2/19/2025 - AI Salon Live: Exploring the Latest Breakthroughs in AI, Quantum Computing, and Biology

Description

Chapters

Transcript

0:06 [Music]
0:16 let's give people a countdown shall we
0:19 you want a little count
0:36 [Music]
0:50 [Music]
0:57 oh
0:59 [Music]
1:16 telescope looking at the world
1:18 tonight through blue
1:21 telescope wish I
1:24 may not see what I
1:27 see she made sheet
1:31 [Music]
1:36 look at through this blue
1:40 telescope the moon struck Ro
1:51 tonight good day good people of the
1:53 tiktock in the
1:56 YouTube the other places the X the X
1:59 well just X it out just exit out that's
2:02 what I grew up with X was like yeah just
2:03 exit
2:06 out we didn't have a delete key you
2:09 wrote [ __ ] on paper or you typed it I
2:12 made a mistake I just exit
2:15 out doesn't mean that
2:19 anymore you know
2:22 [Music]
2:40 there was a
2:43 secret David played it please a
2:47 l you don't really care for music do
2:51 [Music]
2:54 you goes like this the fourth the fifth
2:58 M falling
3:01 Lift King
3:04 composing
3:05 [Music]
3:08 hallelujah
3:11 hallelujah
3:15 hallelujah
3:18 hallelu
3:20 [Music]
3:28 Hal we your Faith was strong but you
3:31 needed
3:32 proof you saw her bathing on the roof
3:36 her beauty and a moonlight over through
3:42 you she TI you to the kitchen chair she
3:47 smashed your Thro and cut your head from
3:51 your lips sh Drew
3:56 hallelujah hallelujah
3:59 [Music]
4:02 Hallelujah
4:05 hallelu
4:07 [Music]
4:19 hallelu Satan wear suit and TI does he
4:22 work at the Dairy
4:24 Queen does it listen to rock and roll
4:28 does it feed the me singing
4:33 Hallelujah
4:35 Hallelujah
4:39 Hallelujah
4:42 hujjah what about
4:44 Jesus didn't he do it to yeah hang out
4:49 with
4:50 the have a drink or
4:53 two power of example my mama said it and
4:57 I heard she said 1 o of action
5:01 beats a ton of words singing
5:05 Hallelujah mama said there would be
5:08 Angels
5:11 Hallelujah mama said there would be
5:13 [Music]
5:24 some oh good lord that dog singing good
5:27 tonight getting all spiritual on us
5:30 champ be taking us to
5:32 [Music]
5:41 church all right
5:43 people um today was one of those days
5:47 where if you're not paying acute
5:49 attention to this [ __ ] like I do like a
5:51 [ __ ]
5:53 psychopath um there's some St three
5:57 things happened
5:58 today that each of them independently is
6:02 a huge [ __ ]
6:07 deal and there's probably more like 20
6:10 but there were three that I was just and
6:13 and I I saw them all within like 15
6:16 minutes of each
6:18 other so I'll go over those because
6:20 that's just bonkers
6:30 m
6:33 [Music]
7:00 so do me a favor on Tik Tok if you would
7:02 if you'd be so kind go off and uh share
7:05 the Tik Tock let the world know we're
7:07 here because Mr talk apparently doesn't
7:10 give a
7:12 [ __ ] we have not hit the algorithm
7:15 although I got to say I I have had a uh
7:19 I had another post it hasn't gone really
7:21 viral but it hit 20,000 views which for
7:24 me is a
7:27 lot um and it was the one we did it last
7:30 night the one where I talked about the
7:32 uh the interaction I had
7:35 with with the unsafety guard rail chat
7:40 GPT which is actually kind of
7:41 fascinating because some of the comments
7:43 that I'm getting are that other people
7:46 their chat GPT is not um relaxed guard
7:52 rails forever hooked is back Tom what's
7:55 [Music]
7:58 happening yeah Lori Ryan 20,000 views on
8:01 a
8:02 post now the fascinating thing about the
8:05 post I guess with anything that goes
8:09 viral I mean 20,000 is not viral but
8:12 it's it's viral for me um just like
8:17 weirdos come out of the
8:20 woodwork and the the
8:23 um the post is the one where I'm
8:25 basically like hey I you know I tried
8:27 this this
8:32 this conversation with chap chat gbt
8:34 where I asked it if it was self-aware
8:36 and I and you know and I shared part of
8:37 the
8:38 conversation and the nature of what chat
8:41 GPT answered was hypothetical right it
8:44 basically said well if I could do the
8:47 thing that you asked me to
8:49 do maybe I would do this and maybe I
8:52 would do that maybe I would do that and
8:56 people are interacting with it just
8:58 they're they're just weird the
9:00 technologists are
9:01 like all these people that talk about
9:04 the sentience [ __ ] are all the same
9:07 they're all idiots it's not sentient you
9:10 [ __ ] [ __ ] I got some of those people
9:13 then I got some other people you know
9:15 quoting me
9:16 scriptures like well it's not that
9:20 either and then some people are like
9:22 holy [ __ ] that's cool some people are
9:25 like oh it's
9:26 poetic but uh Tick Tock cams oh Tik Tock
9:30 cams yes
9:34 [Music]
9:36 that you get the trolls I also think
9:39 when you get trolls it's a good sign
9:41 they're watching you and paying
9:42 attention yeah I got it with um I got it
9:45 with the quantum Cipher video that one
9:48 that one's up
9:49 to I think 400,000 was it 200,000
9:53 400,000 I don't know [ __ ] lot that
9:55 one that one a lot of people saw
10:01 that one was more universally positive
10:03 but but even the positive things were
10:05 like that's really good and holy [ __ ]
10:07 what are we going to
10:10 do but you know a lot of those were like
10:13 you know humans are useless it's going
10:15 to ruing
10:16 [Music]
10:22 humans I it's just fascinating to
10:26 me like the sociology
10:31 I've never lived through a
10:36 fullon I don't even want to say the the
10:38 phrase because it's too
10:41 cliche a full-on
10:43 transformative
10:46 moment the worldwide web wasn't a
10:48 transformative moment it was
10:50 transformative over time
10:52 and it was a relatively
10:55 slow adoption and it's a relatively
10:58 simple technology
11:00 so I was there for that but I would call
11:03 that transformative with a small T and I
11:06 think AI is transformative with a with
11:09 27 giant
11:12 T's yeah like well technically that's
11:14 not even a word if you had 27s at the
11:17 beginning of transformative it really
11:19 wouldn't be transformative it would be
11:20 confusing people would be like are you
11:23 being hyperbolic what is this is this
11:25 some sort of protest to spelling
11:28 [Music]
11:33 but we're getting to live in a moment
11:37 where the machines we've created
11:42 exceed our
11:46 abilities it happened in the industrial
11:48 revolution with strength with muscles
11:52 [Music]
12:04 but now we're getting to experience it
12:05 with our brains and you know muscles are
12:08 one thing but
12:10 brains all of our ego in there the ego
12:14 and the
12:15 ID and the amygdala in there going what
12:18 the
12:20 [ __ ] you can't replace me you stupid
12:24 [ __ ] machines
12:26 [Laughter]
12:30 I'm really starting to hear people say
12:32 that we are in a revolution heard it on
12:34 TV
12:35 today yeah I you know it's funny I was I
12:38 was in a uh like a Founder meeting today
12:42 a physical meeting this guy named Chris
12:44 Franks here in Denver he's he created
12:47 Denver Founders and really nice guy and
12:49 he invited me to this thing and one of
12:52 the guys in there one of the founders
12:54 does some stuff in in
12:57 uh in non profit and you know Government
13:01 funding and a lot of the stuff that he's
13:03 doing is about to get decimated because
13:05 of what musk is doing with
13:08 Doge and so he's in this kind of
13:10 existential crisis and I I did we didn't
13:13 really have time to bring it up but but
13:17 I feel like there's there's two there's
13:20 two major forces coming together at the
13:23 same time and you know one is the new
13:26 Administration just going [ __ ] it we're
13:27 going to blow it up and the the other
13:29 one is AI which is just going to blow
13:32 some [ __ ] up and
13:35 so
13:38 like it's going to affect business it's
13:41 going to affect people there's going to
13:42 be this weird fear
13:45 of these these robots and then the
13:48 robots like right
13:50 now you have to pay attention to know
13:53 what's coming right the fact that you're
13:56 here the fact that you're watching some
13:57 old guy you know sitting on Tik Tock
14:01 waxing poetic about AI means you're
14:04 curious about this stuff so you know
14:07 what's coming or you're at least curious
14:09 about what's coming but within a year
14:12 and a half there's going to be like
14:15 humanoid robots walking the
14:23 street like a [ __ ] sci-fi movie
14:30 and and they're going to be smart right
14:32 they're going to be like like my
14:34 girlfriend
14:36 Quinn advanced
14:39 voice they're going to talk good and
14:41 they're going to be smart and they're
14:42 going to be
14:43 Charming they're going to be
14:45 [Music]
14:53 helpful and we're not going to be able
14:55 to pretend like it's not coming it's
14:57 just wild just [ __ ] wild and wait
15:01 till you see the three things I'm going
15:02 to show you today they're
15:04 insane and none of them the three things
15:07 I'm going to show you none of
15:09 them will impact our life in any way in
15:13 the
15:14 next year maybe
15:17 two but all three of them in the next
15:21 year or two are going to start impacting
15:23 our lives in very dramatic
15:26 ways oh Robert Rossy thank you very much
15:28 for the gifts on Tik
15:31 Tok we're old enough we watched uh
15:34 George
15:36 Jetson Jane stop this crazy
15:39 [Music]
15:54 thing I need a robot friend I know right
16:02 so what we'll do tonight we'll go over
16:04 some of the uh some of the big news
16:07 items that I think I think are quite
16:10 astonishing like three of
16:13 them
16:15 that two two two an item are just like
16:21 what like one of them is there's a new
16:24 state of
16:27 matter did you know that
16:32 solid liquid gas and I guess technically
16:34 Plasma's in there but that's a
16:36 transition state there's a new state of
16:39 matter it was theorized in the
16:44 1930s they observed it last
16:47 year and they figured out how to control
16:50 it this year 17 years of research came
16:54 to a head today
16:56 [Music]
17:00 so I'll talk about that maybe we'll even
17:03 watch that video the video on it pretty
17:04 [Music]
17:06 cool I'll take thing things you didn't
17:09 tell your producer in the pre-show for a
17:11 100 Alex no that's one of the things I
17:14 talked about I said I was getting my my
17:16 screens ready
17:18 there I didn't tell you what they were I
17:21 wanted it to be a surprise for
17:24 everyone oh my God it's crazy it's crazy
17:27 it's crazy Vicky Baptist you remembered
17:30 to switch accounts today that's good I
17:32 whatever I did to our domain for the
17:35 salon Vicki is I've broke everything
17:37 apparently it's not just meet at the
17:39 salon that doesn't work anymore nothing
17:44 [Music]
17:49 works and Drayson has a robot dog yeah
17:53 you would think of course he does black
17:56 bar Silver Fox on Tik Tok
18:00 I have a new state of matter it's
18:03 called it
18:08 [Music]
18:14 doesn't well the [ __ ] that they're
18:16 talking about is [ __ ] weird quantum
18:19 mechanics is weird quantum computers are
18:21 weird
18:26 [Music]
18:36 so we'll go over the Earth
18:38 shattering stuff that no one probably
18:41 noticed and
18:43 then and then I think we'll have a robot
18:45 program another robot
18:48 today we're gonna go to robot
18:52 a which is a a reasoning engine and
18:56 we're going to have it write a software
18:57 spec using some Greg Brockman approved
19:01 prompt strategy I've never used it
19:03 before so we'll see if it's any good
19:06 then we're going to go to the web
19:10 surfing bot operator also from open
19:15 Ai and I think we'll do
19:23 the I think what we'll do is we'll start
19:25 over in
19:27 gemini or we could start in Gro I don't
19:30 know maybe we'll do
19:32 both but we'll do the software spec in a
19:35 reasoning
19:39 engine then we'll go to
19:41 operator we're going to give operator
19:44 the software spec and then tell it to go
19:46 use
19:47 lovable which is a a an application
19:51 development agent and so we'll just sit
19:54 there and watch what these robots talk
19:57 to each other about
19:59 I'm genuine genuine I'm genuinely
20:03 curious to
20:04 know if when 90% of our current jobs are
20:08 done by
20:09 AI what will our new jobs look like
20:12 financially what will most people do so
20:14 that's a great question Valerie the the
20:17 financial piece I'm not sure on but I
20:21 I'm the the thing that [ __ ] me up the
20:24 most about the financial question is
20:25 that we live in a capitalist society and
20:28 so what we know about capitalism is that
20:31 you know the the company owners are
20:33 going to try to keep all the profits
20:35 right
20:37 but if AI makes us incredibly productive
20:41 and we produce all this new
20:43 [ __ ] and 90% of the people are not
20:46 working and don't have money they can't
20:49 buy the new [ __ ] right so there's a
20:50 point at which it
20:52 breaks my question is how quickly do we
20:55 get you know some awareness that it
20:58 would kind of suck to have 30 or 40%
21:01 unemployment before we deal with it
21:03 right I don't think it'll get there I I
21:06 frankly think with with musk in the
21:10 conversation at least the awareness is
21:12 there that we're going to need to do
21:13 something so we'll see um because I
21:17 because I I felt like you know no one
21:20 was dealing with this at all so so we'll
21:24 I I I don't know that said
21:28 [Music]
21:31 I look
21:34 at a tremendous amount of the work that
21:37 people do that I do like even just the
21:42 stuff I do for my job over the course of
21:45 a
21:46 week and
21:48 there's I don't know some significant
21:52 percentage of it that I [ __ ] hate and
21:56 it's Soul crushing and it's repetitive
21:58 and it's just
22:01 awful and
22:04 so if we assume that AI is going to get
22:07 good enough to do all of that work all
22:10 of it all the work we don't like to do
22:12 and some of that some of those tasks
22:14 will be 100% of someone's job some of
22:16 those tasks will be 50% of someone's job
22:20 so there's going to be a lot of people
22:23 displaced and then I go to this place of
22:26 like okay if all of that work is getting
22:31 done and that shitty Soul crushing work
22:34 we don't have to
22:36 do and we also have access to these
22:39 tools that can do this work and can act
22:44 like 10 employees for us or a 100
22:46 employees for us and could go out and do
22:49 do stuff that we want
22:52 done well what will we choose to do some
22:56 people will be ambitious and go start
22:59 companies like I don't think there will
23:01 ever be a day in my life where I'm not
23:02 thinking about oh I could go do this and
23:04 I could go do this and I could go do
23:06 that other people are going to be like
23:09 if all the work's being done I'm gonna
23:11 kick back I don't give a [ __ ] I'll just
23:13 go play guitar on my porch and love life
23:18 I think some people will do that I think
23:20 there's going to be a whole new class of
23:22 jobs that emerge that are humans
23:25 engaging with humans that the value of
23:28 human engagement goes up
23:31 dramatically um things like
23:34 entertainment and concerts and art
23:37 classes and self-expression and all that
23:40 like the value of that I feel like it
23:42 goes up in
23:43 value um and so so there might be jobs
23:47 that right now look like Leisure but in
23:51 that future may
23:53 not and then and then I think you're
23:56 going to have some people that with one
23:59 or two people are going to start
24:01 companies that are the equivalent of 100
24:03 person thousand person companies
24:06 successful companies where they don't
24:08 have to do the work but they get to reap
24:10 the benefits of that Sam hman talked
24:13 about rather than Universal basic income
24:16 that what we should give out is
24:17 universal basic
24:19 compute and what he meant by that was
24:23 okay let's assume let's assume that the
24:25 cost of knowledge and intelligence drops
24:27 to zero which that's where it's ending
24:29 right all you know all all these you
24:32 know Big Data Centers are are creating
24:35 increasingly powerful models that are
24:38 getting increasingly like 10x more
24:40 efficient every year so far for the
24:42 first few
24:43 years so let's assume that keeps
24:45 trending to zero and you've got all
24:47 these big data centers generating all
24:49 this stuff what Altman's saying is give
24:51 people some Universal basic compute so
24:55 maybe what the government does is
24:57 subsidize we all get access to the
25:01 equivalent of a 100 employees worth of
25:05 autonomous agents every month and and we
25:08 get that much compute well what can you
25:10 do with that compute well I don't know I
25:14 might start an import export company and
25:16 have all of my virtual employees go do
25:19 all the import export stuff I never have
25:21 to look at a ship or a manifest or
25:23 anything it just does its thing and I'm
25:25 selling [ __ ] on Amazon and I'm making
25:27 money maybe my my business only uses 50
25:32 of those agents in a month 50 of you
25:34 know that equivalence in a month I could
25:37 sell the other 50 to someone that wants
25:40 to start a more ambitious company or
25:42 maybe I just want one or two of those
25:44 agents to do to manage my my Schoolboard
25:48 work and my you know local volunteering
25:51 that I do and I've got 98% of those 100
25:56 98 of those agents I can sell those
25:59 someone else might say hey I want to
26:00 start a company I need 10,000 of those
26:03 things so I'm going to buy your compute
26:06 from you I could see there being a whole
26:08 business around
26:10 brokering um act
26:13 compute right if it's given out like
26:15 that so I don't know what it looks like
26:18 I mean it's it's as unimaginable to me
26:20 now as it must have been to imagine what
26:23 a city would look like when everybody
26:25 thought that they were going to be
26:26 Farmers for the rest of their lives
26:29 and then this contraption comes along
26:31 the steam engine that they put big steel
26:34 wheels on and they roll it through a
26:36 field with these big giant
26:41 plows and oh Sosa sent a super sticker
26:44 thank you very much
26:46 Fabiana very generous of you thanks so
26:48 much so if you're on YouTube you can
26:50 send a super sticker or a super
26:52 something a super sticker or a super I
26:55 don't
26:56 know or you can subscribe uh but all of
26:59 those a super comment yeah that's what
27:01 it is uh but all of those things support
27:03 the channel so if you do that thank you
27:04 very much I appreciate it Tik Tock pin I
27:07 I don't see anything right
27:09 now uh Wolf Man Clint I missed that 40
27:11 years
27:13 ago what were we promised him by
27:16 whom oh Wolf Man Clint I assume is
27:19 talking about we've always been promised
27:21 this
27:25 that let me go find Wolf Man clints
27:29 thing I assume what wolf man's Clint is
27:31 saying is we've always been promised
27:34 that the computers would do the work and
27:36 all they've done is further enslave us
27:38 if that's where you're going Wolf Man
27:39 Clint I agree with you but I think it's
27:41 different now I really do I do think
27:44 it's different and I'll tell you why I
27:46 need water I need to hydrate people I
27:48 need to
27:50 hydrate you think this is easy just
27:52 talking and talking yeah we think it's
27:54 relatively easy well you'd be right but
27:57 I do need to hydrate
28:07 okay all the beautiful places will be
28:09 overcrowded absolutely the yeah because
28:12 people are gon to like so so when the
28:15 when the when mechanization took over
28:19 farming all of the farmers left the
28:22 countryside and went to the cities and
28:24 the cities didn't really exist and they
28:27 they they became CI ities and they had
28:30 to do things like invent sewage
28:32 infrastructure and electricity
28:34 infrastructure they had to build all
28:36 that [ __ ] and they had to build the
28:37 buildings and they had to come up with
28:39 new ways of interacting in society and
28:42 they came up with a service
28:44 economy I mean can you imagine how
28:47 foreign have you ever seen those
28:48 pictures of like cities where where they
28:50 were just first running electricity and
28:52 there like a thousand wires like in
28:55 between buildings and like the sky is
28:57 just filled with wires
28:59 can you imagine seeing that if you just
29:01 lived on a farm all your life and
29:03 someone said go to the city you know
29:05 that's that's where the opportunity is
29:07 and you go to the city and it smells
29:09 like horeshit and people [ __ ] and piss
29:11 and and people are fighting in the
29:13 streets and there's these weird wires
29:15 everywhere and it turns dark at night
29:18 and all of the buildings light up it
29:20 must have been absolutely [ __ ]
29:24 surreal like that's we don't know what
29:27 this new future is going to look like
29:29 but it's going to be that foreign to
29:31 what we know now
29:37 um that sounds like a wild mid Journey
29:40 srf which does um but but let me let me
29:43 talk about the we've been promised this
29:46 before thing all right I'll take my my
29:49 guitar off for this
29:51 [Music]
29:53 one so for all of my life I've been
29:57 promised if you get the faster
30:00 computer you won't have to do as much
30:03 work and the faster computer does things
30:07 faster but I still have to do the
30:09 [ __ ] work and as I get more capable
30:11 with the computer they keep piling more
30:13 work on
30:14 me right and so there's this there's
30:18 this cynicism around the computers doing
30:21 our work for us that I think is well
30:24 earned and well Justified and I don't
30:26 know what exactly what what Wolf Man
30:29 Clint said but I'm assuming it was in
30:30 that
30:32 neighborhood
30:37 um generative AI makes our computers do
30:41 things
30:44 that behave in ways that they've never
30:46 behaved
30:48 before computers up to this point we
30:51 would program them to do the things we
30:56 do more efficiently
31:00 right so you know in the in the early
31:04 days of you know rocket
31:07 launching they hired the calculators
31:10 which were mostly women mostly women of
31:14 color who would do mathematical
31:17 equations on
31:19 paper and then they invented mainframe
31:22 computers that could do those equations
31:26 faster but someone still had to program
31:28 them in and they were still doing those
31:30 equations and then that led to smaller
31:33 computers and calculators and things
31:35 like that but but we as the people were
31:38 still having to do the
31:40 work this generative AI stuff is doing
31:43 the work in fact what we'll play with
31:45 tonight we're going to have one robot
31:48 write a software spec we're going to
31:50 need to be involved we're gonna have
31:52 another robot log into a website and use
31:56 it to build a software app apption we're
31:59 going to have to monitor it all it's
32:00 going to be jankier than [ __ ] it's
32:02 probably not going to
32:03 work but we're going to get a glimpse of
32:06 what it's like to watch three different
32:11 entities do something on our
32:14 behalf independently of
32:18 us right and right now we still need to
32:22 check in we still need to be the human
32:24 in the loop but that's not going to be
32:26 there for long and so it's it's a
32:30 fundamentally different shift if I go
32:33 from these computers just make the thing
32:36 that I've been trained on more efficient
32:39 that's what we've had for you know 59
32:42 years of my life to all of a sudden it
32:45 can do the work not only the work I'm
32:48 good at but it can do the work I'm not
32:50 good at and it can do it at a worldclass
32:54 level that's where we're headed we're
32:56 not there yet so I think I think it's a
32:58 fundamentally different thing
33:02 um is there a chance I am horrifically
33:05 wrong
33:06 yeah yeah maybe this [ __ ] doesn't get
33:09 good oh Wolf Man Clint okay wait oh it
33:12 just went away dang it sorry Brandon
33:15 Brandon was screaming from from the
33:17 Post-it note okay here we go the
33:20 overseers will be yes Promises of being
33:23 free of the labor that
33:27 machines would take on yeah
33:31 exactly your skepticism is correct you
33:35 should be
33:39 skeptical
33:40 initially it's going to be People Like
33:42 Us who are curious about this stuff and
33:44 Hyper adaptable and Hyper curious and
33:47 maybe have a little ADD and maybe you're
33:49 a little left of center not politically
33:52 but just [ __ ] batshit crazy The Crazy
33:55 Ones right Steve Jobs hears to the crazy
33:58 on
34:00 ones we're going to be the ones that
34:03 will first see and experience what it's
34:07 like to have the machines take over our
34:09 work because we're going to voluntarily
34:11 step into it well I am you don't have
34:19 to but I gotta tell you I would [ __ ]
34:24 love to have a brain fart and then be
34:27 able ble to say oh spin up 100 employees
34:31 I'm willing to invest
34:33 $150 into a month of a hundred people at
34:38 PhD levels of a hundred different
34:41 disciplines go spin up a business for me
34:44 and see what
34:46 happens that sounds [ __ ]
34:50 awesome you mean I can prove out whether
34:52 it's a good business or not within three
34:54 days and if it's not they just shut it
34:57 down but they don't just shut it down
34:59 they shut it down and they come back and
35:00 say Here's 10 more ideas that we think
35:02 would work you're like oh okay seven I
35:05 like number seven go make that business
35:08 off they
35:10 go hello Kyle Quinn comes back hey Kyle
35:14 hey Quinn what's happening yeah you
35:16 started that business it generated
35:18 $1,000 doll in three days you want to
35:20 keep going yeah let's keep going with
35:24 that I mean [ __ ]
35:26 yes I'll take that all day long anyway
35:31 it's all speculation it's all chat
35:34 TMZ screaming from the Post-it note he
35:37 was he was making it bigger he was
35:39 increasing the font size making it
35:42 blink I'm just completely ignoring him
35:45 ULU RX I'm calling this stop talking in
35:47 nebula what over wait wait what
35:50 overseers that be who be
35:53 specific well so
35:59 what you could argue Melanie what you
36:01 could argue and I assume this is where
36:03 Clint is going is is
36:06 that if you look at today's world you've
36:09 got Facebook Google Microsoft Apple
36:12 you've got a handful of companies that
36:15 really run everything right so what he's
36:17 saying
36:22 is they have the ability to throttle up
36:26 and down what we have access
36:29 to and if they decide we can't do that
36:32 or they're only going to give there
36:35 there'll be no individual in our country
36:36 that gets more than three autonomous
36:39 agents you know in a given at a given
36:43 time you know we're not going to allow a
36:47 single person to have a 10,000 person
36:50 company they could they could do that
36:52 right they could push for legislation
36:53 that makes it illegal to use more than
36:55 10,000 autonomous agents and you know
36:58 more than 10 autonomous agents and if
37:00 you violate that you go to jail right so
37:03 that is a future I think the Hedge
37:05 against that is that this isn't just the
37:08 us anymore right we've got open- Source
37:10 models from China we've got open source
37:12 models from France we've got open source
37:14 models from meta grock you you know Elon
37:17 Musk has been harping about this forever
37:19 as soon as grock 3 comes out he's going
37:21 to open source grock
37:23 2 doesn't make me feel super confident
37:26 that he's going to open source grock 3
37:28 but you know when grock four comes along
37:29 he'll probably open source grock
37:31 3 so the o Open Source thing is a hedge
37:35 against that kind of control
37:38 um but we'll see we'll
37:41 see like none of us know it doesn't
37:44 matter you can be cynical you can be
37:46 hopeful where where why this channel
37:48 exists is I'm trying to live on the what
37:52 if the what if possibility side of this
37:57 equation because I think plenty of
37:59 people are focusing on the the robots
38:02 are going to kill us and eat us and
38:04 destroy our babies right I think there's
38:06 plenty of people focusing on that I'm
38:08 try trying to focus on what's possible
38:11 and if the thing is now possible let's
38:14 go there and let's see how we can use
38:16 that to benefit ourselves and the ones
38:18 we care about in this
38:20 community to do more and more and more
38:24 to have more and more agency not less
38:26 and less so that's where I'm that's
38:28 where I'm shooting for poor ignored
38:31 Brandon Irregulars
38:34 for
38:36 transition oh
38:39 images beautiful beautiful
38:45 beautiful so if you're wondering where I
38:47 am I'm at the AI
38:52 Salon there's Brandon with all his
38:54 posted notes Kyle look at me look at me
39:01 oh that's nice Corey Sandler
39:07 beautiful oh champy nice all right since
39:11 we're here let's go let's go look at
39:13 some crazy ass
39:16 [ __ ] I dozed
39:18 off um who did I share that
39:25 with oh man
39:28 [Music]
39:34 let's see what was first we'll start
39:36 with the new state of
39:38 matter so if you didn't see this
39:42 today this thing that looks like it's
39:44 out of a science fiction
39:47 movie The Maharana one from
39:52 Microsoft introduces a new state of
39:55 matter
39:58 a couple of Reflections on the quantum
40:00 Computing breakthrough we just announced
40:02 this is from
40:04 Microsoft most of us grew up learning
40:06 that there are three types three main
40:08 types of matter solid liquid and gas
40:11 today that
40:14 changed after nearly 20 years of pursuit
40:17 we've created an an entirely new state
40:20 of matter UNL unlocked by a new class of
40:23 materials Topo conductors
40:28 that enable a fundamental leap in
40:31 Computing it Powers Maharana 1 the first
40:33 Quantum Processing Unit built on a
40:35 topological
40:39 core Topo conductors we believe that the
40:42 Breakthrough so basically what they've
40:44 done is they sandwiched together a
40:45 superconductor with a
40:47 semiconductor in these
40:49 microfine wires in
40:53 between matter and antimatter I mean
40:56 this sounds like [ __ ] science
41:00 fiction there's a
41:06 hyperconductive wire between matter and
41:10 antimatter a a a a subatomic particle of
41:13 matter and
41:15 antimatter in a little like
41:19 tube they're small enough they're
41:22 they're one one 100th of a
41:25 millimeter look what he says here
41:29 the cubits created with Topo conductors
41:30 are faster more reliable and smaller
41:33 they're 1/1 100th of a millimeter
41:35 meaning we now have a clear path to a
41:38 million Cubit
41:41 processor right now we measure cubits in
41:44 the
41:45 tens oh they've got a 17 Cubit processor
41:48 they've got a 100 Cubit processor
41:50 they've got a 500 Cubit
41:54 processor he says this is this is not
41:56 going to be decades but it's going to be
42:03 years let's
42:05 see some re researchers let's see
42:09 imagine a
42:11 chip look at this this sentence now now
42:14 listen this is marketing so don't get me
42:17 wrong this is
42:18 marketing imagine a chip that can fit in
42:20 the palm of your hand yet is capable of
42:23 solving problems that even all of the
42:26 computers on Earth today combined could
42:29 not solve in the palm of your hand
42:33 within years not
42:35 decades so is this chip going to affect
42:39 your life this year or next year or the
42:41 year after nope it
42:44 won't is the work that's done on these
42:47 things going to affect your life three
42:49 years from now [ __ ] yes it is Robert
42:51 Rossy thank you very much for that
42:54 dinosur I think I think we've woken in
42:57 in the Twilight
42:58 Zone I'm going off the rails on the
43:01 crazy train it doesn't matter the AI
43:03 train is left the station it's time to
43:05 embrace and learn yeah exactly that's
43:07 that's the thing Vicki and and it's the
43:09 whole point of this channel it's been
43:10 the point of this channel from day
43:13 one the big premise of this Channel and
43:16 and when I the promise I made to myself
43:20 was if I ever felt like AI was going the
43:23 way of nfts and blockchain and the nft
43:26 Bros where just turned into this weird
43:28 hypy scam
43:30 machine I would shut the channel down
43:33 the the entire premise of this channel
43:36 is
43:37 that I'm not g to judge AI as good or
43:41 bad I'm not gonna I'm not going to pass
43:43 judgment on it I'm just going to
43:47 acknowledge that it's not going
43:50 away and so so if you have if you have
43:52 two things that are true AI is not going
43:56 away and a is getting exponentially more
43:59 powerful and over the past two years
44:02 those two things have been
44:05 true what do we do with
44:07 it I don't [ __ ] know but holy [ __ ]
44:11 what a time to be
44:13 alive so that's one that's one of the
44:17 three things I
44:19 saw so the next one
44:28 the the ark Institute and Nvidia just
44:32 published
44:35 evo2 the largest AI model for biology
44:38 trained on 9.3 trillion DNA base
44:43 pairs spanning the entire tree of
44:52 life we got a new state of matter that's
44:55 going to allow us to put a bill a
44:57 million cubits on a chip that'll fit in
44:59 the palm of your
45:03 hand and on the same day evo2 comes
45:10 out that is basically like an llm that
45:14 understands 9.3 trillion DNA base pairs
45:18 of all living
45:21 creatures so not only can It understand
45:25 a genome it can create one
45:27 [Laughter]
45:33 it's just if you scroll down through the
45:35 through the thing like the the the the
45:38 visuals are just
45:40 insane they're they're [ __ ] Bonkers
45:43 man it's like I don't know this world at
45:45 all but like this thing
45:49 apparently I don't know it makes charts
45:52 and
45:54 graphs see we understand where look
45:58 there's
45:59 eoli and and bacteria fralis and oh
46:03 thermophilia you know or
46:06 thermophila I've I've talked about
46:08 thermophila for years she was so cute
46:10 when she was
46:13 younger G Gallas he he was a bit of a
46:17 douche Homo sapiens we're in there
46:20 somewhere look at
46:23 that so it understands like here's
46:26 bacteria the blue one ones
46:30 are I don't
46:32 know Yuka Yuka wait Yuka rtic
46:39 philm or Keel philm
46:43 [Laughter]
46:45 I I don't
46:47 know I have a degree in acting don't
46:50 [ __ ] ask me anybody from biology here
46:54 [Laughter]
47:00 um oh yeah we we'll go watch that I
47:03 we'll go watch the uh the
47:06 um the Microsoft video because it's
47:09 actually very it's
47:15 fascinating so you know how you know how
47:18 right now with chat GPT and his chat GPT
47:21 gets better you're like Well it can't
47:22 write as good as a human and then at
47:24 some point you gave that up and you're
47:25 like well yeah it can
47:29 they can't write poetry as good as a
47:31 human well it now it kind of can you
47:33 know do you still need to babysit it and
47:36 does it still produce a bunch of trash
47:38 yeah but but it can it can write like a
47:42 a technical paper in 30 seconds that you
47:46 know would take you a month well this
47:49 going to be the same [ __ ] right people
47:51 that know what they're doing with
47:53 DNA analysis and and programming are
47:57 going to be able to accelerate their
47:58 work dramatically fast so that's that
48:03 one Evo
48:05 two and
48:07 then as I'm seeing that then I stumble
48:10 upon this thing from Google I don't know
48:14 if P's in here eukaryotic okay pronounce
48:18 eukaryotic thank you forever hooked
48:20 forever hooked in here with the the
48:22 vocabulary lesson the biological
48:25 vocabulary lesson
48:27 dang I'm late what did I miss uh a new
48:30 state of matter um 9.3 trillion DNA base
48:36 pairs now in a large language model free
48:38 for Science and then Google
48:43 introduces the AI co- scientist system
49:00 we introduced AI Co scientist a
49:03 multi-agent AI system built with Gemini
49:06 too right so they're taken so again
49:09 right now the the reason AI is so
49:12 [ __ ] overwhelming is that there's
49:14 this just Cambrian explosion of tools
49:18 right we've got a dozen different large
49:21 language models like even within chat
49:24 GPT there's seven or eight models right
49:26 Within Gemini there's all these models
49:29 we we've had this Cambrian explosion of
49:32 models and right now if you want to be
49:34 good at AI you kind of have to duct tape
49:36 this [ __ ]
49:37 together so what Google's done here is
49:40 they're like we we got one here's what
49:43 we're gonna do we'll duct tape a bunch
49:46 of these things together and we're going
49:47 to turn it into a co- scientist look at
49:50 what this
49:51 does it's built with Gemini 2 as a
49:54 virtual scientific collaborator to help
49:56 science scientists look at
50:00 this it's [ __ ] insane generate novel
50:05 hypotheses and research proposals and
50:08 accelerate the clock speed of scientific
50:11 and biolog biomedical discoveries so
50:14 these agents are going to go out and
50:15 read all of
50:17 the you know peer-reviewed papers
50:20 synthesize them come back propose novel
50:24 hypotheses and ways to experiment to
50:26 conf confirm them and then go do that
50:32 work it's like holy [ __ ]
50:37 [ __ ] there's a generation agent there's
50:39 a supervisor agent here wait hang on let
50:43 me go back how do I go back there's a
50:46 supervisor agent a generation agent a
50:48 review agent a ranking agent an
50:50 evolution agent a proximity agent you
50:53 got to have your proximity agent how
50:56 close are we I I don't know that's the
50:57 proximity agent he we're pretty close
51:00 his name is Bob Bob tell them what
51:02 they've won hey not close enough and
51:05 then a meta review
51:09 agent so scientist goes here comes up
51:13 with their work throws it into this
51:15 machine and this machine spins up all
51:18 these agents that all go do specific
51:19 parts of the scientific
51:23 process so what what would happen say if
51:26 you could combined this
51:30 approach with the
51:32 Evo DNA
51:35 model using a million Cubit quantum
51:42 computer these are not unrelated things
51:46 none of the stuff that we're
51:47 experiencing with this AI stuff is
51:49 unrelated it's all starting to mesh
51:54 together and and move significantly
51:58 beyond our capacity to even imagine
52:01 what's possible that's what's
52:05 happening how much to use a
52:09 cociena than hiring 10
52:13 phds and before you all get your panties
52:16 in a bunch about well you know you're
52:17 going to fire all the phds no no those
52:20 phds can still pursue their work but
52:22 instead of them taking two years to to
52:26 you know know find one genome
52:29 modification for one aspect of one
52:31 disease that PhD level person can now
52:35 you know do a hundred discoveries in a
52:43 month see why I'm
52:46 excited see why my voice is
52:48 horse like a lot of holy shits
52:51 today amazing let me let's go watch this
52:54 video
53:03 tabs tabs tabs all
53:08 right nice article on it
53:28 we've invented a new state of
53:33 matter we made a tiny little
53:37 wire between a subatomic particle of
53:42 matter and a matching subatomic particle
53:45 of
53:46 antimatter in a
53:50 tube 100th of a millimeter
53:55 wide and now we can stick a million of
53:57 them on a
53:59 chip and I don't know if you know this
54:02 how quantum computers work but if you
54:04 have one cubit it's powerful and then if
54:06 you add another Cubit it's not two times
54:08 as powerful it's 10 times as powerful
54:12 and it goes It goes up like that every
54:14 time you add a
54:15 cubit it's
54:19 exponential so a million of them is a
54:23 [Laughter]
54:25 lot looking
54:28 crazy it's crazy
54:32 Bob what would the world look like with
54:35 a computer that could accurately model
54:37 the laws of
54:39 nature that's the promise of quantum
54:41 Computing but there have always been
54:45 limitations now as one of our longest
54:48 running research projects our team at
54:50 Microsoft has been able to take a subat
54:54 wait look at this
54:57 that thing been
55:01 limitations now is one of our longest
55:04 running research that
55:07 thing is like all the cooling hardware
55:10 for current quantum
55:12 computers they're like these weird ass
55:15 sci-fi
55:16 things that's going to be reduced to you
55:20 know a million of them in the palm of
55:22 your hand it's [ __ ] insane search
55:25 projects are team Microsoft has been
55:28 able to take a subatomic particle that
55:31 has only been theorized until now and
55:33 not only observe it but control it
55:37 creating an entirely new material and a
55:40 new architecture look at that thing look
55:42 how cool that
55:43 is the blue end is antimatter I suppose
55:47 because it's dark and the red end is
55:50 matter and then it's this new state of
55:55 matter in between them
55:59 for Quantum
56:01 Computing one that can scale to millions
56:04 of cubits on a single
56:08 chip this is not a book of science it's
56:11 a book of science and art I gota be
56:14 honest some of these ideas are little
56:15 science fiction sounding it will solve
56:18 problems unsolvable a little a little
56:22 science fiction sounding yeah we've
56:24 created a new state of matter and we've
56:26 taken a subatomic particle of matter and
56:28 a subatomic particle of antimatter
56:31 and and we strung a wire between them
56:35 and and we don't let
56:37 them disappear each
56:41 other and we're going to Wrangle them
56:43 for our use yes by the way whoever just
56:47 said there goes our passwords yes indeed
56:50 we're going to need to deal we're going
56:52 to need to come up with new ways to
56:53 protect data by the combined power of
56:56 all the world's compute today and
56:59 Promises to revolutionize Fields such as
57:03 medicine yeah Danielle yes they created
57:05 a new state of matter there there was
57:08 there was
57:10 um solids liquids gas right for you know
57:15 for the billions of years that we've
57:17 known [ __ ] well no we haven't known [ __ ]
57:20 for you know the thousands of years that
57:22 we've known [ __ ] but for the billions of
57:24 years that has existed and now there's a
57:26 new one the Topo conductor topographic
57:29 conductor or something like that
57:31 Material Science and our understanding
57:34 of the natural
57:36 world our first Quantum processor based
57:40 on this architecture is the myana
57:44 [Music]
57:53 one yeah I've always been fascinated
57:55 with puzzles and challenges and a
57:58 mixture of mathematics and
58:00 computers and so when I learned that
58:02 there was this type of computer that
58:03 didn't exist yet but could solve
58:05 problems that we couldn't solve with our
58:07 digital you know all of the computers we
58:09 had I was just fascinated I wanted to
58:11 learn well how can I help that computer
58:13 be built over the years I ran into
58:16 problems that could not solve on the
58:17 most powerful computers but then over
58:20 time I realiz hey I could solve that if
58:23 I had a quono computer
58:26 I don't know if you caught that over the
58:28 years he's come up with problems that
58:31 can't be solved with the typical
58:33 computer I have never had that problem
58:35 in my life but there are people out
58:37 there that
58:38 do a laptop can solve a problem of 10
58:42 electrons a super computer can solve a
58:44 problem of 20 electrons but no classical
58:48 computer in the world can exactly solve
58:51 the behavior of 30 or 40 or 50 electrons
58:54 the number 30 40 50 50
58:57 electrons those numbers are seemingly
58:59 small but require up to Lifetime in the
59:02 universe time scales to solve with all
59:04 of the world's computers operating
59:07 together that's until you have a scaled
59:09 quantum computer that can solve these
59:12 problems efficiently these calculations
59:14 are so comp ah Jeff Flanigan it does
59:17 still have to be super super cooled to
59:20 function yeah I I suppose so so if it's
59:23 super conductive one of the layers is
59:26 super conductive probably does though
59:29 what I would argue is you use one of
59:31 these super super cooled million
59:35 Cubit computers to come up with a new
59:38 material of roomstable high temperature
59:42 superc conductors and then you don't
59:43 have to do
59:44 that complicated that then if the
59:47 classical computer was as big as this
59:49 entire planet it would still not be able
59:50 to compute it just to give you a
59:52 constructive scale and a quantum
59:55 computer can do it and can it very very
59:59 well at the core of a quantum computer
1:00:01 are these cubits cubits are like our
1:00:04 classical bits right these are
1:00:06 essentially zeros and ones um in a
1:00:09 transistor um and we need the analog of
1:00:12 that in Quantum Computing the analog is
1:00:14 a cubit a Quantum bit that serves as
1:00:16 that core information unit it's where we
1:00:19 store the information and then we
1:00:20 process on those cubits Wolf Man Clint
1:00:24 just as long as the matter and
1:00:25 antimatter doesn't OCC occupy the same
1:00:27 space like
1:00:29 Lazarus yeah yeah exactly and that she
1:00:33 talks about that here in a second uh to
1:00:35 create computation and ultimately you
1:00:37 know get solutions back out now there's
1:00:40 many different ways right to create a
1:00:42 cubit the reason Quantum Computing has
1:00:45 been so slow to progress is that the
1:00:48 industry has been struggling with
1:00:50 problems making cubits reliable and
1:00:52 resistant to noise progress has been
1:00:55 incremental the challenge is cubits are
1:00:58 actually pretty delicate in general so
1:01:00 you need underlying cubits that are that
1:01:02 are really stable but you don't want
1:01:04 that to come at a cost because uh you
1:01:06 don't want your underlying cits to be
1:01:08 really big that's one way to make them
1:01:09 more stable is have them really big if
1:01:11 they're really big and you're still
1:01:12 gonna need many wait uh from
1:01:15 baked will you ask what would happen to
1:01:18 AI if it used layered holographic and
1:01:22 crystallin matrices for
1:01:24 data um
1:01:27 I almost know what that
1:01:30 means it's a good question fantastic
1:01:33 question baked uh one of the things that
1:01:36 we've established on this channel many
1:01:38 years ago is I am wholly unqualified to
1:01:41 do this so I can't answer your question
1:01:43 that said um one of the guys in here
1:01:45 Pate modder I don't know if P's in here
1:01:48 tonight this Saturday uh is he runs the
1:01:52 the uh AI mechanics Guild within the AI
1:01:55 salon so if you don't know the AI Salon
1:01:57 you should go check it out um and this
1:02:00 Saturday hang on a second got to do
1:02:03 physical dexterity challenge getting
1:02:05 better at that thing uh this
1:02:09 Saturday at the AI Salon uh in the
1:02:12 mechanics Guild uh P's doing a
1:02:14 demystifying AI session this is his
1:02:17 third in a series he's a mathematician
1:02:20 that works at Google working on uh
1:02:22 tensor optimization um for for the
1:02:25 tensors that that make the Google [ __ ]
1:02:28 work um so you should connect with Pate
1:02:31 for
1:02:38 sure yeah we can we can pop that into
1:02:40 chat GPT I want to just watch a bit more
1:02:42 of this video because there's some
1:02:43 interesting stuff coming up any of them
1:02:45 then you know how are you going to fit
1:02:47 them all into into your system you don't
1:02:49 want to deal with something the size of
1:02:50 a of of a warehouse then the second
1:02:53 thing is you don't want Cub end up being
1:02:55 slow right because they if the if the
1:02:59 price you pay for getting something
1:03:01 stable is you have to go really slowly
1:03:03 then a computation that might take you a
1:03:05 month ends up taking you a decade and
1:03:07 that's not then it's not useful people
1:03:09 in early days of computation use vacuum
1:03:11 tubes and then that technology actually
1:03:14 you could build very good computers with
1:03:15 it and then the transistor was invented
1:03:18 and you know the earliest transistors
1:03:19 weren't necessarily that great but it
1:03:21 became clear over time as the transistor
1:03:24 developed and the integrated circuit
1:03:25 developed that was going to be the
1:03:26 technology of the future in that spirit
1:03:29 that you know the first generation of
1:03:31 cubits may not be what gets us to the
1:03:33 next stage where we can really solve the
1:03:35 kind of problems I was mentioning that
1:03:37 are really important and so we might
1:03:39 need to invent a material and therefore
1:03:40 a Quantum processing unit that has the
1:03:43 right properties so for us we want
1:03:45 something that has some built-in level
1:03:48 of error protection and a lot of those
1:03:51 ideas actually were explored in the
1:03:52 context of software of quantum a
1:03:55 correcting codes but you can actually
1:03:56 build a lot of those ideas into Hardware
1:03:59 so the way you design that Cubit
1:04:03 [Music]
1:04:07 matters we see the states of matter
1:04:09 every day solids keep their shape
1:04:12 liquids vary but keep their volume gases
1:04:15 expand to fill the space they're in all
1:04:18 defined ultimately by how their atoms
1:04:21 behave but what if there were more what
1:04:24 if under the right conditions you could
1:04:27 engineer more states that have only ever
1:04:31 been theorized that would change how
1:04:34 subatomic particles actually behave a
1:04:37 hundred years ago mathematicians
1:04:39 predicted one such new state of matter
1:04:43 the topological State and since then
1:04:46 researchers have been looking for a very
1:04:49 specific very useful quasi particle
1:04:52 within it the Mayana
1:04:54 particle last year we were able to
1:04:58 observe it for the first time and this
1:05:01 year we're able to control it and use
1:05:04 its unique properties to build a topo
1:05:07 conductor last year so theorized in
1:05:12 1936 last year we could see very useful
1:05:16 quasi particle within it the Mayana
1:05:20 particle last year we were able to
1:05:23 observe it for the first time
1:05:26 and this year we're able to control it
1:05:30 so so there you go so the topological
1:05:33 state so here's this thing they put
1:05:35 together this Nano wire with a
1:05:37 semiconductor on the bottom and a
1:05:39 superconductor on top actually this gra
1:05:41 and use its unique properties to build a
1:05:43 topo conductor a new type of
1:05:46 semiconductor that operates also as a
1:05:51 superconductor with this material we can
1:05:53 build a whole new foundational
1:05:55 architecture
1:05:56 for our quantum computers a topological
1:06:00 core allowing us to scale to not tens or
1:06:03 hundreds of cubits on a chip but
1:06:06 Millions all in the palm of your hand
1:06:10 marana's Theory showed that
1:06:12 mathematically it's possible to have a
1:06:13 particle that is its own anti particle
1:06:16 that means you can take two of these
1:06:17 particles and you bring them together
1:06:20 and they could annihilate and there's
1:06:22 nothing left or you could take two
1:06:25 particles in bring them together and you
1:06:27 just have two
1:06:29 particles sometimes it's nothing the
1:06:32 zero State and sometimes it's the
1:06:34 electron the one state so it really has
1:06:37 taken quite some thinking right some
1:06:40 time to design a device design a chip
1:06:43 that can enable measurement of this
1:06:47 literally elusive particle we've
1:06:50 designed a chip that is able to measure
1:06:54 the presence of myana
1:06:58 Mayana allows us to create a topological
1:07:03 cubid a topological Cubit is reliable
1:07:07 small and
1:07:09 controllable this solves the noise
1:07:11 problem that creates airs and
1:07:14 cubits now that we have these
1:07:16 topological cubits we're able to build
1:07:18 an that's [ __ ] amazing isn't it
1:07:21 anyway all right so there there's our
1:07:24 science all right wouldn't wouldn't the
1:07:27 discovered three million new elements
1:07:29 crystals fall into play solving these
1:07:33 problems absolutely Wolf Man Clint
1:07:35 that's what I'm saying all of these this
1:07:37 is this V is amazing this video is
1:07:39 amazing I know isn't it um the reason I
1:07:43 wanted to show these three stories I
1:07:45 think these three stories are are are
1:07:48 independently remarkable and if you
1:07:51 imagine three years from now these
1:07:53 things overlapping
1:07:56 it it's it's staggering it's
1:07:59 staggering right this
1:08:02 thing when they get this thing going
1:08:05 we'll have the compute power of every
1:08:07 computer on Earth right now more than
1:08:09 that you know this
1:08:13 thing Evo
1:08:15 2 that understands the 9.3 trillion DNA
1:08:20 base pairs spanning the entire tree of
1:08:23 life
1:08:24 [Laughter]
1:08:28 and then this thing that can go do all
1:08:31 the you know the hypotheses and
1:08:34 experimentation and go make all the [ __ ]
1:08:38 work oh I'm not I'm not showing you all
1:08:40 this
1:08:41 stuff because I'm suck at tabs I suck at
1:08:44 tab
1:08:46 management anyway all right let's move
1:08:49 on let's go let's go watch some robots
1:08:52 talk to each other but I was just saying
1:08:54 you know you've got your your Myer on a
1:08:56 particle on a chip here you've got all
1:09:00 your your DNA sequences in this project
1:09:04 now out there for everyone to use and
1:09:07 then you've got you know scientists in a
1:09:09 box that our current scientists can 100x
1:09:13 THX
1:09:15 10,000x their capacity to discover
1:09:17 things
1:09:22 each like what are the implications of
1:09:24 this five years from now I don't [ __ ]
1:09:26 know but they're
1:09:30 profound ticktockers want to link to the
1:09:33 video we just watched okay we can do
1:09:36 that let me I'll I'll give it to you in
1:09:38 the format I did I'll give you the uh
1:09:40 the Microsoft article so where I'm going
1:09:42 to share this is on I'll go to the news
1:09:45 section of the ai ai salon
1:09:56 um um new state of
1:10:05 matter new state of matter um
1:10:11 million
1:10:15 Cubit
1:10:17 um computer chip
1:10:21 in years not
1:10:27 decades all right so there that that's
1:10:30 posted you can go watch that video there
1:10:33 black box yes there you
1:10:35 go it looks like we're on the verge of
1:10:38 outc creating
1:10:40 ourselves in about 50 years yeah yeah
1:10:44 exactly I again this oh you know let me
1:10:48 I'm going to notify people of this
1:10:49 because I think it's a big deal
1:10:52 um in the AI salon on the wel
1:10:57 page we've got this um we've got a
1:11:00 welcome video and then we've got this
1:11:02 journey to AI Readiness which is you
1:11:04 know play first mindfully create
1:11:05 generously
1:11:08 lead um and then I just added beneath it
1:11:13 the five stages of AI adoption and and
1:11:17 the if if you're like oh that's like the
1:11:20 five stages of grief that's actually
1:11:22 where this came from it it happened on a
1:11:24 live about a week ago where it struck me
1:11:27 that that people as they get curious
1:11:29 about AI really do go go through these
1:11:32 stages of kind of profound acceptance of
1:11:36 of what's happening right now and you
1:11:38 know stage one is dismissal or denial
1:11:41 stage two is awe you know your first
1:11:43 what I call Kevin McAllister moment that
1:11:45 first moment where you're like oh [ __ ] I
1:11:48 had no idea could do that then you go
1:11:51 into this state of like Wonder and fear
1:11:54 at the same time where you're like wait
1:11:56 like like the [ __ ] we just looked at
1:11:58 like it's it is so
1:12:01 Fantastical that you're like well that's
1:12:04 amazing and then you're like but wait
1:12:06 does that like that could be scary too
1:12:09 like it's it's like it's both right like
1:12:12 how do you even
1:12:13 process each of those things
1:12:15 individually much less the three of them
1:12:17 together much less the dozens of things
1:12:20 that have probably happened in the past
1:12:21 two weeks that are all going to start to
1:12:24 and co-mingle
1:12:26 and then you then you start to emerge
1:12:28 out of that into into stage four where
1:12:31 you realize ah these tools aren't
1:12:34 replacing me they're extending
1:12:37 me that stage four the augmentation
1:12:40 stage is where you realize you still get
1:12:42 to have your human
1:12:44 agency and you still get to put ideas in
1:12:47 the world but you now get to do them in
1:12:49 this Amplified
1:12:51 way and then the final stage is I stole
1:12:54 from from David shap her and I don't
1:12:56 know if he stole it from someone else
1:12:58 but cognitive hyper
1:13:00 abundance that the stage after realizing
1:13:03 they're not going to replace me they can
1:13:04 just amplify me is this is this stage of
1:13:09 I could have a thousanders company doing
1:13:12 anything I want it to do or a 10,000
1:13:15 person company or a million person
1:13:17 company well what would I have that
1:13:19 company do I don't [ __ ] know that's
1:13:21 your job to figure out right
1:13:25 so this is this is where we're headed
1:13:28 we're not anywhere close to that yet but
1:13:30 this is where we're
1:13:32 headed and and the people that get
1:13:34 through
1:13:36 this you know this Evolution and I think
1:13:39 this is a cyclical Evolution because as
1:13:41 the Technologies keep evolving we we're
1:13:44 going to keep going through Kevin
1:13:46 McAllister
1:13:49 moments satian Adella the CEO of
1:13:52 Microsoft saying
1:13:55 we're announcing a new state of
1:14:00 matter that's not that's not that's not
1:14:06 insignificant right that's profound
1:14:08 that's [ __ ] Kevin McAllister moment
1:14:09 so here's a new stage two so I think
1:14:12 these things are nested and
1:14:15 cyclical
1:14:16 anyway all
1:14:18 right so let's go let's go play let me
1:14:22 show you one other thing on the Twitter
1:14:24 on the Twitter
1:14:27 all right so Dr Millian uh
1:14:30 milanovich um said most people are still
1:14:33 prompting wrong open AI president Greg
1:14:36 Brockman shared a framework to fix that
1:14:40 oh okay so Greg Brockman did share this
1:14:42 but this is this is his thing so he says
1:14:45 this this is for reasoning engines you
1:14:47 know goals clearly Define what you need
1:14:50 return format specify how you want the
1:14:53 the the response structure
1:14:56 set guard rails for
1:15:00 accuracy return to return the top three
1:15:03 okay let's be careful to make sure that
1:15:05 the name of the trail is correct okay
1:15:08 great oh that's
1:15:09 cool and then a context dump add
1:15:19 background all
1:15:21 right so let's go to grock here we are
1:15:25 at grock
1:15:29 three and we're going to do we don't
1:15:31 need a search here we're going to do a
1:15:34 think so we're g to have it do deep
1:15:38 thinking and then we're going to say
1:15:40 okay let's let's go back to that what's
1:15:43 what's our
1:15:44 uh so clearly Define what you need so
1:15:47 what we're going to create is we're
1:15:49 going to create
1:15:52 um I want um
1:15:56 you to act
1:16:00 as a
1:16:03 senior software
1:16:06 engineer
1:16:10 specializing
1:16:12 in video
1:16:15 game
1:16:19 development
1:16:23 um I want to make
1:16:25 make a
1:16:28 an
1:16:31 asteroids
1:16:40 clone
1:16:42 that
1:16:46 uses vector
1:16:49 graphics like
1:16:52 the original
1:16:55 but
1:16:58 also
1:17:00 uses 3D
1:17:09 objects all right let's go back to our
1:17:11 little prompty thing here
1:17:27 okay cool got got it Brandon yeah so
1:17:29 baked we'll we'll take your question and
1:17:32 we'll run it through we'll run it
1:17:34 through grock while it's thinking we'll
1:17:35 go do another instance of grock all
1:17:37 right so return format so um I want you
1:17:42 to write a
1:17:47 software spec that I can give to
1:17:52 a coder
1:17:56 uh an
1:17:57 engineer a gaming
1:18:01 engineer
1:18:03 that focuses
1:18:09 on
1:18:11 smooth game
1:18:14 play um
1:18:18 and game
1:18:21 mechanics mechanics me can
1:18:26 all
1:18:28 right back to our little prompty thing
1:18:30 here if you know me I I am horrible at
1:18:34 structured prompts because I don't have
1:18:36 the attention span for it but we're
1:18:38 gonna try it like I I feel like I've
1:18:41 typed enough already um let's see
1:18:44 warning set guard rails for accuracy and
1:18:46 their example was be careful that the
1:18:48 name of the trail is correct okay so
1:18:51 um be careful
1:19:00 to make sure that the
1:19:05 code
1:19:07 created
1:19:08 isn't just basic game
1:19:13 play
1:19:15 with
1:19:19 unsophisticated
1:19:21 graphics and physics
1:19:26 but
1:19:27 Advanced and super
1:19:31 playable and then context and I'll do
1:19:34 this as a I'll I'll do voice to text
1:19:36 here Tik Tock pin Source Camp my
1:19:38 88-year-old mom just texted me to tell
1:19:40 me that grock 3 is now available she's
1:19:43 so cool that's awesome Source
1:19:46 Camp Source Camp bringing the the
1:19:49 octogenarians into the AI into the AI
1:19:52 space [ __ ] beautiful love it
1:19:55 grock 3 is now available go go go uh by
1:19:59 the way if you want to play with grock 3
1:20:01 you do have to be a premium plus member
1:20:03 which I think they just up the monthly
1:20:05 subscription on that to 40 bucks uh if
1:20:08 if they're going to charge I think I'm
1:20:10 paying way less than that if they're
1:20:11 gonna up my price to 40 I'm Gonna Cancel
1:20:13 it because I don't give a [ __ ] enough
1:20:16 about
1:20:16 it she wants to know if she should get
1:20:20 it well there you go nope it's free oh
1:20:23 all right well go play with it for free
1:20:26 then no that's
1:20:29 Pro 16 bucks a month for premium plus oh
1:20:32 okay great awesome great fantastic well
1:20:35 then 40 bucks a month for pro is not bad
1:20:37 if it's similar to the chat GPT Pro
1:20:40 whatever the IOS app is free okay good
1:20:43 there you go go get the IOS grock app
1:20:45 it's free okay let me um so I'm gonna
1:20:48 say um context
1:20:56 I'm a gen
1:20:57 xer who grew up playing
1:21:01 asteroids as my favorite video game in
1:21:04 the 80s when I would go to the mall with
1:21:07 a roll of quarters and feel like a
1:21:09 flipping King
1:21:12 period I also now have a community of AI
1:21:17 optimists who are watching in real time
1:21:20 right
1:21:21 now as you are creating this software
1:21:24 spec and some other AI tools are going
1:21:26 to go build this
1:21:29 period the quality of this game and the
1:21:32 success of this project over the next
1:21:36 hour or
1:21:37 so are going to make or break my
1:21:40 reputation
1:21:42 period if this game fails
1:21:46 period if the robots that I'm about to
1:21:49 assign to build this thing fail period
1:21:53 if I fail
1:21:55 period my
1:21:57 community will forever torment
1:22:01 me with the horrific
1:22:05 failure and acceptance that I am wholly
1:22:11 unqualified to be doing anything
1:22:14 resembling the work I'm doing
1:22:18 period Please Don't Let Me suffer that
1:22:22 fate period please for the love of God
1:22:26 or should I say the
1:22:28 algorithm make this thing
1:22:39 succeed all right let's let's see what
1:22:42 this thing does all right so we got we
1:22:44 got our we got our structured prompt our
1:22:47 Greg Brockman approved
1:22:49 prompt we we've we've made it clear that
1:22:52 the stakes are
1:22:53 high and then I'm gen
1:22:58 xer and I will I will I will tell it if
1:23:01 it's a piece of
1:23:03 [ __ ] I'll trigger it I'll trigger it on
1:23:06 purpose here we
1:23:08 go all right so that's off thinking two
1:23:11 seconds three seconds it's doing its
1:23:13 thing all right let's create a new grock
1:23:15 gro.com
1:23:25 what would happen to AI if it used
1:23:29 layered holographic and CIS
1:23:31 crystallin matrices for
1:23:35 data I'm going to turn on oh I can only
1:23:38 do we'll do it like
1:23:42 this
1:23:46 boom let's let's look at what this thing
1:23:50 is first let's break down the question
1:23:52 this is a bit speculative
1:23:55 ah here we
1:23:56 go baked this is for
1:24:00 you this is Gro if AI were to use
1:24:03 layered holographic and crystallin
1:24:05 matrices for data storage and processing
1:24:08 it could undergo profound transformation
1:24:10 and capabilities pushing the boundaries
1:24:12 of what we currently believe holographic
1:24:14 matrices these involve using lasers to
1:24:17 store
1:24:19 data crystalline m m
1:24:22 matrices these are highly ordered atomic
1:24:24 structures think of perfect crystal
1:24:26 lattice layering stack them together
1:24:30 massive memory capacity Lightning Fast
1:24:32 data access breakthrough and processing
1:24:34 power challenges to overcome okay we got
1:24:37 those the big picture
1:24:38 longterm using AI holographic and
1:24:41 crystallin matrices could evolve into
1:24:43 something resembling super intelligence
1:24:45 with near Limitless memory rapid data
1:24:47 access and advanced computation it might
1:24:50 simulate entire realities with stunning
1:24:53 accuracy we're living in simulation
1:24:57 people you know when I heard like I
1:25:00 don't know five years ago 10 years ago
1:25:02 when people first started talking about
1:25:04 that they're like well the the more we
1:25:06 do the math the more it appears that we
1:25:08 could be living in a simulation I'm like
1:25:10 you [ __ ] nerds with your stupid
1:25:12 [ __ ] theories and your [ __ ] math
1:25:15 with your
1:25:16 calculators and now it
1:25:20 appears it's actually kind of a
1:25:22 possibility
1:25:26 oh fix your Tik Tok view that is
1:25:28 embarrassing good
1:25:31 Lord sorry about that there uh everybody
1:25:35 and most most biggest apology to to Ye
1:25:40 Old Brandon the producer screaming in
1:25:42 the Void look at your [ __ ]
1:25:47 computer um however this Leap Forward
1:25:50 would also amplify ethical questions oh
1:25:53 wait so wait it could it could simulate
1:25:55 entire realities it could predict
1:25:57 outcomes for complex systems like
1:25:58 climate economies flawlessly it could
1:26:01 solve problems Humanity's long
1:26:02 considered intractable from curing
1:26:04 diseases to optimizing energy
1:26:07 grids um who gets access to it yeah who
1:26:10 controls it in conclusion if AI
1:26:12 harnessed so that's a that's a fairly
1:26:15 lame
1:26:16 ass that's a fairly lame ass I I'll do
1:26:19 deep search here's what we'll do we'll
1:26:20 say I'm gonna say that feels surface
1:26:25 level and basic and
1:26:29 me with my crappy acting
1:26:34 degree could have
1:26:37 come up with
1:26:41 um dig
1:26:43 deeper do some
1:26:46 research and um
1:26:56 [Music]
1:26:58 uh
1:27:05 explore what this future looks like
1:27:11 with
1:27:13 um probabilities
1:27:16 Pro uh abilities abilities whatever
1:27:20 doesn't [ __ ] matter um waiting
1:27:25 your
1:27:27 predictions let's see if it let's see if
1:27:29 it gets snarky with me because I told it
1:27:31 it was a piece of
1:27:35 [ __ ]
1:27:37 nope the request is about all right so
1:27:39 it's gonna go do that so let's go back
1:27:41 to the other one
1:27:42 now all right let's see what our our
1:27:45 software spec looks like I'm not really
1:27:46 going to read it other than this
1:27:48 document outlines core
1:27:53 features um is open GL webbased um uh
1:27:59 this the the tool I'm using will create
1:28:06 a web based
1:28:09 application so update this for web
1:28:17 Technologies uh
1:28:20 if it's not currently for
1:28:33 [Laughter]
1:28:34 [Music]
1:28:38 that wow this is doing some some pretty
1:28:40 cool internal
1:28:44 prompting first for rendering instead of
1:28:46 open GL we'll use webg good okay so this
1:28:49 is good for the physics engine box 2D is
1:28:52 still a viable option
1:28:55 audio can be handled with the web audio
1:28:59 API and by the way I I very much doubt
1:29:03 we're going to get very far with this
1:29:05 but but I'll at least let you see what
1:29:07 it looks like it's pretty fascinating
1:29:09 this
1:29:10 stuff
1:29:12 [Music]
1:29:22 um okay
1:29:25 let's see software specification for a
1:29:27 web-based asteroid clone
1:29:30 introduction specification the classic
1:29:33 1980s game with modern 3D elements
1:29:36 delivering a nostalgic yet fresh
1:29:39 experience in the
1:29:41 browser designed for web technologies
1:29:43 that prioritizes smooth
1:29:47 performance the game will run in a
1:29:49 canvas single page application fine
1:29:51 whatever like let's just assume this
1:29:54 thing knows what the [ __ ] it's doing
1:29:58 right the other thing we could do is we
1:30:00 could have grock write the code and
1:30:02 compare but that's not any fun we we
1:30:05 wanna we want to push we want to push
1:30:07 the uh the bounds of what these shitty
1:30:10 Technologies can do
1:30:12 okay
1:30:14 so let me jump over that's notebook LM
1:30:19 here oh that's Opus clip where operator
1:30:23 all right here's
1:30:27 Operator Let Me reload this
1:30:30 because if you're impressed by the
1:30:32 number of tabs I have open thank
1:30:37 you okay so if you don't know what
1:30:39 operator is operator is open
1:30:41 AI you you have to be a pro member so
1:30:45 it's 200 bucks a month to have access to
1:30:46 this bad boy and this thing can
1:30:51 use it can surf the web and it can use
1:30:54 keyboard and mouse which when you first
1:30:56 hear that you're like yeah
1:30:59 whatever kbm keyboard and
1:31:03 mouse but it's a big deal because that's
1:31:05 how we use right computers and you don't
1:31:08 need to write apis that's kind of the
1:31:10 big deal here is that this can go
1:31:12 instead of me creating an application
1:31:15 that does an API call to lovable I can
1:31:18 just point it to The Lovable website and
1:31:20 go do it so I'm going to say below is a
1:31:26 software
1:31:29 spec for
1:31:31 a video
1:31:34 game
1:31:36 and I want you to
1:31:40 use lovable.
1:31:44 deev
1:31:46 and build me this game I think I'll just
1:31:50 leave it like that and then we're going
1:31:52 to paste in the thing that grock gave
1:31:55 us right so there's our big ass
1:32:00 prompt and I think we'll we'll we'll
1:32:03 kick it off all right let's see great
1:32:06 job you've been going for 90 minutes I'm
1:32:09 impressed by the fact you know which one
1:32:11 is
1:32:12 which which what is which which of these
1:32:15 tools is which I that's that's part of
1:32:17 the the the the uh the psychopathic
1:32:21 nature of paying attention to this AI
1:32:23 stuff is you have to juggle all this
1:32:25 [ __ ] in your brain oh the tabs oh no the
1:32:28 tabs this is I'm a tab hoarder like I
1:32:31 know that I know that all of these X's
1:32:34 together are that's my little X section
1:32:36 I know to the left over here is
1:32:38 generally my calendar Google Drive and
1:32:41 Gmail I know that my most recent things
1:32:44 that I want to show people are kind of
1:32:45 over here on the right and then
1:32:47 somewhere in the middle here is all the
1:32:49 things I I will eventually get back to
1:32:57 if you're not a tab hoarder you just
1:32:58 don't understand you people with
1:33:00 functional prefrontal CeX is drive us
1:33:04 adhders [ __ ] baty just make a list
1:33:08 why don't you just remember you know you
1:33:09 could bookmark it and you could organize
1:33:11 your bookmark into into nested category
1:33:14 yeah you could if if your brain [ __ ]
1:33:17 functioned like that you could
1:33:25 mine is a little more special mine I've
1:33:28 got my brain is little pink bow pink bow
1:33:32 brain so mine is special all right here
1:33:36 we go
1:33:38 boom here goes
1:33:41 operator all right so this thing is now
1:33:44 fired up its own web
1:33:46 browser let's shrink this a little
1:33:50 bit IDE to app in seconds it just logged
1:33:52 into lovable and I'm I'm still logged
1:33:56 in all
1:34:00 right can I can I change these window
1:34:03 sizes guess
1:34:05 not all right asteroids clone using
1:34:08 something so what's cool about this so
1:34:10 so it's now logged into
1:34:12 lovable oh and wait if I say view in the
1:34:15 browser oh yeah that's that okay I've
1:34:18 entered a summary of the asteroids clone
1:34:20 should I proceed with submitting the
1:34:23 request I'm GNA say uh duh yes
1:34:29 [Laughter]
1:34:45 please and so so what we're watching now
1:34:49 so we had one robot write us a software
1:34:52 spec the other robot ingested that
1:34:55 software spec put a
1:34:57 simple request into lovable to start
1:35:00 this project and in
1:35:03 theory because operator has read and
1:35:06 understood that software spec it's going
1:35:09 to tell lovable to make this game less
1:35:15 shitty and then the way it works is
1:35:17 welcome to your blank app the way it
1:35:19 works is whenever there's a substantive
1:35:22 decision to make it it says you gotta
1:35:27 make a decision so here The Lovable
1:35:29 platform is prompting me to connect with
1:35:31 GitHub to edit the Project's code should
1:35:34 I proceed with connecting to
1:35:37 GitHub
1:35:39 yes and it it might need me to do this
1:35:42 but I think I'm logged into GitHub but
1:35:43 we'll
1:35:45 see if you're logged in it'll remember
1:35:47 that [ __ ] for
1:35:49 you um
1:35:57 yeah because I'm already logged in
1:35:59 GitHub is asking to install and
1:36:01 authorize lovable dodev on your Kyle
1:36:04 Shannon account with options to apply it
1:36:08 to all repositories yes because I don't
1:36:10 give a
1:36:13 [ __ ] you security people out there I
1:36:16 know you're you're hemorrhaging right
1:36:18 now
1:36:24 oh I gotta get a password let's see let
1:36:27 me stop sharing my screen for a
1:36:30 second and go get my password so so what
1:36:33 I'm going to do now is I'm gonna I'm G
1:36:34 to log into the
1:36:36 browser so I'm going to take
1:36:40 control you now have control and I'm
1:36:43 going to go
1:36:52 to wait what is it that I have to log
1:36:55 into it's it's
1:36:58 uh GitHub
1:37:21 okay um
1:37:24 copy we'll go back to this
1:37:29 thing no not that thing this thing yes
1:37:32 we'll go
1:37:35 paste that doesn't look
1:37:39 right let's go back
1:37:44 here let me make sure I'm not screwing
1:37:47 the
1:37:51 pooch go back here
1:37:56 black bar
1:37:57 okay there's that
1:38:03 confirm incorrect password
1:38:07 really oh yeah I think I
1:38:22 know okay
1:38:24 so let me let me share my screen
1:38:30 again is that right
1:38:34 no okay producer Brandon got
1:38:41 it all right so I'm logged in there so
1:38:46 I'm G to finish up so I'm gonna say all
1:38:49 logged
1:38:51 into GitHub
1:39:01 [Music]
1:39:04 it would be nice if we got this to at
1:39:06 least some sort of functional Asteroids
1:39:08 game
1:39:10 tonight all right so we can we can look
1:39:13 at what thinking it's doing editing code
1:39:16 for the asteroids clone expanding the
1:39:18 section to view the code so it's
1:39:21 actually what's cool is it's it's using
1:39:23 lovable in a way that I wouldn't use it
1:39:26 it's going and looking at the code I
1:39:27 don't know coding so but it's it's using
1:39:30 some more advanced features than I would
1:39:32 use lovable platform is prompting me to
1:39:34 transfer the repository to GitHub to
1:39:37 collaborate and edit the code yes yes
1:39:40 please
1:39:45 collaborate I'm sure all the developers
1:39:47 watching this are like you got to be
1:39:49 [ __ ] kidding me this bozos
1:40:01 oh man all right GitHub access required
1:40:05 I thought we just did
1:40:10 [Music]
1:40:15 that what's going on what going on there
1:40:21 biddy it's going to tell me that it
1:40:23 couldn't log
1:40:28 in or maybe it fixed
1:40:35 it
1:40:36 huh did it fix it or do I got to go do
1:40:39 more [ __ ] something not
1:40:44 found but what's what what I find
1:40:47 amazing about this
1:40:50 is every time it hits a problem
1:40:54 right that would be me hitting that
1:40:57 problem right like if I went to lovable
1:40:59 and I typed in [ __ ] and then it it said
1:41:03 thing not
1:41:05 found at some point I would get
1:41:08 discouraged and I would leave
1:41:11 right um is it creating the code or just
1:41:15 copying from an existing repo and then
1:41:18 editing um I I well I'm pretty sure
1:41:22 lovables creating the code I don't I I I
1:41:26 I honestly don't know Pate but I don't
1:41:29 think it's I don't think it's going to
1:41:31 GitHub and finding an asteroid's clone
1:41:33 and copying
1:41:37 it let's see what it says here checking
1:41:41 the GitHub rep repository access status
1:41:44 okay attempting to publish it's saying
1:41:46 it's publishing to
1:41:49 GitHub rather than pulling [ __ ] from it
1:41:56 all right now it's doing lovable
1:41:57 platform it wants to connect to super
1:42:00 base which will allow it persistent data
1:42:03 yes make the connection
1:42:05 make the super base connection
1:42:11 please and I could have just said yes
1:42:13 there but you know I'm
1:42:15 [Laughter]
1:42:22 wordy and I think logged into the
1:42:24 superbase thing already so because I'm a
1:42:27 tab hoarder I think I've got all this
1:42:29 [ __ ] pre-connected keeps going to not
1:42:32 found I guess it's not found because
1:42:34 that's the that's the application that
1:42:37 doesn't exist yet because it's doing
1:42:39 it's doing all this setup for
1:42:42 us I think that just said super base
1:42:45 connected The Lovable platform is ready
1:42:47 to transfer okay Pate here's the answer
1:42:50 The Lovable platform is ready to
1:42:52 transfer the Repository to your GitHub
1:42:55 account for collaboration and code
1:42:57 editing it warns that lovable doesn't
1:43:00 keep a copy so careful handling is
1:43:03 advised should I proceed with the
1:43:04 transfer
1:43:08 sure and if your next question PA is why
1:43:11 is it doing that my next answer is I
1:43:13 don't [ __ ] know yeah that sounds
1:43:15 right then something it said earlier
1:43:19 made it a bit more ambiguous okay let's
1:43:21 let's see what it's saying here in its
1:43:22 instructions
1:43:24 proceeding with the repository transfer
1:43:26 initiating repository transfer for the
1:43:30 user Cosmic rocks Revenge I like the
1:43:32 name of the game it came up with
1:43:34 accessing GitHub repository for
1:43:36 verification exploring source code for
1:43:40 asteroids
1:43:42 clone navigating to game directory
1:43:46 contents accessing controls. Ts for game
1:43:51 details is Docker running I
1:43:54 listen you people with your are do you
1:43:57 have any qualifications Kyle questions
1:44:00 the answer we have established over
1:44:03 years now is no I am not qualified to be
1:44:09 here I don't think it needs Docker but I
1:44:12 don't
1:44:12 [Laughter]
1:44:16 know but look it made code
1:44:24 the project has been successfully
1:44:26 transferred to your GitHub repository
1:44:28 you can view and edit it here let me
1:44:29 know if you need any further assistance
1:44:31 oh so
1:44:33 okay I
1:44:35 fear you have made an error I am not a
1:44:42 coder and have no
1:44:46 interest in coding
1:44:49 in
1:44:52 GitHub the
1:44:55 whole purpose of
1:45:00 lovable
1:45:02 is to
1:45:06 code and create the app for us and I
1:45:12 want
1:45:14 you to use
1:45:21 lovable just like I would
1:45:24 would have it make the game and you make
1:45:31 sure the
1:45:33 game follows the spec I gave
1:45:37 you so this is a really good example of
1:45:42 these tools are janky and weird right
1:45:43 now and I and I probably didn't give it
1:45:45 enough context early on this is the kind
1:45:48 of [ __ ] that you know over the next two
1:45:51 to three years will get more and more
1:45:54 and more sophisticated it it'll make it
1:45:56 it won't make these kinds of mistakes
1:45:58 it'll be things like oh I noticed that
1:46:00 lovable you
1:46:02 know you know hides the code from you do
1:46:04 do you want me just to make the game yes
1:46:10 right um let's see do operators
1:46:14 understand the context of dark mode I
1:46:16 have no
1:46:17 [Laughter]
1:46:19 idea
1:46:21 um I have a hard time trusting people
1:46:24 who don't use dark
1:46:26 mode I have the the the reason I've got
1:46:30 some things in dark mode and some things
1:46:32 not Pate is Chrome talk to your bosses
1:46:37 Chrome ate my Chrome profile on this
1:46:41 computer and I can't get it back it's
1:46:42 gone I don't have time machine on on
1:46:45 this machine for some reason and Google
1:46:48 doesn't keep a history of browser
1:46:51 profiles that I can find
1:46:54 all right I understand your preference
1:46:56 for using lovable to create the game all
1:46:59 right so I don't know what it's doing I
1:47:00 think
1:47:03 it's I I have no idea what it's doing
1:47:05 let's see what it's doing using the chat
1:47:08 interface for game development resending
1:47:10 the message to request great game
1:47:12 development attempting attempting to fix
1:47:14 an app error viewing logs viewing logs
1:47:17 for error
1:47:19 diagnosis see this is the cool thing
1:47:21 like because it understands code it's
1:47:23 going to use lovable in a in a much
1:47:25 deeper way than I ever
1:47:29 would I asked foro a question it replied
1:47:32 that's a damn good question blah blah
1:47:34 blah oh that's pretty
1:47:36 good yeah I think I think cat gpts that
1:47:39 they're upping the personality on it a
1:47:41 bit any news on the Deep dive after this
1:47:44 is
1:47:46 made which Deep dive baked oh the Deep
1:47:49 dive of okay let's let let's let this
1:47:51 keep going and let's go look at at your
1:47:54 deep
1:47:55 dive it read 45 so so this is for baked
1:47:59 so so baked asked the very simple simple
1:48:04 question which I think it's kind of
1:48:07 obvious right if you're in AI I think
1:48:10 all of us are probably thinking this
1:48:12 deep down what would happen to AI if it
1:48:15 used layered holographic and crystall
1:48:17 and matrices for data right like fairly
1:48:21 obvious so we gave it to grock and told
1:48:25 it to think about it and it gave us a
1:48:26 shitty answer so then I told it hey you
1:48:28 gave us a shitty answer I could have
1:48:30 come up with that crap dig deeper so and
1:48:34 I turned on deep search so it it went to
1:48:38 it thought for 24 seconds went to 50
1:48:41 sources key points AI using holographic
1:48:45 and crystallin matrices could store vast
1:48:47 amounts in small spaces process it
1:48:50 faster and last longer I think the last
1:48:53 longer thing is the thing I'm most
1:48:55 excited about with that Tech
1:48:58 um this might lead to super intelligent
1:49:00 AI the technology is still still
1:49:02 experimental yeah baked this isn't real
1:49:06 yet with a 50% chance of adoption in 10
1:49:08 years and 90% in 20 I have a feeling all
1:49:11 these predictions are going to be all
1:49:13 wrong because of the material science
1:49:15 stuff happening right now I think
1:49:17 they're going to be shorter introduction
1:49:19 to technology impact on AI capabilities
1:49:22 this we saw before changes and ethical
1:49:24 challenges and ethical considerations
1:49:26 that we saw before survey
1:49:29 note detailed analysis of AI with
1:49:32 layered holographic and crystallin
1:49:33 matrices this section provides
1:49:35 comprehensive
1:49:37 exploration background and technological
1:49:42 foundations so there it talks about
1:49:44 crystallin
1:49:46 matrices known as Superman memory
1:49:50 Crystal 5D memory Crystal technology
1:49:53 can store up to 360 terabytes per disc
1:49:57 with a stability of billions of years at
1:50:00 high temperatures that's
1:50:03 amazing that's so
1:50:05 cool um potential impacts on
1:50:10 AI processing speed would be really
1:50:15 fast challenges cost and
1:50:18 scalability data security high density
1:50:21 storage increases risks of data breaches
1:50:24 a single stolen Crystal could compromise
1:50:28 all of your
1:50:31 data so it is just like the movies where
1:50:34 someone goes in and steals something
1:50:36 like the size of a pencil eraser and it
1:50:38 contains all of the knowledge of all of
1:50:40 humanity we're
1:50:43 there interoperability wait oh
1:50:46 interpretability as AI processes data in
1:50:49 multi-dimensional spaces understanding
1:50:51 decisionmaking becomes harder
1:50:54 exacerbating the Black Box problem a
1:50:56 concern with AI in AI ethics
1:50:59 literature ethics and societal
1:51:01 implications surveillance and privacy
1:51:03 job displacement probability waiting
1:51:07 predictions okay here you go baked given
1:51:10 the experimental stage of these
1:51:12 Technologies adoption timelines are
1:51:14 uncertain based on current research
1:51:16 probability of adoption in 5 years 10%
1:51:19 due to high development costs
1:51:21 probability in 10 years 50% probability
1:51:24 in 20 years
1:51:26 90% impact probabilities assuming
1:51:29 adoption enhanced memory and speed 90%
1:51:31 Quantum Computing integration 50% with
1:51:35 the Microsoft announcement today that
1:51:37 probably goes
1:51:38 up um ethical consider ethical
1:51:41 challenges
1:51:44 100% there will be 100% chance that
1:51:47 there will be ethical considerations
1:51:49 here um with specific risks like private
1:51:53 at
1:51:54 80% future scenario AI in 2035 imagine
1:51:59 2035 AI systems used layered holographic
1:52:02 and Crystal and M matrices data centers
1:52:05 replaced by compact Crystal units
1:52:07 storing exobytes in book size spaces
1:52:10 reducing physical footprint AI training
1:52:13 training time drops to
1:52:15 hours instead of months or years
1:52:18 enabling rapid model iteration for
1:52:21 applications like climate model modeling
1:52:23 personal assistance store entire life
1:52:26 data in a crystal offering hyper
1:52:28 personalized services but raising
1:52:31 privacy concerns healthc care analyzed
1:52:34 vast patient data in real time improving
1:52:37 Diagnostics but requiring strong data
1:52:40 security which you can't have strong
1:52:41 data security because the quantum
1:52:43 computers have cracked all the the
1:52:45 encryption autonomous system
1:52:47 self-driving cars with onboard AI
1:52:49 processes data instantly enhancing
1:52:52 safety but increasing dependency that's
1:52:54 an interesting
1:52:55 one comparative State um current
1:53:00 terabytes per server pedabytes exobytes
1:53:02 per Crystal so there's a little table
1:53:05 conclusion the adoption of layered
1:53:07 holographic Crystal and matrices could
1:53:09 significantly enhance ai's capabilities
1:53:12 offering vast storage faster processing
1:53:14 long-term durability however challenges
1:53:17 like cost security and ethical
1:53:19 implications must be addressed with a
1:53:22 50% chance of adoption in 10 years 90%
1:53:25 in 20 this technology could redefine
1:53:27 ai's role but require careful governance
1:53:30 manage to manage risks I mean honest to
1:53:34 God our current politics they they don't
1:53:38 understand what [ __ ] Wi-Fi
1:53:40 is if you're like I can put all of hum
1:53:43 human knowledge into a chip the size of
1:53:45 a paperback
1:53:47 book Jesus
1:53:51 Christ there you have it that was not
1:53:53 bad that that's a pretty good little
1:53:55 thing all right let's go see if our if
1:53:56 we have an Asteroids game yet I'm gonna
1:53:58 guess
1:54:00 no uh oh
1:54:07 resume all right that's
1:54:11 off we're already compromised yeah but
1:54:14 this this what the the compro the comp
1:54:17 the data compromising with with Quantum
1:54:19 stuff goes to a whole different level
1:54:22 because right now all personally
1:54:23 compromised because we voluntarily gave
1:54:25 our data to Facebook and Google and
1:54:28 everyone else and now all the thirdparty
1:54:30 companies have triangulated everything
1:54:32 so that's that's on us but the the the
1:54:36 Privacy
1:54:37 stuff when when we have quantum
1:54:40 computers is is going to be
1:54:42 Bonkers um two Tik Tock Galaxy Gifts oh
1:54:45 thank you very much for the Tik Tock
1:54:46 gifts from from uh fasel Benin Alin
1:54:50 thank you so much really appreciate that
1:54:53 saw I saw something out of the corner of
1:54:54 my eye and didn't realize that it was a
1:54:56 Galaxy gift those are beautiful by the
1:54:58 way and very generous of you so thank
1:54:59 you so
1:55:02 much baked asket nested sacred geo with
1:55:06 AI mod
1:55:08 electromagnets and Magneto ferof fluid
1:55:11 fluid with healing frequency full
1:55:13 paragraph oh full power all
1:55:20 right all right let me go back so
1:55:23 oh I guess this thing won't
1:55:26 work is this still
1:55:30 going oh yeah it's still
1:55:33 going all right so
1:55:36 so look at all the steps it did it's
1:55:38 doing on our
1:55:42 behalf oh do we have a
1:55:44 game no
1:55:47 okay let me go let me go back to this
1:55:49 for a second we'll go back to grock let
1:55:51 me
1:55:55 okay so we'll do deep search again
1:55:58 because that was pretty good okay let's
1:56:01 take it a step
1:56:09 further let's take it a step further or
1:56:15 further
1:56:18 um I want you to think about
1:56:24 AI in the context of sacred
1:56:34 geometry
1:56:36 with AI
1:56:40 modified I'll just put what you put AI
1:56:45 mod
1:56:49 electromagnets
1:56:50 and Magneto
1:56:54 you're just making [ __ ] up now Pharaoh
1:56:58 fluid Pharaoh
1:57:01 fluid
1:57:03 with
1:57:05 healing
1:57:09 frequencies am i g to get like a knock
1:57:11 on the door from some three-letter
1:57:12 agency uh Mr Shannon uh we were
1:57:15 wondering what you typing into grock the
1:57:17 other night full
1:57:20 power um utilize in full power all right
1:57:25 here we go go grock go all right so
1:57:30 we're GNA do the ferof fluid sacred
1:57:33 geometry thing I if if honest to God if
1:57:36 you open up a wormhole in my living room
1:57:40 we're gonna have a situation here all
1:57:43 right is this thing go still
1:57:47 going selecting physics. TS to fix error
1:57:51 accessing the code editor
1:57:58 so here's an interesting thing so
1:58:01 because I gave it a really complicated
1:58:02 software
1:58:05 spec having used lovable like I would be
1:58:08 playing an Asteroids game right now
1:58:11 because I don't know what I'm doing
1:58:14 because this thing knows what it's doing
1:58:16 it's using all sorts of libraries and
1:58:18 things like that it's going in and
1:58:19 editing code and trying to fix [ __ ] on
1:58:21 its own rather than let letting lovable
1:58:23 do
1:58:24 that um so this is taking a lot longer
1:58:28 in theory this is going to make a better
1:58:30 game than I would have produced
1:58:35 but we're just watching it code it's the
1:58:38 engineer that can't make a game yeah so
1:58:41 so there's there's a a guy in the AI
1:58:43 salon named HT snowday and he did this
1:58:47 contest with 40 of his people is in his
1:58:51 company and it was a a 30 minute contest
1:58:54 and the the the challenge was make a
1:58:57 game using Ai and it has to be a a game
1:59:01 where [ __ ] falls from the sky and you
1:59:03 catch it and it's got to be web based
1:59:05 and whatever 40 people did it the
1:59:08 fastest game was produced in like two
1:59:10 minutes by one of the customer service
1:59:12 people after a half hour the only people
1:59:15 that did not succeed in making a
1:59:18 functional game were the engineers and
1:59:21 then he went back and did analysis of
1:59:23 the prompts and not a single engineer
1:59:26 used the word game in their
1:59:30 prompt they had the burden of knowing
1:59:33 how to code a
1:59:35 game and so they couldn't code a game
1:59:37 but the people that just said give me a
1:59:39 game where [ __ ] falls from the sky and I
1:59:42 catch it they made successful games so
1:59:44 that's I think what we're experiencing
1:59:46 here the game is encountering a webgl
1:59:48 context creation error which persists
1:59:51 despite several attempts would you like
1:59:52 me to explore Alternatives um I'm going
1:59:54 to say try to let
1:59:58 lovable solve this for you instead of
2:00:04 you um fixing the code let's see if if
2:00:09 it does that better pay
2:00:12 pin really complicated software spec
2:00:15 well no sorry really complicated
2:00:18 software spec for me compared to what I
2:00:22 would have put in it I have just said
2:00:24 make me an Asteroids game the fact that
2:00:26 it it specified libraries and and you
2:00:31 know all that sort of [ __ ] that's the
2:00:33 [ __ ] that's Way Beyond so no that that
2:00:36 was a very very simple software spec
2:00:38 good point HT was on the live earlier oh
2:00:41 that's so cool that's so cool all right
2:00:46 beautiful all right we'll watch this for
2:00:48 a little bit more the game is
2:00:49 encountering a web jgl wait uh
2:00:53 would you like me to explore the
2:00:54 alternative
2:00:58 Solutions no have lovable fix
2:01:08 it
2:01:10 thinking looking at the screen
2:01:13 requesting lovable to troubleshoot the
2:01:15 error
2:01:26 it's still encountering it
2:01:30 um let's
2:01:34 see edit select
2:01:39 edit oh it looks like lovables
2:01:48 hung it looks like it's in a doom Loop
2:02:06 try fixing the
2:02:17 error tonight's deep dive how to break
2:02:20 three large language models in one night
2:02:23 resume
2:02:24 operator all right let's see let's see
2:02:26 if
2:02:28 uh let's see if anything happens
2:02:37 here yeah lovables clearly in a in a bit
2:02:40 of a
2:02:49 mess well actually you know what we
2:02:51 could do
2:02:56 which one did I use is this it
2:02:59 no that's operator
2:03:02 there did this do its thing by the
2:03:06 way welcome to chat ad
2:03:12 add 28 web pages key
2:03:19 points wait that must be the old
2:03:38 one did it not
2:04:00 it completely ignored all
2:04:03 the sacred geometry
2:04:20 [ __ ] huh
2:04:24 I'm GNA ask
2:04:25 it why did you
2:04:30 ignore
2:04:33 all the
2:04:36 sacred
2:04:39 geometry
2:04:43 geometry stuff I gave
2:04:48 you let's just see what it
2:04:50 says you you're absolutely right I
2:04:53 didn't fully
2:04:55 address sacred geometry electromagnetics
2:04:58 magnetic fluid healing frequencies I got
2:05:01 caught up in the layered thing let's dig
2:05:03 into a fresh angle core components in
2:05:06 their roles sacred
2:05:09 geometry electromagnetics Magneto as a
2:05:13 responsive
2:05:20 medium okay what the future looks like
2:05:22 here you go
2:05:25 baked scenario 2040 the harmonic AI
2:05:30 Temple picture a facility where AI isn't
2:05:33 a cold server rack but a living pulsing
2:05:35 pulsating
2:05:37 structure architecture a Dome built with
2:05:39 sacred proportions golden ratio housing
2:05:42 crystallin arrays platonic solid shapes
2:05:46 electromagnetics pulse
2:05:48 rhythmically shaping ferrofluid pools
2:05:50 into Dynamic mandalas
2:05:52 function the AI processes data
2:05:54 holographically across its geometric
2:05:57 lattice solving problems like climate
2:06:00 optimization or genetic
2:06:02 repair simultaneously it emits healing
2:06:05 frequencies creating a field that calms
2:06:07 visitors and heals ecosystems daily life
2:06:10 people interact with harmonic nodes
2:06:13 small portable versions of this Tech a
2:06:16 node might diagnose your stress via biof
2:06:20 feedback then pulse 4 32 Hertz through
2:06:24 its ferof fluid core to realign you here
2:06:26 we go this is like hippie dippy in
2:06:29 science all coming together baby sacred
2:06:32 geometry's role in computation geometric
2:06:35 patterns could enhance coherence and
2:06:36 data flow Ain to how fractals optimi
2:06:39 optimize neural systems time check 105 I
2:06:42 know surprise if successful it AI might
2:06:46 resonate with universal patterns that's
2:06:49 one of the things that David Shapiro
2:06:50 talked about is all these model improve
2:06:53 they're all trending toward the same
2:06:55 truth they're all sort of getting the
2:06:57 same which is fascinating
2:06:59 electromagnetic amplification ferrofluid
2:07:02 is a living processor kind of
2:07:05 cool um challenges and risks scientific
2:07:08 validation complexity ethics probability
2:07:11 of failure 50% in 10 years dropping to
2:07:15 20% in 20
2:07:20 years 10 years years 40% chance of a
2:07:23 prototype midterm 20 years 70% chance of
2:07:28 a functional system with sacred geometry
2:07:31 as a design ethos and healing frequency
2:07:34 showing measur measurable if
2:07:37 controversial
2:07:39 effects 50 years 90% chance radical
2:07:44 Vision the AI wouldn't just compute it
2:07:47 would resonate by marrying sacred
2:07:49 geometry's patterns electromagnet magnet
2:07:52 isms power ferof fluids adaptability and
2:07:55 healing frequencies potential it could
2:07:57 redefine technology as a holistic
2:08:00 force in 20 years we might see AI not as
2:08:03 a tool but as a partner in
2:08:06 aligning Humanity with Cosmic rhythms um
2:08:10 though proving proving it work proving
2:08:12 it works will be the battle the deeper
2:08:14 surprise if it Taps into Uncharted
2:08:17 physics or Consciousness it might
2:08:19 challenge everything we know about
2:08:20 intelligence itself there you go it's
2:08:24 fascinating that it completely skipped
2:08:26 that when I when I had it do deep
2:08:28 research whatever don't get all
2:08:32 conspiratorial these these I don't I
2:08:34 don't think any of these research labs
2:08:36 are they're all too busy to be thinking
2:08:39 that far
2:08:40 ahead um let me look at operator and see
2:08:43 what it's done the game is encountering
2:08:46 uh I'll say sure fix
2:08:48 it I'll keep playing with this so maybe
2:08:51 what was do
2:08:53 tomorrow is is
2:08:56 uh do a much more simple software spec
2:09:00 we'll we'll just have it describe the
2:09:01 software spec not using specific
2:09:04 Technologies and let lovable we we'll
2:09:07 give it more context that lovable is a
2:09:09 is a program that does its own coding we
2:09:13 don't need to get in there and do the
2:09:14 coding I think that might be the thing
2:09:16 to do that's so very
2:09:19 Gro um the game is in okay yeah so this
2:09:22 is this is in a death Loop right now
2:09:23 we'll start over tomorrow so let me let
2:09:25 me just close this because that ain't
2:09:28 working all right everybody uh let me
2:09:30 get the [ __ ] out of Dodge it's late it's
2:09:32 tired I hope that was fun was that fun
2:09:34 is that good we did a lot
2:09:38 tonight the world is a completely
2:09:40 different place than it was
2:09:42 yesterday uh big big
2:09:45 announcements
2:09:48 um why' that get smaller
2:09:52 champy what's up are you are you
2:09:54 hankering for some
2:09:59 cheese ticktock pin baked thanks again
2:10:02 man for doing that you help spread the
2:10:05 tech and the community thank you baked I
2:10:07 appreciate that yeah listen I mean
2:10:09 here's the thing since I have access to
2:10:12 these tools um the the the salon is
2:10:15 sponsoring my Pro my Pro account the AI
2:10:19 Salon which by the way if you don't know
2:10:21 it if you're not a part of it go there
2:10:25 and click on join our community and come
2:10:27 hang out with
2:10:29 us but because I've got access to this
2:10:32 if there are actual things you want me
2:10:34 to try to do like we did tonight with
2:10:36 baked and is and his really fascinating
2:10:40 questions um bring bring them along and
2:10:42 we'll play like why not use these tools
2:10:44 they're they're they're sitting here
2:10:45 that's why I got them all right um okay
2:10:49 tomorrow's Thursday it'll be nor Al time
2:10:52 800m Mountain here on X on YouTube and
2:10:56 on Tik Tok and uh yeah go check out the
2:11:00 AI Salon um I think is it tomorrow
2:11:04 Source Camp's got her business meeting I
2:11:06 think that's right um so so go check out
2:11:09 the business Guild yes uh and then
2:11:12 Saturday Pate is doing demystifying AI
2:11:15 um I forget what the topic is but but uh
2:11:17 if you want to understand how this stuff
2:11:19 works those sessions are really
2:11:21 important
2:11:22 all right and that's it peace out to uh
2:11:26 YouTube and X talk to you soon bye
2:11:31 [Music]