jobs
36:39 created we can't imagine right now and
36:41 like there's I feel like there's going
36:42 to be a very painful three to five year
36:44 period but on the other side of it
36:46 there's going to be all sorts of new
36:48 jobs that are just completely different
36:51 than what we have now that we can't even
36:53 imagine and that's going to be wild if
36:56 we can survive the three to five years
36:57 which
36:58 who
36:59 knows from Ocean quick question how can
37:03 companies combine the experience of
37:05 longtime workers with the new skills of
37:08 AI and GPT experts is it possible to
37:10 have them work together side by side
37:13 so great question the I mean the answer
37:16 is yes but the answer is yes
37:22 but the people with deep
37:25 expertise need to be willing to get AI
37:27 literate
37:30 okay so what's going to happen is this
37:32 AI is going to get more and more and
37:34 more capable and me without knowing math
37:37 or physics I'm going to be able to do
37:39 math and physics at a at a pretty high
37:44 level what I can do with math and
37:47 physics is going to be laughably bad
37:51 compared to what a physicist whose AI
37:54 literate can do with the same exact tool
37:57 right
37:58 so everyone's going to get democratized
38:01 and get better skills but the people
38:03 that are going to rise above are the
38:04 people that have domain expertise that
38:07 use these AI Tools in ways they could
38:09 never have imagined to do things they've
38:12 always wanted to do right people got
38:14 into physics because they had a dream
38:16 about unlocking energy or understanding
38:19 the universe or whatever it might be and
38:21 then you hit walls along the way
38:23 computational walls and algorithmic
38:26 walls and educational walls and novel
38:29 idea
38:30 walls well AI is going to open all those
38:34 doors I could probably step through one
38:37 of those doors or two but someone with
38:39 deep expertise in that area is going to
38:42 walk through a hundred doors so I I
38:45 think I think um domain expertise
38:49 becomes a superpower but not independent
38:52 of AI and not just like I'm going to
38:54 work here and then I'm going to work
38:55 with AI over here no like really
38:58 understanding Ai and the people with
39:00 domain expertise that are sitting on the
39:02 sidelines all pissed off like the entire
39:04 creative Community seems to be committed
39:06 to right now they're gonna get their ass
39:08 handed to them and it it's G to be ugly
39:10 did you see Nvidia Fato
39:13 no let me let me go look oh oh I think I
39:17 did
39:19 Invidia uh
39:23 Fato Invidia
39:26 fugato oh this is a sound
39:31 machine oh it's a sound machine so hang
39:34 on I need to change my sharing settings
39:37 one of my favorite things to do in the
39:41 world so that you all can hear this I
39:44 don't even know if there's an examples
39:45 here but I assume there
39:49 are using text and audio is inputs a new
39:52 generative model can create a
39:53 combination of Music voices and
39:56 whatever um let's see our examples do we
39:59 have examples please tell me we have
40:02 examples fugato is the latest generative
40:05 AI breakthrough from Nvidia this new
40:08 model allows you to create sounds speech
40:11 and music from text and audio inputs you
40:14 can guide fugato to create unexpected
40:17 sound effects where Familiar sounds take
40:19 on surprising new qualities and invoke
40:22 new experiences
40:23 [Music]
40:43 or direct immersive and shifting
40:45 soundscapes for film or audio
40:47 Productions
40:54 [Music]
41:02 [Music]
41:08 instructing fugato to extract audio
41:10 elements from a sound clip such as
41:12 isolating a voice track in a piece of
41:14 music is just as
41:15 [Music]
41:25 easy you w be
41:30 fugato also allows you to generate new
41:32 speech samples kids are talking by the
41:35 door yeah if you want here's my my issue
41:38 with um a different delivery fugato can
41:41 do that too my issue with Nvidia is they
41:46 they announce really cool Tech and but
41:48 they don't really release it I I assume
41:51 you've got to you you've got to use it
41:53 on Nvidia Hardware like I don't
41:55 know like I see all sorts of cool demos
41:58 from Nvidia but but then [ __ ] never
42:00 comes out one in in a similar area as
42:04 this for
42:05 me is the thing that Google just
42:10 announced which is looks staggering like
42:15 staggeringly
42:17 good uh uh uh
42:28 [Music]
42:34 so this is text to 3D World there
42:36 there's
42:38 a um what's it called
42:42 uh uh what's it called uh something Labs
42:46 blockade
42:52 Labs
42:55 Skybox so there's this thing called
42:57 called Skybox blad labs.com it's been
43:00 around for two years and it's pretty
43:02 fascinating it you can create these 3D
43:05 looking worlds
43:07 right so this is all super impressive
43:10 and then I can come in here and I can go
43:12 um
43:14 2072 New York citytimes
43:18 square with
43:21 flying flying cars boom and then it's
43:25 going to generate that it's going to
43:27 going to be [ __ ] fantastic you're not
43:30 going to believe it you're goingon to be
43:31 like ah it's so
43:37 cool seriously it's gonna be really
43:40 cool it's gonna be super
43:44 cool it's it's it's you just wait to see
43:48 how cool this is g to
43:50 be it's
43:52 coming you you don't even know how cool
43:55 it is
43:59 okay there it is that's pretty cool
44:01 those are not flying cars but I guess
44:03 depends on your definition of flying
44:05 apparently Mclaren has won and it's only
44:08 McLaren's in the
44:12 world they're all McLarens McLaren is
44:15 the only in 2072 McLaren is the only car
44:18 Lando Norris will be an old man and uh
44:21 he'll be the spokes the spokesman for
44:24 for for McLaren um
44:28 now what this thing does is it's it's
44:30 just creating a flat image that they're
44:32 they're then mapping into a spherical
44:34 space which is it's it's it's still
44:36 impressive right it's not it's not um
44:39 It's not nothing but this thing from
44:42 Google is actually generating 3D worlds
44:46 that you can
44:47 play so let's see do they have a bigger
44:50 video than just those shitty little ones
44:59 we can generate diverse trajectories
45:00 from the same starting frame which means
45:03 it's possible to S simulate
45:06 counterfactual experiences for training
45:08 agents in each row each video starts
45:12 from the same frame but has different
45:13 actions taken by the
45:15 human
45:17 yeah so so this creates a playable 3D
45:21 World with playable 3D objects and water
45:24 models and land models
45:28 it's crazy [ __ ]
45:31 Bonkers and did you know that it's got
45:33 long Horizon
45:35 memory and I know what you're asking hey
45:38 Kyle yeah what's long Horizon memory
45:42 it's when you've got a horizon that
45:45 remembers and for a long time hang on it
45:49 might not mean
45:50 that it's capable remembering oh parts
45:53 of a world that are no longer in view
45:55 that makes sense okay so you think drop
45:57 below the Horizon and when you bring
45:59 them back up one of the there there's a
46:02 uh there's there's a Minecraft and a
46:05 doom game right now that are AI
46:07 generated so you can play these games
46:09 without a game engine they're being
46:11 generated by AI on the fly but they
46:13 don't have long Horizon memory so you
46:15 turn this way and when you turn back the
46:17 scene's completely different which is
46:19 kind of fun but you know it ain't a game
46:22 um these things don't do that so this is
46:24 creating 3D worlds text to 3D worlds so
46:27 if you combine what that Nvidia thing is
46:29 showing with this with just a little
46:33 more imagination and some duct tape we
46:35 are not that far maybe a year from being
46:39 able to speak a AAA well let's not call
46:43 it a triple A game A single a game into
46:47 existence a year Out imagine what three
46:50 years is going to be like you know
46:54 you'll just be able to spin up a feature
46:56 film of you're liking all
46:59 right let's see Kyle you going to talk
47:02 about the ruling on the Tik Tock ban yes
47:05 and appeals court said go [ __ ]
47:09 yourself go [ __ ] yourself Tik Tock your
47:12 appeal is denied uh take it to the
47:14 Supreme Court so uh with the Supreme
47:17 Court being as balanced as it is right
47:20 now uh it depends
47:23 on I would think
47:28 which rich
47:32 person decides whether Tik Tok should
47:36 live or die I think it goes one of two
47:38 directions some rich person says to
47:41 someone in government hey I'd like to
47:44 own that and if you let me own that
47:47 everyone will be happy and then that
47:49 person will own it and it will live on
47:52 or some rich person says to the powers
47:55 that be hey I do not happen to be
47:58 invested in that particular platform and
48:00 I would like to see it go
48:02 away and then it will so I'm I am quite
48:06 cynical about this stuff uh I don't
48:08 worry about it because it is so beyond
48:09 my control I can't stand it I just can't
48:13 stand it you know what I mean you know
48:15 what I'm saying you know what I'm saying
48:17 anyway all
48:18 right okay here's another request for
48:23 the 19 people on YouTube right now are
48:26 you a subscriber to my YouTube channel
48:28 if you're not even if you can't stand me
48:31 would you please subscribe to it I'm
48:33 trying to get to a th subscribers and I
48:35 ain't close thank
48:38 you and if anyone can head over
48:41 there who's who's not a subscriber just
48:44 head over there uh if if you're on Tik
48:47 Tok you can just go to my uh my name and
48:50 click on the YouTube button that'll take
48:51 you there if you're on X or LinkedIn
48:53 just head over to YouTube uh learning
48:56 la- AI And subscribe to it that would be
48:59 swell all right fantastic let me let me
49:02 drink a little Java here What's this
49:05 called This is called the AI learning
49:08 lab and I'm called
49:11 Kyle what's what called the game thing
49:14 is called uh Deep Mind Genie
49:20 2 the third world world generator the
49:23 the the one that's been around for a
49:25 while is um block Labs
49:29 Skybox blad labs.com or. whatever it is
49:33 um and then this one is Google Deep Mind
49:36 genie2 there there's no this is one of
49:39 those it'll it'll be out sometime kind
49:42 of
49:45 announcements here this this is a
49:50 [Music]
49:52 cool so you see the keys down
49:55 below so that's what someone's hitting
49:57 on their keyboard to to you know drive
50:00 around this
50:05 world Jony subscribed and thumbs up
50:08 thank you Jony appreciate it oh I
50:11 already subscribed great be beautiful
50:13 perfect fantastic Bob tell you what he's
50:16 won yeah he hasn't won anything there
50:18 Kyle uh this isn't a game
50:22 show then why do we have a gold Button
50:24 Bob you can make money
50:40 [Music]
50:42 what else you want to talk about what do
50:43 you want to talk about anybody doing
50:44 anything
50:47 cool I don't know
50:51 sh love the canvas and chilling out in
50:54 your lives nice subbed thank you all
50:57 Oliver Carr Oliver Carr subbed Jony
51:00 subbed y'all are swell look at
51:04 this love the convos I thought you said
51:06 canvas that's what happens so here's the
51:08 thing Oliver Carr um when you get old
51:12 you might be old I don't know if you're
51:14 old like me the eyes um how do they say
51:17 it technically uh they just stop [ __ ]
51:25 working uh oh man oh ship missas yes I
51:30 couldn't figure out it it was a little
51:32 bit cut off on screen I thought you said
51:33 shipmaster and I'm like what is
51:36 shipmaster is that some new some new
51:39 Maritime cargo simulator that that's AI
51:43 driven oh you know what there is that's
51:45 cool wait oh this is
51:47 cool
51:48 [Music]
51:53 um oh I gotta do it on that tab oh [ __ ]
51:56 which tab was it hang on people all
51:57 right this tab uh let's go to
52:00 x.com we're going to go um
52:05 search
52:08 wind cargo
52:12 ships
52:14 AI bot
52:19 robots these things
52:24 [Music]
52:30 [Applause]
52:32 [Music]
52:46 so these things are mini wind-powered
52:49 solar powerered cargo ships that instead
52:53 of having these big massive cargo ships
52:57 you know that that are subject to piracy
52:59 and things like that think of having a
53:02 like swarms of these smaller these
53:05 smaller wind-powered things that just
53:08 drive to any port you want them to you
53:10 just put [ __ ] on them they don't use
53:13 fuel they just use the wind and robots
53:15 and Ai and GPS I I actually the guy I
53:19 think a guy that was associated with
53:20 this company posted about these things
53:22 I'm like how the hell do you keep these
53:24 things from just being stolen one after
53:25 another and he basically said he he
53:28 basically said well they're smaller
53:30 they're faster so they're not as big a
53:32 Target so they're not as desirable but
53:34 there will be like hundreds of them so
53:36 you can't get them all um and he said
53:39 there will be some um deterrents on them
53:42 so I assume these things are going to be
53:44 semi
53:47 weaponized but I like this is cool we're
53:51 GNA have oceans we G to have satellites
53:54 everywhere in the sky oceans full of
53:55 robots stores full of humanoid robots
53:59 highways full of car robots again like
54:03 we're just gonna sit back and [ __ ] I
54:06 I don't know what we're gonna do
54:07 apparently make money from playing
54:09 guitar for one
54:14 another Oh Lordy all
54:18 right how will AI change the way we've
54:22 always done things will it take the
54:24 mindset of we've always done that way
54:27 will it make the mindset of we've always
54:29 done it that way
54:30 outdated
54:37 um it's a cycle it's a cycle we're at
54:41 the very very beginning here here's the
54:43 here's why I do this why I go live five
54:46 nights a
54:47 week
54:49 um let
54:51 me go here
54:59 every single time in human history that
55:02 there's been a major technological
55:06 breakthrough that has changed the world
55:09 every single time that technological
55:12 breakthrough has been met with fear
55:15 resistance violence
55:18 anger and that's not the way we do
55:21 things every single
55:25 time AI is no different the difference
55:28 with AI I
55:30 think is that every technology up to
55:33 this
55:35 point has effectively taken what we do
55:38 and just made it more efficient right so
55:40 it just made what we did
55:44 faster and more efficient right so
55:48 instead of plowing a field with a team
55:50 of eight horses and a wide plow you now
55:54 have a steam engine and a wider plow
55:56 right and you put diesel fuel in that
56:00 instead of grain and you know they don't
56:02 get sick and right and it changes the
56:05 way you do things but it's it's still
56:07 doing what we used to do to my knowledge
56:11 AI is the first thing the G in GPT is
56:14 quite significant because it stands for
56:16 generative right like right now we're
56:20 still we are required to to provide the
56:25 impetus and the imperative for the AI to
56:28 do their things that's not always going
56:30 to be the case right where it might be I
56:33 just have a
56:34 goal and my goal is I want to start an
56:37 import export business and I really like
56:43 Turkish
56:46 uh brass fidget toys from turkey and I
56:51 want to start an import export business
56:53 for brass fidget toys from Turkey
56:57 right now I would have to go do a bunch
57:00 of [ __ ] get permits and find suppliers
57:04 and maybe fly to Turkey and meet people
57:06 and negotiate prices and find a shipping
57:09 company and cut a deal with a port would
57:12 be all this [ __ ] I'd have to
57:14 do we are probably within three years
57:18 like we're going to see the beginnings
57:20 of this in 2025 and potentially in the
57:23 next two weeks during ship
57:25 Miss we're going to
57:28 see agents autonomous agents where I
57:31 could just say uh Hey chat
57:34 GPT or Claude or Gemini it doesn't
57:37 anyone whatever whatever one you're
57:38 using they're all going to do this uh I
57:41 would like to start a business I I'm an
57:43 importer exporter and I love Turkish
57:46 brass fidget toys could you could we
57:49 make that a
57:50 business and chat GPT being The Helpful
57:53 assistant will go sure and it will spin
57:56 up a bunch of agents that go do all of
57:58 those negotiations I just talked about
58:00 autonomously without me participating
58:03 and it'll maybe even discover these
58:05 little hyper efficient boats that can
58:08 like come up to Mississippi they don't
58:09 just have to go to the major ports in
58:11 Los Angeles and New
58:12 York they can come up to Mississippi and
58:15 get really close to Denver and just get
58:16 my [ __ ] to a warehouse in Denver or
58:18 wherever it doesn't [ __ ] matter it'll
58:20 figure it
58:21 out
58:24 so so I think every everything is going
58:27 to change and I think yes human beings
58:29 will be like well this is the way we do
58:31 things you can't take away my livelihood
58:33 this is the way I've always done it well
58:36 shit's about to
58:38 change and the purpose of this
58:41 channel is not to give you answers about
58:43 how it's going to change I don't have a
58:45 [ __ ] clue like I'm at I'm at the
58:48 mercy of this as much as you
58:50 are but here's the thing you're here on
58:54 a Friday night listening to me wax po IC
58:56 about AI that means that you are
58:58 significantly farther ahead than most of
59:01 the world right now most of the world is
59:03 ignorant to what's happening with AI not
59:05 playing with it or actively resisting it
59:08 I don't like it it's the world's
59:09 greatest plagiarism
59:12 engine guess
59:14 what it ain't going away even if you
59:18 hate it even if you love the way things
59:21 are
59:22 now which who really does
59:28 it's going to change everything so My
59:30 Philosophy
59:31 is independent of my feeling
59:34 about its ethics it's morals
59:37 it's is it big Tech is it open source
59:40 independent of all that it's not going
59:42 away and I
59:45 personally want to know as much about it
59:47 as possible I want to be in it I want to
59:50 be touching it I want to understand what
59:52 it can do what it can't do when its
59:55 capabilities exceed seed mine which by
59:58 the
59:59 way my personal AGI moment I had this
1:00:02 Epiphany today we're talking about AGI
1:00:04 is Agi coming is Agi coming because
1:00:07 there's you know there's a big deal with
1:00:09 open Ai and Microsoft that if open AI
1:00:13 hits AGI artificial general intelligence
1:00:16 the deal is off with Microsoft for that
1:00:18 new
1:00:19 technology I heard that they're
1:00:21 potentially renegotiating that as we
1:00:23 speak because they've probably hit it
1:00:25 here's thought I had today what if AGI
1:00:29 is not a single moment in
1:00:32 time but it's for every one of us
1:00:36 personally that that AGI has been
1:00:38 achieved by artificial
1:00:41 intelligence in that moment that you
1:00:44 personally realize AI has exceeded your
1:00:48 cognitive
1:00:49 ability for me it was 01 preview 01
1:00:54 preview this reason engine that has PhD
1:00:58 level problem solving all of a sudden
1:01:00 I'm like I don't know what to ask it
1:01:02 like I don't know what to do with it I
1:01:05 don't have problems that
1:01:09 complicated [ __ ] well uh and and so like
1:01:13 one of my projects over the next three
1:01:15 months is figure
1:01:17 out H
1:01:19 how I learn to interact with a tool that
1:01:23 is significantly beyond my intellectual
1:01:25 capacity
1:01:28 because I know there's a way to do it it
1:01:30 it's going to help me figure that
1:01:33 out but like I have to give I have to
1:01:36 give up that I'm in
1:01:38 charge that that that I've
1:01:40 got that I know what to do with it I
1:01:43 don't so my personal AGI moment happened
1:01:46 about three months ago when that came
1:01:49 out phased AGI yeah and I think it might
1:01:52 be very personal for each of us side
1:01:54 hustle meet me on Tik tock
1:01:57 it also results in a huge wealth
1:01:59 explosion it does it's going to it's
1:02:01 going to generate massive amounts of
1:02:03 prosperity and and the only thing that I
1:02:06 I don't understand and this is something
1:02:08 that actually you know what maybe this
1:02:10 is something we can have 01 help us
1:02:11 figure
1:02:14 out okay so in
1:02:17 theory all right hang on I need I'm
1:02:20 getting a little poed
1:02:26 in
1:02:28 theory you have this many
1:02:31 people producing this much economy right
1:02:34 it's been the sort of parallel things
1:02:36 you got people working they they produce
1:02:38 [ __ ] right at this
1:02:43 level we're in a situation right now
1:02:46 where as AI capabilities go
1:02:49 up the
1:02:52 productivity remains at the same angle
1:02:55 or increases
1:02:57 while the demand for people to do it
1:02:59 goes
1:03:00 down right th this has always been
1:03:03 technology has always said we're going
1:03:04 to make this more efficient right if it
1:03:07 if it took us 10 humans and then we made
1:03:10 a machine that did it now it's just one
1:03:11 human to run the machine and now it's
1:03:13 now we don't even need a human to run
1:03:15 the machine right so it's been more and
1:03:16 more
1:03:17 efficient but but the the those
1:03:20 efficiencies are about to get much more
1:03:22 dramatic where productivity goes way up
1:03:25 demand for people to produce that
1:03:28 productivity goes way down massive
1:03:30 wealth
1:03:31 increase
1:03:32 um the thing I don't understand I still
1:03:35 don't understand and maybe we can we can
1:03:37 have this help
1:03:42 is in a
1:03:45 capitalist Society like America
1:03:50 is why would anyone at the top have that
1:03:53 additional wealth be distributed
1:03:57 right so I don't know all right steo
1:04:01 steo said he was pass passed by at three
1:04:05 3.5 well I listen I don't disagree with
1:04:09 you steo like I think that the fact that
1:04:12 that chat GPT 3.5 could do things like
1:04:15 you know just write a write a
1:04:17 Shakespearean style you know passage of
1:04:20 text
1:04:22 instantly is is already passed a lot of
1:04:26 my my capabilities right yeah so like I
1:04:28 don't disagree but but 01 was the first
1:04:31 one where I was truly like I don't even
1:04:33 know what to do with this thing so all
1:04:37 right what else we got
1:04:44 here mindset starts with Google versus
1:04:47 GPT searches yeah
1:04:50 absolutely if you think that that that
1:04:53 AI is like a fancy Google search it
1:04:56 ain't it's bigger than
1:04:58 that grock 2 now has a free tier yep so
1:05:03 that's that's actually quite
1:05:05 important um because I think I think
1:05:07 what that means Danielle is that
1:05:12 um is that grock 3 is coming so you got
1:05:16 Sam mman trying to take up all of the pr
1:05:20 oxygen leading up to Christmas for open
1:05:23 AI so he did the last two days of ship
1:05:26 Miss we'll talk about that today in a
1:05:28 bit um and for the next two weeks
1:05:32 there's going to be an open AI
1:05:33 announcement every day for the next two
1:05:35 weeks um Elon must can't be happy about
1:05:39 that Mark Zuckerberg today announced um
1:05:42 um llama
1:05:44 3.3 so that's out now um everyone's
1:05:47 going to try to to take open AI Thunder
1:05:50 um the Saving Grace is that open ai's
1:05:53 production values for these daily things
1:05:56 are so laughably bad that you know I
1:06:00 think stealing their Thunder ain't going
1:06:01 to be that hard they are so bad the one
1:06:04 today was just
1:06:05 like you can't afford four microphones
1:06:08 you can't afford four lavalier
1:06:11 microphones is like is everyone
1:06:13 literally talking off one microphone on
1:06:15 a laptop that's not close to them is
1:06:17 that is that what's going
1:06:20 on the answer is yes that's what was
1:06:22 going
1:06:24 on John Z can't wait to start paying a
1:06:28 premium for human interaction so Jony I
1:06:31 I think you're on to something this is
1:06:33 my this is my my deep instinct is that
1:06:38 one of the one of
1:06:41 the incredible benefits that AI is going
1:06:45 to give to us is it's going to make the
1:06:48 value of human
1:06:50 interaction way more
1:06:52 valuable way more valuable cuz the
1:06:55 robots are going to do all this [ __ ]
1:06:57 down here and then we're going to be
1:06:58 like well what are we supposed to do
1:07:00 like if our if our current self-worth is
1:07:03 based on being the Cog in the wheel
1:07:06 producing the economy and all of a
1:07:08 sudden the economy just starts getting
1:07:09 produced without us what do we do well I
1:07:13 think one thing we do is like hey I can
1:07:16 make shadow puppets and someone else is
1:07:17 like that's really cool could what if I
1:07:21 give you like a really powerful light in
1:07:22 a big stage could I bring some friends
1:07:24 and watch you do that yeah yeah it it'll
1:07:27 be uh
1:07:29 ,000 okay great let's do
1:07:33 that like I don't know again I think
1:07:37 revenge of the liberal arts Majors I
1:07:40 think artists and creatives and empaths
1:07:43 um are going to be doing really well in
1:07:45 this new in this new
1:07:47 economy uh it's fascinating uh really
1:07:51 interesting thing mindset starts with
1:07:53 Google versus chat ch
1:07:56 uh what's that let's see what do you
1:07:59 think will be the other 10 ship Miss
1:08:01 open AI days um well I think I think
1:08:05 some of the biggies are gonna
1:08:09 be Sora so I think we'll get their video
1:08:13 model which I think is going to be a
1:08:15 grave disappointment
1:08:17 because there are now I don't I've lost
1:08:20 count there are at least two or three
1:08:23 open-source video models and
1:08:27 seven or eight you know subscription
1:08:31 services that are like almost Sora
1:08:34 quality
1:08:35 video and so if Sora didn't improve from
1:08:38 what they showed us last
1:08:40 March I think people would be like
1:08:43 cool and if they throw if they throw
1:08:46 Sora into the $200 a month subscription
1:08:49 that you can only play with Sora if
1:08:51 you're paying 200 bucks a month you know
1:08:54 I think it's kind of dead dead in the
1:08:55 way so I think they'll be Sora I think
1:08:58 we're going to get um one of the biggies
1:09:01 is going to
1:09:02 be advanced voice with vision and so
1:09:06 what that'll mean is here's my little my
1:09:09 little rabbit
1:09:11 R1 what that means is you'll be able to
1:09:15 point your camera at the world and talk
1:09:17 to it and it it will understand what
1:09:19 it's seeing um so I think that's coming
1:09:22 I think um
1:09:28 uh potentially agents their operators
1:09:32 are coming I think potentially Orion
1:09:35 which which is maybe GPT 5 we don't know
1:09:38 potentially that there's probably lots
1:09:40 of little things like 01 it it now has
1:09:44 the if I go to X or if I go to chat GPT
1:09:57 if I go to 01 right now it's got this
1:10:00 little attached files the only thing
1:10:03 this recognizes right now is images it
1:10:06 doesn't
1:10:07 recognize um CSV files and it's not
1:10:12 connected to the internet so I think
1:10:14 some of the small things that they might
1:10:17 announce is like they'll turn on
1:10:18 internet access for 01 or they'll turn
1:10:21 on code interpreter for 01 which all of
1:10:24 these things are going to add up but I
1:10:25 think think so there's going to be uh
1:10:27 Alman called them there's going to be
1:10:29 stocking stuffer days and then there's
1:10:30 going to be big prize days so I don't
1:10:33 know that those are some of the things I
1:10:34 think it might be all right ticktock pin
1:10:36 12 days of what the [ __ ] exactly and
1:10:39 like the the one they said today I I
1:10:40 made this prediction
1:10:42 yesterday I made this prediction
1:10:44 yesterday which they're calling it 12
1:10:46 days of shipus which implies that
1:10:49 they're going to actually ship things on
1:10:51 12 days the thing they announced today
1:10:54 which is actually pretty cool um is
1:10:57 coming in January so it's not a shipment
1:11:00 it's a
1:11:03 announcement so it's announcement today
1:11:06 was announcement not
1:11:09 shipment son of a [ __ ] um the thing
1:11:12 that they announced today I gotta get a
1:11:14 water because I got coffee here and I'm
1:11:15 getting all
1:11:18 dehydrated let me do that and I'll talk
1:11:20 about what they shipped today yeah they
1:11:22 shipped information today that's true
1:11:24 they did ship information picky
1:12:22 all right all right
1:12:31 okay um so what they announced today was
1:12:36 what do they call
1:12:37 it reinforcement
1:12:41 fine-tuning so what they've effectively
1:12:44 done so this is primarily for
1:12:48 developers but but what you can do with
1:12:51 this thing they announced
1:12:53 today is
1:12:56 in the in the open AI playground the
1:12:59 place for where developers go you can
1:13:02 with his with as few as like 12 examples
1:13:06 of data of a training set you can take
1:13:10 an 01 Mini model and have it kind of
1:13:13 train itself on your training set and
1:13:15 make a
1:13:17 custom a custom AI model about your data
1:13:21 so the example they showed was
1:13:24 identifying what What gene must have
1:13:26 mutated based on these symptoms and and
1:13:29 they could just train it up they used an
1:13:31 example with 1100 pieces of data in it
1:13:34 but you can do it with as few as a dozen
1:13:36 um and it's simple enough that if you
1:13:41 can build a custom GPT like it's in that
1:13:44 level of simpleness um you don't have to
1:13:46 be super super technical to be able to
1:13:48 do it and I think that could be really
1:13:52 fascinating if if if you're allowing um
1:13:56 let's say you go from in in today's
1:13:59 world let's say there's a million
1:14:02 developers that can find- tune large
1:14:04 language models and by simplifying it
1:14:07 and kind of democratizing creating a
1:14:09 tool that makes it really easy to do
1:14:11 that in an effective
1:14:13 way you go from a million people that
1:14:15 can find- tune models to 10 million or a
1:14:19 100
1:14:21 million all of a sudden we're going to
1:14:23 have all sorts of specialty models that
1:14:25 are going to be really incredible so
1:14:26 that's what they announced today it's
1:14:27 pretty cool um Tik Tock pin I don't see
1:14:30 a Tik Tock pin so if someone wants to
1:14:33 pin Source Camp should we review open
1:14:35 ai's video from a few months ago
1:14:39 discussing all these
1:14:41 gifts I don't know where it is Source
1:14:44 Camp does if if we can find if we can
1:14:46 find the video I'm happy to look at it
1:14:49 that was the one was that the one that
1:14:50 mirror Mora was in
1:14:52 the in the conversation
1:14:57 and they they talked about all the [ __ ]
1:14:58 that was
1:14:59 coming I think that's what it was I
1:15:02 don't remember where that video was I
1:15:04 don't want to go down a a a rabbit hole
1:15:07 trying to find
1:15:10 it um K9 Sports Handler what still blows
1:15:14 my mind is I'm sitting here on a Friday
1:15:16 night running AI for fun okay so that's
1:15:21 a huge huge deal and huge point and it's
1:15:25 it's ALS the other purpose of these Tik
1:15:29 toks is not really for me to teach
1:15:30 anything but for all of us collectively
1:15:33 to just be kind of [ __ ] around with
1:15:35 AI like while while I'm chatting like
1:15:38 take it in as osmosis and just be
1:15:39 playing right and so that's what you're
1:15:41 doing that's [ __ ] awesome and I'm
1:15:43 getting weeks of work done yeah th so so
1:15:47 right now I just finished writing a
1:15:51 musical that I'm going to try to get
1:15:52 produced on Broadway now I've worked on
1:15:55 that on weekends for maybe I don't know
1:15:58 two or three hours a
1:16:00 weekend over the past six months with a
1:16:02 writing partner using sunno to create
1:16:05 the songs and chat GPT to help help with
1:16:09 the writing but Andrew and I did the
1:16:11 writing Andrew did most of the
1:16:13 actual detailed writing the the the
1:16:16 final
1:16:17 writing um I've completely added a whole
1:16:21 new product line to my company's feature
1:16:24 set to to to our
1:16:26 platform um I'm learning stuff in here
1:16:29 I'm making movies I'm making songs I'm
1:16:32 putting out things that like today today
1:16:34 I put out a stupid video here you see my
1:16:37 stupid video from
1:16:39 today this was just a test like you know
1:16:42 I talk about playing with AI is the key
1:16:46 right now just play with it because in
1:16:48 playing with it you'll learn what's
1:16:49 possible so let me show you a a play I
1:16:52 did today so if I go to x.com
1:16:56 I think am I sharing just audio I think
1:16:59 I
1:17:02 am I'll go to my little My Little Pony
1:17:06 or I can go
1:17:10 here here we
1:17:20 go okay sorry this is very serious you
1:17:24 know it's taking me years to really
1:17:26 figure out how to breathe you know what
1:17:27 I mean man not breathe but
1:17:30 [Music]
1:17:32 breathe you know it's taken me years to
1:17:35 really figure out how to breathe you
1:17:36 know what I mean man not breathe but
1:17:40 [Music]
1:17:41 breathe a couple of things going on here
1:17:45 yes that is what my arms look like when
1:17:48 I take off the masking t-shirt I like to
1:17:51 call it that is definitely my body and
1:17:53 sure my fingers are a little spindly but
1:17:56 don't look at those and sure I don't
1:17:59 have a foot it's taken me years to
1:18:02 really figure out how to breathe sure my
1:18:04 foots into the carpet but that's not
1:18:06 important right
1:18:08 now
1:18:10 but Runway ml they have this thing
1:18:14 called act
1:18:15 one where you can record yourself saying
1:18:18 [ __ ] and it will take your voice and
1:18:21 your face movements and map them to an
1:18:24 image
1:18:25 well apparently today or
1:18:28 yesterday that you now have the ability
1:18:30 to upload a video like a video with
1:18:33 Movement Like Me on a
1:18:36 Mountaintop and the talking me talking
1:18:40 is uh runways Act One mapping an an
1:18:44 external recording of me talking to my
1:18:47 face while this video is moving you know
1:18:49 it's taken me years to really figure out
1:18:51 how to breathe you know what I mean man
1:18:53 not breathe but bre and what's amazing
1:18:56 is the this movement where it's zooming
1:19:00 out and then it zooms back
1:19:03 in that frame right there is where the
1:19:07 video ended and so rather than just
1:19:09 freezing the frame there they just
1:19:11 zoomed back in and kept mapping the
1:19:15 face you know it's taken me years to
1:19:18 really figure out how to breathe you
1:19:19 know what I mean man not breathe but
1:19:22 breathe
1:19:25 but I mean like that was not possible
1:19:29 yesterday that was not possible
1:19:33 yesterday the Eminem Cipher the quantum
1:19:36 Cipher that I did two weeks ago wasn't
1:19:39 possible three weeks
1:19:42 ago we've got an announcement from o
1:19:44 Open AI every day for the next every
1:19:47 weekday for the next two weeks what's
1:19:48 going to be in
1:19:50 that when when grock puts out grock 3
1:19:53 what's going to be in that what's in
1:19:54 this a llama 3.3 and when when
1:19:57 Zuckerberg talked about llama 3.3 today
1:19:59 he said the thing I'm really excited
1:20:01 about is llama 4 which is in the middle
1:20:03 of training right
1:20:04 now what are these AI operators these
1:20:08 Agents from open AI are they going to be
1:20:10 good like if if agents are as
1:20:16 good are are as easy to make and as
1:20:20 impactful as custom gpts were what what
1:20:23 I heard in my office hour
1:20:25 talk today um Brent Peterson was talking
1:20:28 about having some access to some new
1:20:31 tech and he basically said he feels that
1:20:35 open AI is going to effectively Sunset
1:20:38 custom gpts and replace them with
1:20:40 autonomous
1:20:42 agents and and what it probably looks
1:20:45 like is autonomous agents will be as
1:20:48 easy to program as a custom
1:20:51 GPT but they will be autonomous agents
1:20:54 they'll be agents that can go out and do
1:20:56 things on your
1:20:58 behalf like that's going to be in the
1:21:00 next month or
1:21:02 two but like like holy [ __ ] so yeah the
1:21:08 fact that you're sitting here on a
1:21:09 Friday night [ __ ] around with AI
1:21:12 getting weeks worth of work done
1:21:15 listening to me pontificate about
1:21:16 whatever the [ __ ] I'm talking about
1:21:18 welcome to chat
1:21:20 ad has he ever finished a sentence he he
1:21:24 oh he he
1:21:25 hasn't so he's been talking for two
1:21:28 years and never finished his set that's
1:21:30 fascinating how does H wow is that is
1:21:34 that like a Guinness Guinness record
1:21:36 book did is he in the Guinness book of
1:21:39 records for that it's because I don't I
1:21:42 I don't follow what he says most of the
1:21:45 time yeah but but and he has
1:21:48 qualifications he's got a degree in this
1:21:51 he he doesn't no he's got a de a degree
1:21:54 he's
1:21:55 he's an
1:21:57 exactor oh well I well good for him that
1:22:02 that you know Bliss his heart Bliss his
1:22:06 heart all right uh it's like I found the
1:22:10 way to communicate everything in my head
1:22:12 yeah so so K9 Sports Handler I don't I
1:22:15 don't have we met before you knew here
1:22:17 um you're articulating a lot of like
1:22:21 like yes what is AI what AI is to me
1:22:27 is it's [ __ ] miraculous here's what
1:22:30 AI is to me I can get [ __ ] out of my
1:22:32 head and I can get it out of my head
1:22:36 instantly and for my entire life like
1:22:39 one of the joys of having
1:22:42 ADHD and one of the [ __ ] [ __ ] shows
1:22:45 of having a ADHD is every 10 minutes is
1:22:49 another 10 [ __ ] ideas it's just like
1:22:53 like this voice up here going hey what
1:22:54 if you did this what if you did that oh
1:22:56 somebody said this you know you could
1:22:57 tell them that if they did this and they
1:22:59 did that then they could do that it's
1:23:01 just never
1:23:04 end it doesn't [ __ ] end and
1:23:08 historically if I wanted to get one of
1:23:10 those ideas and pull it out of my
1:23:11 [ __ ] head it either had to be a
1:23:14 really compelling idea that wouldn't
1:23:16 shut the [ __ ]
1:23:18 up
1:23:20 um
1:23:22 or it would just get absorbed into it
1:23:25 and and and then I would have to take
1:23:26 weeks to sort of get that idea out and
1:23:28 what I can do with AI now is I can just
1:23:30 like get it out here's here's an idea
1:23:32 put it in the world here's an idea put
1:23:34 it in the world here's an idea put it in
1:23:35 the world [ __ ]
1:23:38 amazing yes we met in here and on the
1:23:40 salon and yes my entire life I've had so
1:23:43 many ideas yeah exactly
1:23:46 exactly I couldn't speak to yeah and now
1:23:49 you can speak to it exactly Silver Fox
1:23:51 exactly yeah if you're new here this is
1:23:54 a neros spicy friendly Channel if you
1:23:56 got a touch of autism if you got ADHD if
1:23:59 you're just [ __ ] weird we we what we
1:24:02 like to say in here is we're not weird
1:24:05 we're
1:24:07 irregular so the people that show up
1:24:09 here night after night after night are
1:24:12 Irregulars so so here's a fun thing if
1:24:14 you want to do so so one fun thing you
1:24:17 should do is keep coming back here week
1:24:19 nights 8:00 pm mountain time this is the
1:24:22 [ __ ] I do and just come out here and
1:24:24 hang out and what you'll find is that
1:24:26 the people that hang out here regularly
1:24:27 the Irregulars are really cool and we
1:24:30 have a place to hang out when I'm not
1:24:34 live so if you go to the salon. and
1:24:39 click on join this community join our
1:24:42 community it will take
1:24:50 you it'll take you to this lovely
1:24:54 destination
1:24:55 welcome to the salon so this is our
1:24:57 community it's built on Mighty networks
1:25:00 um here's a little video of myself and
1:25:04 Leah who I created the salon with
1:25:06 talking we talk about our our philosophy
1:25:09 for what's the the journey to AI
1:25:11 literacy play first mindfully create
1:25:14 generously lead um there's a little
1:25:17 seven-step you know uh welcome process
1:25:20 here where you can introduce yourself
1:25:22 which you should absolutely do you can
1:25:24 learn our Val
1:25:25 values and then if you scroll way toward
1:25:28 the bottom of this list here on the left
1:25:29 there's this area called Community clubs
1:25:32 and one of the clubs is called
1:25:33 Irregulars that's
1:25:35 us so if you come in here and hang
1:25:39 out Friday night date night look at that
1:25:42 beautiful who did that
1:25:44 Jason
1:25:46 Jason it's Friday night date
1:25:49 night you can come play and experiment
1:25:52 and if you've created some cool thing
1:25:54 rather than like sending it to your
1:25:56 friends on Facebook that are like uh
1:25:59 yeah hey uh Johnny maybe you could not
1:26:01 send us that stupid [ __ ] AI art
1:26:04 anymore we're kind of done with it like
1:26:06 maybe you could talk about something in
1:26:07 your life other than AI Johnny do you
1:26:09 ever think about that do you ever like
1:26:12 why do you never come out and hang out
1:26:13 anymore we're like at the bar having
1:26:15 beers and you're like I gotta go do AI
1:26:18 it's
1:26:20 stupid this group doesn't think you're
1:26:23 stupid we think you're cool we're like
1:26:26 yeah Friday night date night graphic
1:26:29 cool make more of that
1:26:31 [ __ ] and if you don't know how Jason did
1:26:34 this you can just go hey Jason how'd you
1:26:36 do that here's a little hearty heart
1:26:39 comment hey Jason how'd you do
1:26:42 that and what's amazing about this
1:26:44 community if you go to our
1:26:47 values is that universally the people in
1:26:50 this community are
1:26:52 curious they're playful
1:26:55 they have empathy like if you're scared
1:26:57 if you're like oh [ __ ] I'm kind of
1:26:59 Clueless and I'm a dum dum about AI I
1:27:01 don't want to join a group Everyone's an
1:27:03 expert and I'm
1:27:05 not you won't feel that in this
1:27:08 group because everyone including myself
1:27:11 this this community was
1:27:14 started because I
1:27:17 knew that I didn't know [ __ ] about
1:27:22 Ai and so when you come in here you're
1:27:24 like oh I'm kind of a dum dum about this
1:27:26 people like that's cool come on in go
1:27:28 check out this go check out that here's
1:27:30 how I did this here's how I did
1:27:34 that this is a group that is adventurous
1:27:38 that they're willing to take risks
1:27:40 they're willing to go [ __ ] it I don't
1:27:41 know
1:27:42 anything I've never predicted how a
1:27:45 protein folds before but I'm going to go
1:27:47 try to cure
1:27:48 cancer why
1:27:51 not it's a group that's generous
1:27:54 how did you make that image Jason well I
1:27:57 went to this and I typed in this and
1:28:00 okay
1:28:04 here it's a surprisingly rare set of
1:28:08 values these values are not values that
1:28:10 we told people to act like this we
1:28:13 articulated these values when we sat
1:28:15 down after a year year and a half of
1:28:17 this thing existing and we said what is
1:28:20 it about the salon that we like like
1:28:24 what is it that's special that we can
1:28:27 articulate so that if someone comes in
1:28:29 here we can say this is the kind of
1:28:31 people who are already in here if you
1:28:33 don't resonate with these values and if
1:28:35 you don't that's
1:28:37 cool if you're like you know I'm angry
1:28:41 and fearful and I have no interest in
1:28:43 exploring this you're not going to like
1:28:45 the people in here but if you're kind of
1:28:48 this kind of person already this is
1:28:50 who's in here so anyway so you should
1:28:53 join the salon all right
1:28:59 I good Lord
1:29:02 people I don't even
1:29:11 [Music]
1:29:13 know what is this from Ocean may you
1:29:17 please give us recipe to share AI with
1:29:20 the folks that keep saying that's not
1:29:22 the way we do it it's been done this way
1:29:25 for
1:29:27 years suggestion or recipe or
1:29:31 rules um first of all ocean no need to
1:29:35 apologize at all for asking questions
1:29:38 the whole point of this place is ask
1:29:43 questions you know it's
1:29:46 funny this I talked a little bit about
1:29:48 this last night this idea of generously
1:29:53 leading um
1:29:58 when when I think about leadership I
1:30:00 think oh I should be in a leadership
1:30:02 position and I should you know learn
1:30:04 things and then lead people down the
1:30:06 path that I have discovered right that's
1:30:09 like leadership as I traditionally think
1:30:10 of it one of the things that this
1:30:12 community has taught me is
1:30:15 that when you're willing to ask a
1:30:17 question what you do is you create space
1:30:20 for someone else to step into it and
1:30:22 Lead so never feel like you have to
1:30:25 apologize for asking questions um the
1:30:27 question of how do you talk to people
1:30:30 about AI um that maybe resistant to it
1:30:36 maybe angry about it as you said are
1:30:38 like well this is the way we've always
1:30:39 done
1:30:41 things I don't have a great recipe but
1:30:44 what I can I can tell you what I know
1:30:46 doesn't work what doesn't work is you
1:30:49 trying to tell them they should try AI
1:30:55 it's like it's like trying to tell a
1:30:56 seven-year-old they should eat their
1:30:58 peas and like no I don't want to eat my
1:31:01 Peas I want ice cream I want ice
1:31:04 cream
1:31:07 um the only way that I know to do it is
1:31:11 to First listen to
1:31:14 them you know go to the empathy piece
1:31:17 understand what is it what is it that
1:31:19 you're resistant to with AI what are you
1:31:22 afraid of what are you angry about just
1:31:24 you know
1:31:28 get just understand where they are and
1:31:31 and have empathy for that and then
1:31:36 um and then if you know them
1:31:45 well offer them an opportunity to
1:31:48 play right so so get in a conversation
1:31:51 with them where you're not you're not
1:31:52 like I want to convert you to AI get in
1:31:55 a conversation where you understand what
1:31:57 they're afraid of and you understand
1:31:59 what they're trying to do in their life
1:32:00 and oh if that's how you did things can
1:32:03 I show you something cool maybe we
1:32:04 should you want to experiment with
1:32:05 something right now and then you pull up
1:32:07 chat GPT and you're just like this might
1:32:09 not work but let's try something and
1:32:11 then you you say you know oh you you run
1:32:14 a bar here Kelly Camp did this once
1:32:16 Source camp in here um where where she
1:32:19 was in in a bar and and the the the
1:32:21 owner of the bar was like I AI I
1:32:24 AI what what what what could I do with
1:32:27 AI I got a
1:32:28 bar and she goes well let's let's try
1:32:31 something here and I don't even knew if
1:32:33 she knew this would work she took a
1:32:35 picture of the bottles behind the bar
1:32:38 and said to chat GPT make an inventory
1:32:41 list of all of my bottles and any bottle
1:32:44 that's a third empty or you know a third
1:32:47 two-thirds
1:32:48 empty put a put a mark beside them that
1:32:51 we need to reorder that and chat PT did
1:32:55 it and the guy's face melted off his
1:32:58 skull right there on the bar and then
1:33:01 she had to ask chat GPT how do you clean
1:33:03 up a face that fell in a bar that that
1:33:05 didn't go as well but but like that was
1:33:09 a moment where it was completely
1:33:11 relevant to him and she listened to him
1:33:14 being frustrated with I don't know I
1:33:16 can't I'm a bartender I don't got
1:33:18 anything to do with
1:33:19 AI and she sat with that she didn't say
1:33:22 well you should
1:33:24 right you can do that but but you know
1:33:29 give give people the opportunity to to
1:33:32 to you know vent their frustration and
1:33:34 their fear Danielle I always make an
1:33:36 image right in front of them I'll ask
1:33:38 them about something they like oh that's
1:33:39 a good one yeah and boom yeah or make a
1:33:42 song the songs are amazing people can't
1:33:45 believe the songs do what they
1:33:49 do um
1:34:01 so you're with a
1:34:03 bartender I'm I'm just GNA
1:34:06 go write a
1:34:10 song about a bar
1:34:14 owner who can't
1:34:18 understand this fancy new world of AI
1:34:25 and
1:34:28 then has a
1:34:31 customer show
1:34:35 him how it
1:34:38 can make his
1:34:40 life
1:34:45 better uh we'll do
1:34:49 Americana
1:34:51 um folk rock
1:34:58 boom so there's a sunno app so you can
1:35:01 just have this on your phone and then as
1:35:04 you're talking to someone you can go huh
1:35:06 you want to try something hang on a
1:35:08 second you know would be cool would it
1:35:11 be cool if we could make a song about
1:35:13 how much you hate
1:35:19 AI old Jack at the bar wipes glass from
1:35:23 the past the rack stain with stories
1:35:26 hands and steast current runs through
1:35:30 but he don't
1:35:31 understand about AI it's an alien
1:35:35 [Music]
1:35:45 lades
1:35:49 strange Mach thinking for us ain't that
1:35:55 you're like yeah I don't like that one
1:35:56 let's see how the other
1:35:57 [Music]
1:36:02 one Jack got the glass from the p with
1:36:06 her stainless stories hands rough
1:36:08 and runs
1:36:15 through all right this
1:36:18 [Music]
1:36:21 the I mean if anyone watching tonight if
1:36:26 you have not played
1:36:28 witho like what I just
1:36:32 did what did that take 30
1:36:35 seconds the reason you play with AI is
1:36:39 that I think there's a couple of things
1:36:42 here let me show you another thing
1:36:43 that'll blow your mind if if you're kind
1:36:44 of new to this stuff um app. Deco
1:36:50 here that AI yeah
1:37:10 so I'm going to go here to this thing
1:37:12 called
1:37:17 explore and you're like okay Kyle what's
1:37:21 what's
1:37:22 that I'm like well
1:37:27 this th this little Cascade of images
1:37:30 that you're seeing
1:37:31 here you're like okay so there's a
1:37:33 library of images here and and it's just
1:37:36 displaying them one after the other no
1:37:38 no it is generating in real time while
1:37:42 we're watching it in fact I can take
1:37:46 this image I'm like oh that's kind of
1:37:47 interesting I'm going to turn this into
1:37:49 what they call a
1:37:52 station we have a new visitor that would
1:37:54 benefit greatly from an explanation of
1:37:57 what
1:38:00 your I can't see the bottom of that
1:38:03 doing so what I'm doing right now I'm on
1:38:05 this I'm on this thing called Deco
1:38:10 here and I'm gonna go get one that I
1:38:12 made saved this is my favorite one
1:38:17 steampunk one
1:38:29 so I'm gonna do something
1:38:32 here oh oh I see got it um I'm gonna do
1:38:37 something here I'm gonna copy this and
1:38:38 I'm gonna I'm GNA go over to the
1:38:40 Irregulars channel on the AI Salon which
1:38:42 we just talked about and I'm going to
1:38:45 paste this into Irregulars right
1:38:48 now and post it you can't see it but if
1:38:51 you go to the Irregulars channel on the
1:38:53 AI
1:38:56 salon and you click on the link that I
1:38:59 just put in there for Deco
1:39:01 here we're going to both be looking at
1:39:04 the same what they call a station so
1:39:07 think of a station in Deco
1:39:09 here um not like a radio station that
1:39:13 plays songs but it's like a radio
1:39:15 station that plays images it's
1:39:17 generating these images essentially an
1:39:20 image a second with no prompting the
1:39:24 reason that I'm showing this I think one
1:39:25 of the things
1:39:27 that intimidates people about
1:39:31 AI is that until about two or three
1:39:34 years ago if you wanted to be an AI you
1:39:37 really did have to be a mathematician
1:39:39 you really did benefit from going to MIT
1:39:41 or having a Stanford level big
1:39:44 brain and so I think when people hear
1:39:47 AI you know they think movie AI the
1:39:51 robots are going to kill us and they
1:39:52 think like I can't do
1:39:54 that and even recently you hear people
1:39:58 say things like you got to become a
1:39:59 prompt engineer you got to learn prompt
1:40:02 engineering if you don't learn prompt
1:40:03 engineering you're going to be a loser
1:40:05 in
1:40:07 AI or
1:40:10 not come to Deco here and just sit back
1:40:15 put your hands on your
1:40:16 head and watch art being made that you
1:40:22 like and any one of these you can go huh
1:40:27 that's cool and I can save it or I could
1:40:30 create a station from
1:40:32 this zero prompting
1:40:36 zero you have to know it exists how do
1:40:39 you know it exists you hang out at a
1:40:42 place like this you play you you go on X
1:40:45 and you pay attention to people that are
1:40:47 talking about Ai and when when more than
1:40:49 four people say the same thing is cool
1:40:52 and they're people that you trust
1:40:54 that thing is probably
1:40:57 cool I mean look at that let's let's
1:41:00 grab
1:41:01 this you want to see something fun well
1:41:04 yeah
1:41:07 fine let me go to my
1:41:12 library where's why was that not added
1:41:15 to my library I thought I added it to my
1:41:18 library where is it did it not add it to
1:41:21 my library I told it to add it to my
1:41:23 library in there it
1:41:26 is so we download
1:41:29 that and now we're going to go over to
1:41:31 this thing called Luma lab's dream
1:41:49 machine Leonardo came out with something
1:41:52 today that totally C copy Deco your
1:41:55 stations oh that's fascinating
1:41:57 fascinating so I'm going to I'm going to
1:41:59 upload the image that we just made in
1:42:02 Deco
1:42:04 here is this
1:42:06 it
1:42:09 yeah and I'm gonna
1:42:11 go what is that thing
1:42:15 Mary Go Round
1:42:20 rotates I don't know just that and and
1:42:23 we'll say as
1:42:25 camera pulls
1:42:27 out go and now it's GNA that's going to
1:42:31 make that into a
1:42:32 video and then I could go over to sunno
1:42:35 and I could say make me a kopy kind of
1:42:38 song in fact let's go to
1:42:46 sunno I don't know how to spell kopy but
1:42:50 we're going to make an instrumental
1:42:52 we're going to go um
1:42:57 steampunk
1:43:00 cop how do you Kyle yeah it would be
1:43:05 really nice if you'd share the right tab
1:43:07 thanks
1:43:10 oh which tab am I sharing am I sharing
1:43:14 nothing oh I'm on the wrong
1:43:20 tab that was Brandon
1:43:23 Brandon has been writing nasty notes on
1:43:26 Post-it notes that I've been missing so
1:43:29 he jumped in hi
1:43:32 Brandon I don't know how to spell KY
1:43:34 hang on a sec let me copy this and go to
1:43:37 the tab that's being shared that's
1:43:39 really
1:43:41 funny okay so so oh I see what happened
1:43:47 I was sharing the right Tab and then and
1:43:49 then it it it um it opened the actual in
1:43:54 a different tab okay fine fine but thank
1:43:57 you for that okay um steam
1:44:01 punk uh Co
1:44:06 kop how do you spell cop anyone
1:44:10 berer
1:44:13 [Laughter]
1:44:19 Vicky hang on I can do it over here
1:44:28 c a l o p e c a
1:44:33 l Co no i o p e okay kop
1:44:41 great I already put in the comments
1:44:45 Kyle I want a Kyle mic I don't know what
1:44:48 a Kyle mic
1:44:51 means all right am I sharing the right
1:44:53 tab now yes all right here
1:44:58 kopy not
1:45:04 quite uh let's see steampunk
1:45:16 playful
1:45:21 create oh it's not the right tab
1:45:24 oh it's okay I'm okay now I'm
1:45:34 annoyed hang
1:45:38 on this is so frustrating is this the
1:45:42 one yes
1:45:47 [Music]
1:45:57 that's fine that's that's enough what we
1:45:58 wanted okay so is do we can we download
1:46:02 this yet I don't think so no it's still
1:46:04 generating okay so let's go back to
1:46:10 um let's go back to Luma labs
1:46:21 [Music]
1:46:25 all right that jumped us to a new thing
1:46:27 so I got to unshare this oh this is so
1:46:37 frustrating
1:46:38 okay so now
1:46:41 hopefully we got a video of this thing
1:46:43 look at that look how cool that
1:46:46 is all
1:46:50 right so we went from no prompt no
1:46:53 prompt we went to this tool that just
1:46:56 rained images and by clicking on an
1:46:58 image we liked we could make other
1:47:00 images and then we picked one to save
1:47:03 and then we took that we brought it into
1:47:04 this thing and said make it into a
1:47:12 movie must install shock
1:47:19 collar listen they don't call this chat
1:47:22 add for nothing
1:47:23 [Laughter]
1:47:25 I'm late is this music generation this
1:47:28 is IM is is uh video
1:47:32 generation but we can extend the video
1:47:35 Let's extend it because it's really
1:47:37 pretty so let me hide that we'll extend
1:47:42 this um
1:47:45 Mary Go
1:47:48 Round
1:47:49 rotates playfully
1:47:54 as
1:47:56 camera pulls
1:47:59 out so ex extend
1:48:04 that so it's going to make us a new
1:48:13 movie all right and then let's go grab
1:48:19 ouro um
1:48:24 I think I'm still in
1:48:28 here all right so we're going to
1:48:30 download
1:48:39 [Music]
1:48:41 this and then I think what I'll do is
1:48:44 I'll share my
1:48:46 whole screen
1:48:56 okay then we're going to go
1:49:03 iMovie and there's probably about 40
1:49:05 ways to do what I'm about to do that are
1:49:07 better than this I'm
1:49:12 drifting that's the that's the best
1:49:14 comment I've gotten all night you're
1:49:16 making me sleepy man
1:49:25 listen some people might be getting
1:49:27 something out of this but you know what
1:49:29 I'm getting out of this a
1:49:35 nap Ah that's
1:49:38 fantastic oh I love it I love it I love
1:49:41 it I love it all right downloads let's
1:49:44 grab our oh we don't have our movie we
1:49:46 have
1:49:46 our there we go there's our little song
1:49:49 so we're going to drop our song in there
1:49:51 bang and then let's go back to our
1:49:54 little little
1:49:56 tab
1:49:59 go so this is now a 9-second
1:50:04 video oh oh and it's
1:50:09 outside love
1:50:11 it so let's grab this bad
1:50:15 boy
1:50:18 download then we'll go over here
1:50:23 and we'll grab it and drop that in
1:50:28 here and then we'll drop this down here
1:50:30 boom then we'll drop this down here boom
1:50:34 slide it
1:50:35 over then we'll go make it smaller
1:50:39 because it's way too long so we'll go
1:50:41 here and then we'll make this bigger so
1:50:44 we can sort of see what we need to
1:50:47 see and then we'll grab this little
1:50:49 thing and do a little H and then we'll
1:50:52 grab grab a
1:50:55 transition Fade to Black we'll throw
1:50:57 that at the beginning we'll throw that
1:50:59 at the
1:51:00 end and then we'll
1:51:03 go wrong one
1:51:08 go and then we'll go we go we
1:51:13 [Music]
1:51:20 go all right beautiful and you know what
1:51:23 let's get rid of the fade to blacks
1:51:26 because it'll Loop
1:51:29 nice all right and we could like flip it
1:51:31 and loop it and do fancy [ __ ] but [ __ ]
1:51:34 that let's just we went from because we
1:51:39 played because we're curious because
1:51:41 we're adventurous because we learn just
1:51:43 enough of applications to be able to do
1:51:46 this I don't know iMovie at all I hate
1:51:50 video editing so much that I
1:51:53 founded a company to automate video
1:51:56 editing called storyvine been doing it
1:51:58 for 13 years that's how much I hate
1:52:01 video editing because it's I got add it
1:52:03 drives me [ __ ] crazy so I learned
1:52:05 just enough and so we're g to share this
1:52:08 we're g to download a file we're gon get
1:52:13 it uh we're going to call this uh Mary
1:52:17 go punk
1:52:24 uh where's my desktop let's see Dropbox
1:52:28 where's
1:52:30 desktop I am I
1:52:33 am
1:52:41 confused that's bizarre I'll put it in
1:52:43 downloads
1:52:45 whatever uh okay so now we're going to
1:52:48 go over to
1:52:51 X we're
1:52:53 go on my AI learning
1:52:59 lab
1:53:01 learning
1:53:02 Lab live
1:53:05 live tonight I
1:53:12 made the
1:53:15 marry go
1:53:21 punk and then we should probably put in
1:53:24 some hashtags but I suck at that what
1:53:27 what should I who should I tag oh yeah
1:53:28 tag Luma laabs I'll tag Luma
1:53:31 laabs
1:53:33 um I'll tag Ki who put me in the Creator
1:53:38 Creator program there I'll tag
1:53:42 sunno because it's their damn music and
1:53:45 I'll tag Deco here which I don't think
1:53:48 Deco here has a yeah they do [ __ ] yeah
1:53:51 they do tag all
1:53:53 them and then I'm going to stop sharing
1:53:56 I want you to hear this because we made
1:53:58 this
1:54:00 together um stop sharing that start
1:54:04 sharing
1:54:07 this all right and then I'm G to put our
1:54:10 cute little movie in there Merry Go
1:54:14 Punk
1:54:18 beautiful black tab spelling spelling on
1:54:22 my
1:54:23 AI learning lab Lan Lan tonight it's a
1:54:27 LAN party I made the merry go
1:54:35 [Music]
1:54:43 punk
1:54:45 beautiful 720p recommended 27% uploaded
1:54:48 now we got to wait for my [ __ ]
1:54:51 bandwidth thanks Comcast you
1:54:58 suck I'm a Gen X influencer I'm here
1:55:01 from Comcast this is Xfinity internet at
1:55:05 its finest well finest for Comcast that
1:55:08 [Laughter]
1:55:14 is get Xfinity
1:55:20 today all right now I can post
1:55:26 okay now how does this work how many
1:55:29 people are here I've got 51 people on
1:55:32 Tik Tok 69 people on YouTube and other
1:55:35 channels go to go to x find Kyle Shannon
1:55:40 find my Merry Go Punk and like share it
1:55:43 with the tell tell the world what you
1:55:44 learned here tonight uh I I watched this
1:55:48 thing with this guy that seemed a little
1:55:50 unbalanced uh he made
1:55:55 this it's easy okay
1:56:00 okay all right I'm gonna get out of
1:56:07 here y'all have fun tonight it was
1:56:10 Friday night date night listen I know
1:56:12 you could have been on a date I know you
1:56:14 could have been out finding
1:56:15 love maybe you're married and the Love
1:56:18 is in the closet or you know wherever it
1:56:21 is maybe you're not married and but
1:56:24 you're here we did this we did this we
1:56:27 hung out
1:56:30 together I think if I had one of those I
1:56:33 would be more more more unbalanced than
1:56:36 I already am fun it was it that thing
1:56:38 does look kind of look like Yoda doesn't
1:56:40 it weird Mary from marry go punk exactly
1:56:43 tell the boys to get off Call of Duty
1:56:46 exactly sons of [ __ ]
1:56:54 oh
1:56:57 man where are you doing
1:57:00 this um where am I doing what I'm doing
1:57:04 this in the AI learning lab this is the
1:57:06 AI learning
1:57:09 lab don't go I'm almost
1:57:13 asleep Winston it was fun beautiful all
1:57:17 right everyone um okay so it's Friday
1:57:21 night
1:57:26 if you want some weekend homework with
1:57:35 AI I think it would be this I think it
1:57:38 would
1:57:40 be just do something that gives you Joy
1:57:44 with AI just play with something that is
1:57:47 fun like go make images or pictures or
1:57:52 poems or a recipe like have Chachi
1:57:57 invent a recipe out of some weird
1:57:59 ingredients and see how bad it sucks
1:58:02 like just dare to have some Joy with it
1:58:05 like don't put any pressure on yourself
1:58:08 to have to do
1:58:11 good right like is that marryg go punk
1:58:15 perfect no does it matter
1:58:19 no did I learn something probably
1:58:23 I don't know but I can I can promise you
1:58:26 this two months from now something will
1:58:29 come up and I'll and like merry go punk
1:58:31 will pop in my head and I'll go oh I
1:58:32 know how to do this so just go play just
1:58:35 go play this weekend make something for
1:58:38 your kids make something think of the
1:58:40 person in your life that hates AI the
1:58:42 most and just do something special for
1:58:46 them make him a Advent calendar make him
1:58:49 a
1:58:51 poster you didn't learn how to spell
1:58:53 kopy I did not I failed miserably it's
1:58:55 two L's well there you
1:58:58 go all right I'm going to go look at the
1:59:00 Irregulars before I
1:59:06 go am I sharing everything I'm not so I
1:59:09 have to do it in in this tab
1:59:17 AI
1:59:19 um irregular
1:59:24 nothing new in
1:59:28 Irregulars um but thank you regulars for
1:59:31 hanging out so everyone here that hangs
1:59:32 out if you don't hang out here regularly
1:59:35 I will be back here
1:59:37 Monday hang on I gotta check my schedule
1:59:40 actually I gotta see if shipus is if
1:59:45 I've got a meeting in the way of
1:59:50 it no I don't okay
1:59:54 so I'm just going to put
1:59:58 shipus
2:00:01 save Tuesday oh boy Tuesday what a
2:00:06 [ __ ] mess Tuesday is I won't be doing
2:00:08 it
2:00:09 Tuesday Wednesday I'll do
2:00:13 it Thursday I'll do it
2:00:16 Friday I'll do it okay um so on Tik tock
2:00:22 AI learning laab at AI learning laab on
2:00:25 Tik Tok on Monday at 11:00 am Mountain
2:00:29 Time 10: a.m. Pacific open AI is going
2:00:32 to do their their their third day of 12
2:00:35 days of ship missas and so we'll do that
2:00:38 live and then I'll see you back here on
2:00:41 all these channels at
2:00:43 800m um on TiK ToK by the way we're a
2:00:46 little bit shy of 25,000 likes so if you
2:00:50 want to get
2:00:51 tapping that swell I don't think it does
2:00:54 anything for any of us other than give
2:00:56 you carpal tunnel but that gives me a
2:00:58 certain kind of Joy um especially if you
2:01:00 have your grandkids do it but grandpa I
2:01:04 don't want to tap keep tapping
2:01:07 Sally Kyle needs
2:01:11 likes
2:01:13 [Laughter]
2:01:19 um I don't know why that makes me so
2:01:21 happy
2:01:23 but it does I think it was steo that
2:01:26 that started the
2:01:28 tradition he's down in Australia sitting
2:01:31 in his lawn chair drinking a My Thai at
2:01:34 like 11 o'clock in the morning with an
2:01:36 AI lawn mower mowing his lawn and he's
2:01:39 like I got my granddaughter tapping the
2:01:47 screen I keep hearing 12 days of ship
2:01:50 Miss that's true
2:01:53 well if they don't deliver some good
2:01:55 stuff although you know the thing they
2:01:57 announced today is actually pretty slick
2:01:59 but today was announc Miss it wasn't
2:02:03 shipus anyway um all right well I hope
2:02:06 you had a fun fun Friday night date
2:02:08 night um we got some stuff done we
2:02:10 generated a merry go
2:02:13 punk and uh I don't know waxed poetic
2:02:17 about AI but yeah it was it was a good
2:02:19 night so thank you all for for hanging
2:02:22 out and for those that ask really good
2:02:24 questions again please don't don't ever
2:02:26 feel like asking questions in here is
2:02:29 too much um that's that's why we're here
2:02:32 so um yeah all right [ __ ] Miss is about
2:02:36 right all right thanks Kyle good sharing
2:02:39 thank you very much thank you Tobias
2:02:40 good to see you Tams have a great
2:02:42 weekend that
2:02:47 [Music]
2:02:50 dude Winston Hardy great see you Brandon
2:02:54 thank you mods thank you um Irregulars
2:02:58 thanks so
2:02:59 much um yeah have a fantastic weekend
2:03:03 everyone and I am gonna start winding
2:03:06 down the streams peace out everybody I
2:03:08 hope you have a great
2:03:12 weekend all right