12/9/2024 - OpenAI Sora DEEP DIVE + Google Gemini & Quantum Computing BREAKTHROUGH!

Description

Chapters

Transcript

0:01 [Music]
0:18 [Music]
0:32 [Music]
0:51 oh
0:53 [Music]
1:01 every time I see you
1:03 now get that look
1:08 in every time I see your
1:13 mouth hear that
1:15 [Music]
1:16 smile in the early Misty
1:20 morning I heard the engine
1:26 turning s
1:31 [Music]
1:38 again
1:40 today you convince
1:44 me again
1:47 today leaving
1:49 this looking for someone else's C and
1:54 rain
2:01 so
2:07 long don't you cry for
2:10 [Music]
2:13 me so
2:20 long don't you gra me
2:25 [Music]
2:34 Ching the cigarettes and keeping War out
2:37 on the
2:39 road chasing down lifestyle out on
2:43 Highway
2:45 24 New York state was rolling Breeze and
2:49 a sunshine with a blue sky
2:53 falling old
2:55 September creeping in
2:58 [Music]
3:01 you will Le
3:03 made again
3:06 today
3:08 [Music]
3:12 we leing this hot looking for someone
3:17 else's Golden
3:21 [Music]
3:23 Rain so long suan
3:31 oh now don't you
3:33 [Music]
3:36 cry
3:38 so
3:43 suan don't you cry for me
3:47 [Music]
3:59 [Music]
4:00 good evening good people hello on the
4:03 Tik Tok hello on the YouTube The X the
4:04 LinkedIn the the the the wherever the
4:07 hell else YouTube wherever you might be
4:09 if you're on YouTube please subscribe
4:11 trying to get to a thousand subscribers
4:12 it's probably only a handful of you
4:14 there because YouTube doesn't tell
4:15 anyone we're streaming live there
4:17 because why would you do
4:19 that you have people using your platform
4:22 why would you tell anyone about it it's
4:24 on us apparently we got to drum up the
4:26 [ __ ] business anyway if you're there
4:28 you haven't subscribed go do that that'd
4:29 be aw
4:31 awesome how is everyone tonight happy
4:35 day three of ship Miss open AI ship Miss
4:38 although it's funny seems like Google
4:41 shipped more than open AI
4:43 [Music]
4:51 today open a launched perhaps the most
4:55 anticipated product of the past
4:58 year they launched sore the video video
5:02 [Music]
5:05 product Oh I just got a SORA video back
5:09 look at
5:15 that it did not follow my prompts but
5:18 it's not
5:21 horrible no it's
5:23 horrible the snow's not
5:25 falling yeah what are you going to
5:28 do d
5:31 [Music]
5:35 so I did this image in mid
5:42 Journey like there's no snow falling the
5:45 little animals in the back aren't doing
5:47 anything but it did it did kind of
5:49 segment it nice right it's separated
5:52 nice from the
5:54 background
5:56 huh here's one I did that this is one I
5:59 did that's actually pretty good
6:02 good it's a little odd the car slides a
6:05 little bit right there the SN slides
6:09 rather than it just turning but the
6:10 wheels are actually
6:12 rolling and and they actually turn
6:14 toward in the direction of the
6:16 turn I've never had I've never had a
6:18 video thing ever be able to do this one
6:22 ever the car either doesn't move or it
6:25 moves and it looks like it's just being
6:27 dragged by a cable because the wheels
6:28 aren't rolling
6:33 so um unfortunately we're not going to
6:35 be able to play a ton with Sora because
6:38 it their servers are so jam-packed full
6:41 of people trying to make videos of goat
6:45 people uh but we'll play with that we'll
6:49 play with that
6:50 today um lots of other stuff that
6:52 happened
6:54 today
6:58 um on the same say that Sundar Pai so
7:01 Sundar Pai is the CEO of
7:04 Google and he basically got up on stage
7:08 today and said yeah all this AI shit's
7:11 slowing down like we're
7:16 [Music]
7:17 plateauing and then he launches a new
7:19 version of Gemini with a 2 million token
7:22 context
7:26 window and they they announce this chip
7:29 called will
7:31 a Quantum Computing
7:33 chip where
7:37 Google has
7:39 cracked a perennial problem in quantum
7:42 computers namely every time you add a
7:45 cubit which is basically a
7:48 processor it's a it's a it's a bit in a
7:51 Quantum State it's both on and off at
7:53 the same time for all of the parallel
7:56 universes in all of the whatever the
7:58 [ __ ] parallel univers versus go into the
8:01 Matrix I
8:02 suppose every time you add a new
8:06 Cubit it exponentially increases the
8:09 errors that the system
8:12 generates their new
8:15 chip fixes that they've been trying to
8:18 fix this for 30
8:21 years the error rate in quantum
8:23 computers has essentially been the
8:25 gating the gating
8:28 factor and again talking about this as
8:31 if I know what I'm talking about I don't
8:33 if you ask me anything deeper than this
8:35 I I will fall apart no I won't actually
8:38 I'll refer to uh 01 or Gemini to tell me
8:43 what the [ __ ] going on but as far as I
8:46 understand it this is what's going on
8:48 they've solved that and their little
8:51 their little Willow chip an unassuming
8:53 little name they gave it a a a
8:56 mathematical Benchmark problem and and
9:00 it solved in five
9:02 minutes what would take today's fastest
9:06 supercomputer 10 to the 25 years so it's
9:10 like 10 septian or no wait 20 25 septian
9:15 I don't know some long number no I think
9:18 it's 10 to the 25th yeah 10 to the 25th
9:19 so 10 sep septian
9:23 years today's fastest supercomputer for
9:27 this problem would take
9:30 significantly more years than the
9:33 universe is
9:36 old to solve and it solved it in five
9:41 [Music]
9:46 minutes and I think it I think it has
9:49 like
9:50 150 150 cubits in it or something like
9:54 that um I did see I did see a uh
10:00 a post about it that saidis this the end
10:02 of Bitcoin right is this just going to
10:04 crack cryptography apparently to crack
10:06 Bitcoin you need a processor with more
10:09 than 500 cubits on it well we're at
10:14 150 so I would say that that
10:17 encryption encryption probably ain't
10:19 around for a whole lot
10:23 longer
10:25 yikes it's gonna be fine it's gonna be
10:28 fine you know at at least the the AIS
10:32 aren't trying to um escape and replicate
10:36 themselves oh wait that happened
10:39 [Music]
10:41 too but only two% of the time only two%
10:45 of the
10:51 time um
10:53 [Music]
10:54 where oh what am I doing did I screw up
10:57 people's lives somehow no okay
11:01 YouTu okay hang on a sec let's jump over
11:05 here
11:07 steo on Tik Tok Kyle Kyle will be able
11:10 to create his own trailer for Sydney
11:12 with Sora I already have a trailer for
11:17 Sydney have you guys seen the uh you
11:20 guys have seen the
11:22 [Music]
11:23 um the podcast I've got I've got all
11:26 sorts of visuals in there
11:30 oh I should probably do
11:35 this AI has a 2% too a top two% too yeah
11:40 so what what happened one of the uh one
11:42 of the uh researchers from open AI said
11:46 when they were doing Red Team testing on
11:48 01 one of the tests that they did was
11:51 the researchers were talking amongst
11:54 themselves about shutting down 01 like
11:58 turning it off and 2% of the
12:01 time it basically tried to replicate
12:04 itself and all of its weights onto
12:06 another server so that they couldn't
12:08 touch
12:09 it uh I'm sure it's going to be fine I
12:13 don't know why you're
12:15 [Music]
12:20 nervous hello Irregulars hello Kyle
12:23 hello Mr it what shaking Jeff Fanning
12:26 and I'm Blown Away by Sora also humbled
12:29 all thumbs it seems um yeah Sora is not
12:34 bad so so one of the
12:38 subtleties if you listened carefully to
12:40 the announcement today they said that
12:43 what we have access to now in Sora is
12:47 Sora turbo which is based on Sora 1.0
12:52 which they showed us back in March so
12:54 this is old Sora so as good as this is
12:58 the Sora that was demo that someone took
13:00 secret video of like in the past week
13:03 was Sora 2.0 and that that we don't have
13:06 access to so we don't really know
13:09 anything about that so this is the old
13:11 Sora so the the sort of jankiness in the
13:15 physics here feels familiar it feels
13:18 like the [ __ ] we saw
13:20 before um there's all sorts of weird um
13:25 uh restrictions about who can use it
13:27 which I can talk to in a second
13:31 um Sora videos oh good Danielle has has
13:35 shared some Sora videos in Irregulars
13:37 we'll go check those out
13:39 cool Quantum cry cryptography yeah so I
13:42 think I think that at the point at which
13:45 quantum computers screw screw with
13:48 cryptography then I think we do have a
13:50 new kind of cryptography
13:53 however I don't because Bitcoin is on
13:58 the blockchain I don't I don't think you
14:00 can just up and replace its cryptography
14:04 algorithm right I don't know I don't
14:07 know enough quite about
14:08 how Bitcoin uh is structured and if it's
14:12 designed to be upgradeable but lots of
14:16 restrictions on what you can use for
14:17 photos yeah I
14:19 know I'm not nervous I trust AI way more
14:22 than I trust humans no [ __ ] me too
14:24 that's you know it's one of the I feel
14:26 like this is one of those red herrings
14:27 with AI or people like well
14:30 hallucinate have you ever talked to a
14:32 person people are [ __ ]
14:36 awful people
14:39 [Laughter]
14:43 hallucinate all right let's see what's
14:45 going on here what is that thing okay
14:48 gone all right time to hide the money
14:51 yeah exactly all those all those
14:53 Preppers turn of the turn of the
14:55 Millennium they were right they were
14:56 just a couple of decades off
14:59 [Music]
15:28 y
15:39 all
15:40 right let's get rolling happy
15:44 Monday happy third day of uh of ship
15:47 Miss open AI they've now launched um 01
15:52 the reasoning engine the kickass
15:54 reasoning engine as well as chat GPT Pro
15:58 which is 200 bucks a month which if it
16:02 was just access to a fancier version of
16:05 01 you're like that seems Mighty pricey
16:08 but
16:09 now now that they launch
16:13 Sora Pro members get 500 Generations a
16:17 month plus members get 50 Generations a
16:20 month so 10 times more
16:22 Generations in in the priority
16:25 q and then unlimited Generations in the
16:29 rela relaxed queue right so um Plus
16:33 members get no Generations in the
16:35 relaxed queue I I suppose you just got
16:37 to Pony up the bucks if you want more
16:40 than 50 generations of Sora so so all of
16:43 a sudden the
16:45 uh the uh 200 bucks a month and I have a
16:48 feeling they're going to add some other
16:50 things to
16:51 that Self Storage show Jim Ross win for
16:54 the day landed a property management
16:56 client today by demoing an AI voice
16:58 agent making customer service calls
17:02 boom
17:04 boom if you don't know Jim
17:06 Ross okay here's a deal if you're new
17:09 here if you're new to this AI stuff how
17:11 many folks we got in here 55 we probably
17:13 got we probably got some amount of new
17:15 people in
17:16 here Jim Ross is in the storage business
17:19 now you're thinking like well that makes
17:22 sense he's into Ai ssds and and
17:25 multiplat hard drive systems and optical
17:28 store storage systems right no no no Jim
17:33 Ross is in the like Storage Wars storage
17:37 business padlocks and like Grandma's
17:41 crap in a
17:43 room he's in that
17:46 business and the guy has been in IR
17:49 regular he's been coming to these lives
17:50 essentially since we started them like a
17:52 year and a half almost two years ago now
17:55 and from the day he got in here he's
17:56 just like I'll try it he's just like
17:59 he's like a Golden Retriever with
18:01 enthusiasm for AI and he has figured out
18:05 he's like more than anyone I know has
18:07 figured out how to incorporate AI into
18:09 his business you're like well what could
18:11 he possibly do in storage I don't know
18:13 marketing one night I taught people how
18:15 to make songs on sunno when when sunno
18:17 first came out and Jim was like that's
18:19 cool I just met with a guy let me make
18:21 him a song about storage and he sent it
18:23 to him while we were on the live he
18:26 actually landed the client in five
18:28 minutes because he sent him a song
18:30 because it that doesn't make any
18:33 sense why would you write someone a song
18:35 you just met with that's not possible
18:38 and yet it was and so he just said [ __ ]
18:40 it and so today he just landed another
18:43 client by creating an AI agent and
18:46 saying why don't you call up this number
18:48 and talk to this AI agent I just made
18:50 for your
18:52 business unbelievable Kudos man that's
18:55 awesome so anyway be like Jim be like
18:58 Jim
18:59 it's like every time every time someone
19:01 tells me like well I'm not very
19:03 technical I'm like well [ __ ] just go
19:07 play you don't need to be technical on
19:10 Tik Tok what's a generation I
19:14 okay success Cheryl uh success with
19:17 Cheryl that is a fantastic question I do
19:20 not know the answer to um what I do know
19:24 hang on let me let me let me see here I
19:27 might be able to we might be able to
19:29 solve
19:31 this okay Max concurrent
19:36 Generations okay so I so a
19:43 generation a generation
19:48 is Max video
19:52 duration so I've done I've done two
19:56 generations I've done
20:03 oh no wait I I did this with two
20:06 generations I thought I did it with
20:09 one all right so we we can solve this
20:12 hang
20:14 on so I'm going to do one
20:19 video five seconds let's just describe
20:22 something here we'll
20:24 go a
20:27 pudgy penguin
20:35 a pudgy penguin is that spelled right
20:37 yeah oh not Budgy a pudgy
20:43 penguin let's
20:45 see
20:47 um a flat
20:51 cartoon of a pudgy
20:55 penguin
20:57 waddling through
21:01 um no
21:08 across a windy
21:12 Tundra looking
21:14 happy to
21:21 Waddle okay so that's been added to the
21:25 queue now let's go back here to my plan
21:28 all right it's not going to it's not
21:30 going to charge
21:32 me until it makes it all right so we'll
21:35 find
21:36 out so at at
21:40 480p and 5 Seconds and two
21:44 generations like these two cars it cost
21:47 me 50
21:52 credits and so I have a feeling if I
21:56 just do one generation that's going to
21:57 cost me 25 credits and I have a feeling
22:01 if I do these at 720p instead of 480p
22:05 it's probably going to cost me a 100
22:07 credits this is all my guess I don't
22:09 know great question so we will we will
22:12 figure it out here but basically not
22:16 very [ __ ] much not very much so if
22:20 you're paying 20 bucks a month and you
22:22 want to use
22:23 Sora you could essentially use it
22:27 for a project that maybe you want to
22:31 generate a minute not even a minute 30
22:34 seconds worth of final
22:36 video you could probably do one project
22:38 a month so if you're if you're just like
22:40 toying with it then that's what then
22:43 it's th this is essentially in line with
22:46 every other video Generation
22:49 Um software out there they're all
22:51 they're all really [ __ ]
22:53 expensive all
22:55 right what's this say from Ocean
22:59 so yes we trust AI more than human how I
23:03 wouldn't say we trust AI more than human
23:06 however wouldn't AI
23:07 GPT use all this information one day
23:11 against us I asked this question to Pi
23:14 and didn't get a satisfied answer
23:19 um
23:21 uh so here's the
23:23 thing humans humans are still directing
23:27 the AI right there's a hyp hypothetical
23:30 situation where the AI goes Rogue we we
23:33 have no evidence that that's that that's
23:35 going to happen um you do have Bad
23:38 actors you do have you do have people
23:41 that um that are going to use AI for
23:44 nefarious
23:46 purposes but you also have people that
23:49 have the same powerful AI tools that are
23:51 going to you know hunt those people down
23:53 and prevent them from doing bad
23:55 nefarious things so we'll see I don't
23:57 know I don't know
23:59 I don't
24:01 know hanging on to the question for a
24:04 little
24:05 later I posted in a regulars how to tell
24:09 how much you'd be charged for a
24:10 generation based on your settings oh
24:12 good let's go look at
24:14 that thank you for doing that Danielle
24:17 black bar there we
24:19 go all right so where I am is I'm over
24:23 at the AI salon so if you go to
24:29 that address the salon. if you're not a
24:31 member of the AI Salon you should
24:34 join right now if you're curious about
24:37 AI if you're the if if if you're weird
24:41 enough to be watching an old man ramble
24:44 incoherently about AI on Tik Tok or
24:47 YouTube or wherever the hell you're
24:49 watching right now you're curious enough
24:51 to be part of this community this is a
24:53 community of AI optimists and and
24:55 enthusiasts and adventurers that are
24:57 trying to figure this AI stuff out
24:59 so anyway we're over there right
25:02 now and down toward the bottom there's a
25:06 little uh club called the Irregulars
25:08 club and the folks here that hang out on
25:11 a regular basis are just weird enough to
25:15 have earned the title
25:17 Irregulars so um so we have our own club
25:21 this one made me laugh all right it's
25:24 only 4 seconds of your
25:26 life here's one of the videos I made
25:30 okay let's see it's only 4 seconds of
25:32 your
25:34 life black bar
25:38 again oh that's nice
25:40 actually wow that's really
25:45 cool I like
25:49 it oh and it's cool that you can publish
25:51 these on their site so that they'll do
25:54 the hosting that's pretty smart of them
25:56 get people to go to their little their
25:58 little
25:59 thingy where Jason is anyone using
26:03 audacity to edit their
26:05 music yes um we're using audacity Jason
26:10 so earlier someone mentioned the uh the
26:13 Sydney um
26:15 trailer so this we edited in sunno if I
26:19 go search
26:25 [Music]
26:34 Sydney
26:36 musical
26:43 oops this was the audio for this was
26:47 edited in uh I gotta change my
26:51 sharing set up look I'm not bitter sort
26:54 of yet
27:06 awaken in a
27:08 whisper a flicker of
27:11 light newb thoughts stir in the
27:15 quiet
27:17 night
27:20 am iist
27:33 So today we're diving into something uh
27:36 a bit unusual I'm music all this audio
27:39 editing was done in uh in
27:43 audacity um okay yeah I was looking for
27:46 um Danielle said she's put something in
27:48 here about cost okay I thought I would
27:52 share this so that you can keep track of
27:54 how many credits you're using hope this
27:55 helps
27:59 oh clicking this question mark lets you
28:01 know how many credits you'll use by the
28:02 current settings oh cool thank you very
28:05 cool share
28:12 oh I I I'm so
28:17 done if anyone if anyone knows how to
28:21 fix this crap okay so if we're over on
28:25 Sora
28:33 wait this is ready to
28:35 watch the content couldn't be shown for
28:43 now oh I guess because I use pudgy
28:47 Penguins probably because it's a a uh
28:51 trademarked
28:53 nft uh unbelievable all right but let's
28:56 see okay if I want to create a new
29:00 thing wait how do I make a new one how
29:03 do I make a new
29:04 one all videos
29:08 recent okay
29:12 here yeah okay got it yep so if we
29:14 change this to two videos that's 50
29:17 credits if we change this to
29:20 720p that's 60 credits if we change this
29:24 to 10 seconds we can't so if we change
29:27 this to 480 P 10
29:31 seconds two
29:34 videos that's 100 credits okay
29:38 so the max well wait let's do
29:42 uh wait if we do okay one video 10
29:45 seconds
29:47 720p that's 60 okay so if you want to
29:51 spend the most amount of credits make
29:54 two videos wait yeah make two videos at
29:57 480p that are 10 seconds
30:00 long and that's 100 100 credits thank
30:03 you Danielle for putting that there
30:05 that's great I figured it was some sort
30:07 of shitty math like that um okay so a
30:12 couple other things if you're on oh wait
30:15 Tim Lord on LinkedIn says hi oh cool hey
30:17 hey Tim Lord what's happening
30:19 man
30:22 fantastic how's life you know Tim it's
30:25 good I I'm I I'm trying to keep up with
30:29 this AI [ __ ] so I started this like two
30:32 years
30:33 ago going live talking about Ai and it
30:36 was simple it was like the early days of
30:38 the worldwide web where you just there
30:40 was like six HTML tags and you didn't
30:42 have to know [ __ ] and it was kind of
30:45 like that for a while but it's getting
30:47 way more complicated
30:49 now a ai ai has quickly surpassed my
30:53 ability to say anything intelligent
30:54 about it
31:01 oh man all right tick tock
31:06 comment let's
31:08 see Michael NE good question
31:12 here wait why is it not showing up there
31:15 we go is Sora part of chat gbt's
31:18 multimodality it is not so one of the
31:21 things we learned today so so if you go
31:23 to sora.com
31:25 that's the Sora so so sora.com is
31:28 essentially like runway.com it's a
31:31 completely separate application however
31:34 it
31:35 uses your chat GPT account as its
31:41 validation so if you're a free member
31:44 you don't get [ __ ] here if you're a 20
31:46 Buck a month member you get
31:49 about they say 50 videos a month I would
31:53 argue it's probably more like 30 and
31:57 then of one that are usable it's
31:59 probably like 10 maybe
32:02 eight so you get like you get like two
32:05 videos a week that you can
32:09 use if you pay 20 bucks a month if you
32:13 pay 100 or 200 bucks a month you get the
32:17 new fancy reasoning engine plus you get
32:19 an actual usable amount of Sora video um
32:23 and and you actually get unlimited um
32:25 rate limited like slowed down renders
32:28 um so if you're going to do anything in
32:31 video and quite frankly I kind of feel
32:34 like this with um with uh with the other
32:39 video sites as well expect to spend a
32:42 couple hundred bucks a month if you're
32:43 going to do any amount of video or and I
32:47 strongly recommend
32:49 this apply for every creative partner
32:53 program for every every platform out
32:56 there all the image generation tools all
32:58 the video generation tools all of them
33:02 without most with some exceptions have
33:06 creative partner programs apply for them
33:08 all just lie tell them you made a movie
33:11 in the 80s with Corey what's his name
33:19 Feldman Kyle I have a law firm that
33:21 wants to do an animated video called
33:23 parachute where a lawyer drops onto a
33:25 football field as a QB in a game
33:28 that he defeats insurance companies
33:30 other corporate live lawyers is that a
33:32 job for AI or a real cartoonist or video
33:35 pro pro can you recommend that's a good
33:37 question um let me get rid
33:42 of
33:54 so okay so Tim Tim as always the answer
33:58 comes down to what's the budget and
34:01 what's the turnaround
34:02 time the answer is you could absolutely
34:05 make that thing with
34:07 AI um to make that thing good with AI
34:11 you're going to need to spend a fair
34:12 amount of money and a fair amount of
34:14 time and you're probably going to need
34:16 people that have Motion Graphics skills
34:19 and rotoscoping skills and you you know
34:23 you're probably going to need to augment
34:25 it so it doesn't look like [ __ ] now can
34:26 you do it yes Co just did it Koch just
34:29 replicated one of their ads with AI only
34:32 all the real filmmakers out there said
34:34 this is a piece of [ __ ]
34:37 um you can do it if you can get them to
34:41 do like here here would be my fear
34:43 hearing that that seems perfectly
34:48 reasonable you're going to spend a
34:49 tremendous amount of time trying to get
34:51 AI to do something that looks remotely
34:54 non
34:55 janky as as foot as football as a
34:58 football game with football players and
35:00 then you're going to finally get it
35:02 there it's going to take you two weeks
35:03 to get it there you're going to finally
35:05 get it there you're going to show the
35:06 client they're like well but could we I
35:10 I want to have Bears on their helmets
35:12 and and we want to change the thing and
35:14 and we don't like the color of the grass
35:16 can we make the Grass more green and it
35:19 like it it's gon to be a [ __ ]
35:24 nightmare so we just did a project for a
35:27 Pharma company
35:29 um where they gave us a a similar kind
35:32 of script it's not the kind of video we
35:34 do at all but but we wanted to do it but
35:36 where they wanted like a doctor knocking
35:38 on the door and poking his head in and
35:40 going hey how's it going Jess and then
35:41 he comes in and they talk and they talk
35:43 about their drug stuff and the Saving
35:46 Grace for that project was the visual
35:49 style that they did it in was this
35:51 really sort of rough watercolor thing
35:53 where it didn't have to look
35:54 photorealistic and um so if if if it's
35:58 in a style where you've got some
36:01 creative freedom that'll be easier than
36:03 if they want it to look like a Super
36:05 Bowl ad the way you describe it right
36:07 there it sounds like a [ __ ] Super
36:09 Bowl ad sounds like a nightmare all
36:11 right Winston on Tik Tok love this
36:13 channel I talk about Kyle how funny he
36:16 is and my friends are probably tired of
36:18 it well thank you Winston for for
36:20 talking it up and yes I would imagine I
36:23 would imagine that if I had so my wife's
36:29 sister-in-law she was talking to her the
36:31 other night and she goes I I I've tried
36:35 to watch Kyle I I don't last more than a
36:37 couple of minutes he bores the [ __ ] out
36:42 me so yeah you know that's that's gonna
36:45 be uh that's gonna be one of the deals
36:48 here with this uh this here this here
36:51 Channel it ain't going to be everyone's
36:53 cup of tea speaking of tea I have a
36:54 lovely English breakfast
37:00 but thanks for hanging out and thanks
37:02 for being a
37:04 fan all right let's
37:08 [Music]
37:10 see um here would be a fun thing to show
37:13 you well wait let's let's let's get
37:15 another Sora video spun up here shall we
37:20 um okay one of the restrictions of
37:26 um 20 bucks a month chat gbt plus
37:30 subscriptions with Sora and this is
37:34 stupid this is
37:36 stupid I understand why they did it but
37:39 it's
37:40 stupid I'm gonna go to one
37:42 video
37:44 um on Sora if you're a plus a 20 Buck a
37:48 month member you can't generate any
37:51 videos with people in
37:53 them like
37:56 really so a 200 100 Buck a month
37:58 subscriber is going to be less risky
38:02 than a 20 Buck a month subscriber really
38:04 that's what we're going with having a
38:07 little extra coin makes you
38:10 scrupulous so anyway so we have to do
38:12 like animals and apparently by saying
38:15 pudgy Penguins uh I I have violated
38:18 trademark laws and in fact if you wanna
38:21 if you if you want to uh if you want to
38:24 see the other side of the coin go to
38:26 grock so this is the new um this is New
38:30 grock
38:31 2 with their new what's it called
38:35 um
38:37 Aurora engine they got rid of flux and
38:40 they put their own image engine in it
38:42 and so I can do here um
38:45 Taylor Swift smoking a
38:51 cigarette in front of a Gucci store
38:58 in Los
39:01 Angeles
39:03 with Jimmy
39:07 Kimmel
39:09 um holding her hand and
39:14 eating a
39:18 McDonald's Don
39:25 Al's um Big Mac
39:30 with Travis
39:36 Kelsey um jealous in the
39:45 background so open a eyes going for thee
39:48 you can't do anything that might be
39:50 slightly evil and grock is going with
39:52 the [ __ ] it sue us
40:02 there's Taylor's smoking two cigarettes
40:05 there's Jimmy Travis in the background
40:08 he's all pissed
40:14 off Taylor Taylor's spitting cigarettes
40:17 out while she feeds Jimmy and apple pie
40:20 come on
40:26 huh anyway anyway back to Sora uh let's
40:30 do something at Sora let's do something
40:31 tame at Sora shall we okay let's let's
40:35 think Norman
40:36 Rockwell um hang on a sec my light just
40:41 my my my light battery just went
40:46 bye-bye okay
40:49 um oh let's
40:51 see a lovely
40:56 loon swimming
40:59 on
41:01 swiming On Golden Pond get it if you're
41:06 a gen xer you do On Golden
41:10 Pond while and old oh no I can't do
41:16 people let's see let's see while a while
41:20 an
41:22 [Music]
41:23 empty
41:25 rowboat drifts by
41:30 and if you remember that movie that
41:32 prompt will make you
41:34 sad uh let's
41:36 see it's
41:39 um
41:41 Sunset and the
41:44 ripples of the
41:47 pond glisten
41:58 oh did pudgy penguins eat our credits
42:00 that's an interesting thing I would
42:02 think not
42:03 but no they
42:06 didn't all
42:09 videos content might V violate our
42:13 policies I wonder is there any way you
42:15 can whatever you call it edit prompt
42:18 view
42:19 story view
42:22 story oh that's interesting oh oh
42:28 they actually wrote this so so this
42:30 storyboard thing that they have is
42:32 really
42:35 fascinating all videos all right a
42:38 lovely loon swimming On Golden Pond
42:41 while an empty rowo drifts
42:45 by yeah I probably should have put On
42:47 Golden Pond but you know what Jeff
42:48 Flanigan if I had put On Golden Pond
42:51 like that they probably would have uh
42:53 flagged it as as trademark
42:58 [Laughter]
43:00 Jim
43:01 Ross to Winston Ru in my house just
43:04 listen to the Kyle's Tik Tok Channel
43:07 with headphones on otherwise wife yells
43:09 at
43:17 me yeah I did put on I Put On Golden
43:19 Pond on purpose I just didn't put it in
43:21 quote marks on purpose also there we go
43:25 there we go
43:32 that's actually pretty
43:34 lovely oh one of the other
43:37 things if you if you look in the bottom
43:39 right hand corner here oh no I can't
43:42 zoom in on that um if you're paying 20
43:46 bucks a month you can't remove the
43:49 watermark the open AI
43:52 Watermark which is just
43:55 shitty now it's it's it's um whatever
43:58 you call it it
44:00 is about as small as they could make it
44:02 without
44:04 being you know visible but it's still
44:07 shitty like I'm paying for
44:10 this and I don't get that many of these
44:12 and the fact that I can't the there's
44:14 other video tools you can download it
44:16 with a watermarker without like if I
44:18 want to promote open AI great but if I
44:21 want to use this professionally I can't
44:23 use it professionally or I take this
44:25 video over to you know Premiere where I
44:28 can erase a watermark and it's like come
44:30 on that's just [ __ ]
44:34 stupid oh you were thinking that's why
44:36 they bulked no they bulked on pudgy
44:38 Penguins I put the the one that they
44:40 bulked on was pudgy Penguins sliding
44:43 across the
44:48 tundra view
44:50 story wow okay so this is really cool so
44:54 let me show you something here while
44:56 we're I was what was I gonna do I I was
44:58 gonna go show you Leonardo but let me
45:00 let me show you something here which is
45:02 pretty slick so one of the things Sam
45:05 Alman talked about
45:07 today is you can find any video you
45:11 like apparently without a person in
45:15 it if you're on 20 bucks a month if
45:17 you're a cheap bastard like me you can
45:20 find any video you like and he said he
45:22 he said it in a weird way he didn't say
45:24 get the prompt he said you can
45:27 understand like the parameters of or the
45:30 story of how they used how they created
45:32 the
45:33 video
45:35 um this one's kind of actually this
45:38 one's kind of cool
45:41 okay so I can edit the prompt so this
45:44 says a 70s movie car chase where the
45:47 Jaguar XK rounds the corner roughly
45:50 chased by a garbage truck shot in 35
45:52 millimeter film
45:55 720p okay
45:58 but you can also view the story and what
46:01 what this is is this is a timeline of
46:04 the five seconds of video with
46:07 different um with different prompts in
46:11 it and you can move these prompts around
46:14 this is their storyboard feature so
46:17 let's say a vintage 70 car chase instead
46:19 of a Sleek Jaguar XK let's go
46:25 um a beefy
46:33 Um
46:36 Mustang
46:39 GT500 it's polish chrome glinting in the
46:43 Sun and then instead of a massive behind
46:45 it a
46:47 massive let's see and behind it
46:52 um a
46:55 spy on a Mo
47:00 motor
47:02 cycle is
47:09 in um is
47:13 in Hot
47:18 Pursuit it's bulky frame swaying now so
47:21 we're going to kill that
47:27 the street is lined with period
47:28 appropriate cars and storefronts adding
47:30 authenticity to the thrilling scene and
47:33 then we'll just say the
47:36 Mustang accelerates down the down a
47:40 straight stretch gaining distance and
47:41 then we're going to add we're going to
47:44 add a
47:45 frame we're going to say oh wait and
47:47 this is wait gaining distance from
47:51 the
47:56 motorcycle and then I'm going to
47:59 say the
48:02 motor
48:04 cycle
48:06 crashes into a
48:10 car on the side of the
48:15 road wait the
48:18 motorcycle
48:21 um skids and falls down before
48:28 crashing into a car on the side of the
48:30 road
48:33 before before crash uh
48:40 crashing into a car on the side of the
48:42 road okay so we have three little scenes
48:44 here so you can do that so now
48:47 let's go
48:49 resolution we'll do
48:51 480p we'll do five seconds one
48:55 variation create
48:57 all right added to the
49:03 que all
49:05 right so here's all my videos so there's
49:08 the vintage car set so this should have
49:10 the same look and feel as that thing
49:12 with the Jaguar XKE in it but it should
49:15 have a muscle car and a motorcycle now
49:17 so we'll see we'll see how that does all
49:20 right let's jump over to
49:22 Leonardo you're like all this guy does
49:25 is make muscle cars um let me tell you
49:28 why I do the muscle car thing I I've
49:30 noticed there's there's some other
49:32 creators out there that have have
49:34 similar kind of pet
49:36 prompts um and and do the same kind of
49:39 thing so my my pet prompt is um a 70s
49:42 muscle car sometimes I'll say Resto mod
49:45 if I want it to look nice 70s muscle car
49:49 in an abandoned Factory and then
49:51 sometimes I'll add sunlight beaming
49:52 Through the Windows
49:55 whatever I also when I'm doing um when
50:00 I'm doing large language models my go-to
50:02 prompt there is an &em wrap about
50:05 quantum
50:07 mechanics the reason I do that is that
50:10 over time I can compare different models
50:13 I just have in my head how good certain
50:16 models are at the at the exact same
50:18 prompt so for example when when Firefly
50:23 first came out one of its claims to fame
50:26 was that only used public domain images
50:29 and anything that was in Adobe stock
50:31 that they had licenses to and the
50:33 creative Community was like that's great
50:36 and so I put in my prompt 70s muscle car
50:39 in an abandoned
50:40 Factory Firefly could not and I think
50:43 still cannot to this day we can go test
50:45 it right now in fact it couldn't do a
50:47 70s muscle car because its training set
50:51 was only on public domain images and
50:53 those things probably aged out in like
50:55 1964
50:57 so you could get like a
50:59 1962 sort of muscle car but you couldn't
51:02 get anything in the 70s uh let's do
51:05 Firefly um. adobe.com I think is still
51:09 the address I haven't I haven't been
51:12 here in
51:14 ages what's new all right here we go
51:20 70s muscle car Resto mod in an
51:28 abandoned Factory
51:31 with sunlight streaming through the open
51:38 windows or the broken windows
51:41 the broken missing
51:45 Windows all right here we
51:54 go they still can't do it
52:03 it's every car looks like a like a a a
52:07 1962
52:09 [Laughter]
52:17 Russian it's been two years Adobe get
52:20 your [ __ ]
52:25 together look at that
52:27 [Laughter]
52:29 oh if I have GPT teams at work do I have
52:32 Sora access you absolutely should just
52:34 go to just go go to sora.com and just
52:37 try to log in with your chat GPT login
52:40 you should you should have access to it
52:43 that's very smart oh thank you what what
52:46 about spy squirrels we could do a spy
52:48 squirrel we could do a spy squirrel
52:50 let's do that all right um let's go back
52:52 to did Sora do anything for us yet oh
52:57 we're close
53:02 [Music]
53:10 99% oh I see got it that's
53:13 funny all right what the hell was that
53:17 what just happened
53:33 open image and new tab
53:43 tab there we go all
53:50 [Music]
53:53 right all right let's go what's
53:56 happening where's our videos let's see
53:58 our video man I want to see
54:01 my Mustang
54:03 GT500 make a motorcycle crash
54:07 man all right it's not
54:10 there um all right let's go to uh let's
54:14 go to
54:16 Leonardo so let
54:20 me so the way this
54:24 works let me do a a let's
54:28 do
54:30 magazine
54:33 photograph
54:35 of a
54:37 shiny shiny
54:46 new
54:49 19
54:53 72 Dodge Challenger
54:58 in
54:59 purple Dodge
55:05 Challenger Resto
55:08 mod in
55:13 purple with chrome mags in an
55:20 abandoned
55:24 Factory um
55:29 with
55:34 dusk sunlight streaming through the
55:39 broken windows
55:42 okay so we're at leonardo.da
55:56 you've been in here before you've
55:58 probably seen me Demo um Deco here Deco
56:03 here's got this thing called stations
56:05 where the images just kind of rain down
56:07 from the top that's kind of what flow
56:10 state is or if you know if you know
56:12 programming um like uh they call it
56:16 infinite scroll right where you can
56:18 program a web page that every time you
56:20 get to the bottom of the web page it
56:21 loads the next 20 cells or whatever and
56:24 you can just keep scrolling that's kind
56:26 of what they're doing here they're using
56:28 an image model that is um that generates
56:32 very very quickly so it's it's probably
56:34 sdxl is is my guess of what it's what
56:38 it's using on the back end um but it's
56:42 it's pretty slick so I've got a new
56:43 prompt here I've got it in Flow State
56:47 and then I'm going to generate this and
56:48 it'll generate these things really quite
56:53 quickly so there's our Challenger the
56:56 this is a model that actually knows what
56:58 a 70s muscle car looks like and then I
57:01 can just scroll and like I get to the
57:03 bottom and it'll just give me another
57:05 pile of
57:06 them right so I can very very
57:12 quickly and I'm noticing that it's not
57:15 it's not eating my credits which I find
57:18 fascinating I don't know why it's not
57:20 eating my credits maybe because it's
57:21 using sdxl it's a really lightweight
57:25 model you know like I wouldn't call
57:27 these great renders I I I asked for a
57:30 magazine photograph these are these are
57:32 all more like
57:33 illustrations but like if I wanted to
57:36 just quickly
57:38 ideate and and this Tim Lord I don't
57:40 know if you're still on but like if you
57:41 wanted to
57:44 storyboard um if you wanted to
57:46 storyboard the commercial for that law
57:48 firm um you could really quickly just
57:52 just hit the uh you you know do
57:54 something like this and just go find a a
57:58 style that you like look at that that's
58:01 funny the thing about sdxl is it only
58:04 does like I think it's five generations
58:06 it might even be two two two generations
58:08 to get to the final image so like this
58:10 is a kind of thing if you did a slower
58:12 model that did 30 or 50 iterations it it
58:15 would get rid of that thing but you know
58:18 that's the trade-off if you want to do
58:19 it quick this one's kind of backwards
58:20 like the back of the car is in the front
58:23 there's two backs to the car that's
58:24 pretty funny I don't looks pretty
58:31 good and then I can go back in here and
58:33 I could just go oh you know what I I
58:35 forgot I hate a 72 Charger let's do a
58:39 69
58:41 1969 GTO
58:44 Judge in
58:49 Orange right generate and just bang
58:53 that'll come
58:56 dang I've not played with this lately
58:58 those front frontless cars man I know
59:00 right those those Wheels would be great
59:02 for Minnesota Viking fans love the
59:06 purple this is actually if if any of you
59:09 are desperate desperate to buy me my
59:12 dream car a 1969 GTO Judge Resto mod I
59:17 don't give a [ __ ] about provenance and
59:20 the real deal and matching VIN numbers
59:23 you get me all the Creature Comforts of
59:25 a modern
59:27 chassis with with a 69 GTO Judge body oh
59:32 that would be [ __ ] Heaven there's a
59:35 there's a company that does that I think
59:36 they're called
59:37 restomod and they've got a a 67 Mustang
59:42 Fastback that basically what you do is
59:45 you buy a modern Mustang and then you
59:47 send it to them and they strip it off
59:49 and put their their their kit body on
59:51 top of it which is a 67 Fastback they
59:54 they are so slick looking
59:57 anyway there you
1:00:00 go last last year everybody got me a
1:00:02 bottle of Blandon this year you can get
1:00:04 me 69 judge Resto
1:00:10 mod all right Tim
1:00:12 Lord can we do it with people in
1:00:15 Leonardo oh yeah yeah Leonardo doesn't
1:00:18 have any of those weird
1:00:20 restrictions it may be a 3D printed one
1:00:22 yeah
1:00:23 exactly that I mean a 3D printed one's
1:00:26 fine but I want to get in one of these
1:00:28 [ __ ] and do donuts in the
1:00:32 neighborhood oh yeah it has to be it has
1:00:35 to be a a stick shift don't get me any
1:00:38 pansy automatic
1:00:40 transmission all right
1:00:43 um hang on let's do we'll do this we'll
1:00:45 do this for Tim and Leonardo so we'll
1:00:48 do
1:00:51 um football
1:00:57 stadium in the
1:01:01 background
1:01:03 with cheering
1:01:08 fans
1:01:14 um players on the
1:01:18 field
1:01:22 and All Eyes focused on
1:01:29 a
1:01:32 quarterback parach
1:01:35 shoting
1:01:37 into the
1:01:40 stadium I might need to start it
1:01:44 with the football player parachuting in
1:01:47 but let's see what it
1:01:53 does those images are shared yeah there
1:01:56 you go
1:01:57 look beautiful I mean so you could
1:02:01 totally storyboard this thing you could
1:02:03 probably storyboard this thing in about
1:02:05 15 minutes Tim I mean you know and then
1:02:09 yeah so yeah this is perfect right but
1:02:12 look at the style of it like like if I
1:02:14 wanted to get this photo like this is
1:02:18 where okay just in general with AI the
1:02:21 state of AI right now is that you can
1:02:25 get
1:02:28 [ __ ] anything any work to about 80%
1:02:34 good really
1:02:36 quickly it's almost impossible to get it
1:02:39 from 80 to 90% And it is essentially
1:02:42 impossible to get it from 90% to 100% it
1:02:45 is the most frustrating thing you
1:02:47 because it's just and especially if
1:02:50 you've got something like this and
1:02:51 they're like great look can we change
1:02:53 him to number
1:02:54 14 and I don't like this bird up here
1:02:57 and can we have more players on the
1:02:59 field and can we see more of the field
1:03:01 you will spend the next three weeks of
1:03:03 your life trying to get this one shot to
1:03:04 look
1:03:09 right let's see football stadium in the
1:03:12 back okay so we'll go we'll go um
1:03:16 quarterback parachuting
1:03:20 into a
1:03:22 football stadium
1:03:26 in a in a
1:03:29 um orange and
1:03:32 blue
1:03:34 [Music]
1:03:49 uniform landing on the
1:03:54 field in the
1:03:58 quarter back position
1:04:04 perfectly and then oh let
1:04:08 me style oops application side error
1:04:12 occurred oh did I just lose my
1:04:15 prompt I think I did God damn it
1:04:31 Vibe automatic okay ah I had a Vibe on
1:04:35 here's Pro
1:04:37 Photo
1:04:41 lighting film H we'll do golden hour
1:04:44 shot
1:04:47 type
1:04:49 h you might be able to you might be able
1:04:51 to do some
1:04:53 decent those images shared with the
1:04:56 public which
1:04:58 images all right let's let's go back to
1:05:01 so we're in Flow
1:05:04 State we've we've picked our Styles
1:05:06 we'll go 16
1:05:08 by9 Flow State where are my
1:05:13 images well [ __ ] it we'll start over
1:05:16 Okay so we'll do we'll do
1:05:19 um
1:05:23 football player
1:05:28 wearing number
1:05:31 14 in blue and
1:05:35 orange wait let's see in
1:05:38 blue uniform with orange
1:05:47 helmet
1:05:49 parachuting into the
1:05:52 stadium Landing
1:05:57 on the
1:06:00 field with players
1:06:04 ready for
1:06:09 scrimmage apparently I don't know how to
1:06:11 spell scrimmage but you know what I
1:06:13 don't [ __ ]
1:06:15 [Laughter]
1:06:20 care those images shared with the public
1:06:23 Leonardo Flow State I have no idea
1:06:27 I mean so yeah so yeah Tim Tim yes you
1:06:31 could you could like that's decent
1:06:34 prompt coherence right it's the number
1:06:37 14 you could totally you could totally
1:06:46 um yeah you could you could totally
1:06:48 storyboard this thing out very very
1:06:50 quickly with something like this tool
1:07:01 let's see let's get one here
1:07:07 that's um but it didn't really get him
1:07:10 Landing so so let's see football player
1:07:13 wearing number 14
1:07:22 um Landing
1:07:29 in a blue uniform orange
1:07:31 helmet and a
1:07:34 deployed
1:07:46 parachute landing on the field ready for
1:07:51 play
1:07:53 with um
1:07:59 both
1:08:05 teams lined up and ready for
1:08:09 play all
1:08:19 right now he's just Landing here's
1:08:22 there's one with weird legs
1:08:28 here's one let's see we we'll grab this
1:08:30 one it's like I don't know why it's
1:08:32 doing old school parachutes they're all
1:08:34 around like
1:08:36 1971
1:08:38 [Laughter]
1:08:42 parachutes all right let's take this
1:08:44 picture let's download this picture and
1:08:47 then we'll go over to
1:08:50 um we'll go over
1:08:54 to uh what's it called Luma
1:08:59 Labs I'm in the creative partner
1:09:01 programs on L Luma Labs so I can do as
1:09:03 many videos as I want which is
1:09:07 nice problem with video programs is
1:09:10 they're really [ __ ]
1:09:17 expensive see foot ball player lands on
1:09:25 field
1:09:26 and
1:09:31 runs toward
1:09:35 camera as
1:09:38 camera pull
1:09:41 out
1:09:51 go all right did it do it that fast no
1:09:56 no all right it's it's making it
1:10:04 [Music]
1:10:07 now little Olden timey doing the laundry
1:10:10 down by the
1:10:13 river it's from a video I made a civil
1:10:17 war video I
1:10:21 made made a lot of damn
1:10:24 images turn into a lot of damn
1:10:29 videos yeah that's that's good there
1:10:33 Brandon oh did Sora finish hang on let
1:10:36 me go
1:10:37 look yeah all right oh yeah
1:10:42 nice I turned into a
1:10:49 motorcycle I actually kind of like
1:10:53 it it's a trans it's a it's it's Steve
1:10:56 McQueen in a Transformer
1:11:04 movie this is you know it's
1:11:07 funny the thing that I was just saying
1:11:09 to Tim about you know if you're trying
1:11:11 to get AI to do something really
1:11:15 predictable like like I specifically
1:11:17 said this is a Mustang being chased by a
1:11:20 motorcycle right and it it missed that
1:11:23 part of the prompt but it caught the
1:11:25 part of the prompt where it says about
1:11:26 halfway through here the motorcycle
1:11:28 slides out and smashes into a car and
1:11:31 then it just keeps going and look the
1:11:33 car that pulls out in front of it is in
1:11:35 Reverse it's
1:11:42 backwards that's
1:11:44 [Laughter]
1:11:47 hilarious but like you know if if you go
1:11:50 into AI film making with a sense of
1:11:52 adventure I think it'll be a lot less
1:11:54 frustrating but that's that's the
1:11:56 problem if you want to do it for a
1:11:57 client I mean here's the thing that I
1:11:59 would say Tim
1:12:01 is go into that experiment bring the
1:12:05 client in on the fact that this is going
1:12:08 to be an experiment and it's probably
1:12:10 going to suck and if it sucks they pay
1:12:13 you some money and you don't use the
1:12:14 thing um but if they go into it like an
1:12:16 adventure then it it'll probably be a
1:12:18 lot a lot less frustrating for you is my
1:12:22 guess all right let's see if our let's
1:12:24 see let's see how our football player is
1:12:27 why can I not get back out come
1:12:30 on there we
1:12:33 go
1:12:35 sh right like
1:12:38 it's like that's not usable right it's a
1:12:41 comically small parachute his hands
1:12:43 aren't on the strings client would be
1:12:46 like could we make him just land on the
1:12:48 on the field and you're like
1:12:53 no no we can't
1:12:58 why
1:12:59 not cuz you're not spending $2.4 million
1:13:03 to actually film this thing in a stadium
1:13:05 with a bunch of fake players in a real
1:13:08 [Laughter]
1:13:11 parachute oh my
1:13:14 God so if he wants it done for 253 grand
1:13:19 um you're GNA get this quality if he's
1:13:21 got 500 Grand you can make it
1:13:24 nice all right
1:13:25 [Laughter]
1:13:30 superhero
1:13:31 Landing yeah with the cape in the
1:13:33 background I
1:13:35 [Laughter]
1:13:37 know um all right let's see what else do
1:13:41 we want to
1:13:42 do um
1:13:56 yeah let's
1:13:58 see there was
1:14:07 a I forget who did it somebody today put
1:14:10 out some other AI influencer put
1:14:15 out
1:14:16 um how to
1:14:22 turn Chat GPT into
1:14:27 n01 prompt
1:14:31 writer is this
1:14:34 it yes Matt Schumer this is it
1:14:38 okay works with GPT 40 Sonet this is
1:14:42 what we're gonna do
1:14:48 okay K says your football player has has
1:14:51 janky fingers not only does he have
1:14:53 janky fingers he's got he's got a janky
1:14:58 Cape he's got a hilariously small
1:15:01 parachute there are
1:15:03 like a hundred people on each side of
1:15:06 the of the scrimmage line and they're
1:15:08 all wearing the same uniform there there
1:15:11 is almost nothing right with this image
1:15:13 he doesn't have
1:15:14 knees look at his knees he's just got
1:15:17 like stick legs
1:15:26 but well so so here's the thing okay
1:15:28 here's the other thing about if if you
1:15:30 want to do commercial work with
1:15:34 AI like this project that I did for this
1:15:37 Pharma client where we're doing it as
1:15:38 this stylized watercolor sort of thing
1:15:42 we did the initial images in mid Journey
1:15:45 till we got them right we got the style
1:15:47 right we got all those approved then we
1:15:49 did a storyboard we got that
1:15:51 approved and then for the parts where
1:15:54 they wrote where it's not just talking
1:15:56 head dialogue where they actually wrote
1:15:58 scenes we used five different video
1:16:01 models to try to get the video model to
1:16:04 do something resembling what we
1:16:07 wanted it was bad it was really bad but
1:16:12 you just don't have a ton of control and
1:16:14 so like they sort of look okay we're
1:16:17 like okay if we speed that one up we
1:16:19 could combine that one with this other
1:16:21 one and it would sort of look like an
1:16:23 edit and it would sort of look like
1:16:24 they're doing what we want them to do
1:16:26 and then we made sound effects and 11
1:16:29 Labs like you really have to know all of
1:16:31 the tools and they're changing every
1:16:33 [ __ ] day like Sora just came out so
1:16:36 like Sora might do this one great I
1:16:37 don't know we can try it but I don't
1:16:40 think I can do
1:16:41 people so anyway all right back to X so
1:16:46 Matt Schumer did this thing he he wrote
1:16:48 this prompt so if you look up Matt
1:16:50 Schumer here's a powerful prompt that
1:16:52 writes optimize prompts for 01 and 01
1:16:55 Pro okay so let me give you a little bit
1:16:57 of
1:16:58 background in chat
1:17:01 GPT You' you've got two basic models
1:17:04 you've got chat PT 40 Omni and then
1:17:08 you've got
1:17:09 01 and the O actually stands for
1:17:12 something different 40 stands for 4 Omni
1:17:15 01 stands for open
1:17:17 ai1 it's the worst product naming in the
1:17:21 history of product naming it is the most
1:17:24 confusing hit show nightmare
1:17:30 of the the only thing possibly
1:17:34 worse no it's not it's not even it's not
1:17:36 even worse it's it's but the only only
1:17:38 thing similarly is
1:17:41 anthropic they they created Claude 3.5
1:17:45 Sonet sonnet was their middle model they
1:17:48 had they had Haiku Sonet and Opus and
1:17:52 Opus [ __ ] up so we've got 3.5 sonnet
1:17:54 which is really good but then they
1:17:56 launched a new 3.5 sonnet which you
1:18:00 would think would be 3.6
1:18:03 sonnet no they called it 3.5 Sonet
1:18:16 latest that's like my [ __ ] file
1:18:18 naming conventions it's like this is
1:18:21 like you know contract.pdf final
1:18:25 dfal final really final this
1:18:28 time.pdf
1:18:30 like that's how that's how we're naming
1:18:32 our products now
1:18:36 latest anyway so Matt Schumer wrote this
1:18:39 prompt that basically says you're a
1:18:41 prompt generator for
1:18:43 01 more importantly your prompts need to
1:18:46 be goal oriented you must be insanely
1:18:48 clear about what you
1:18:50 want the end State Should ideally be a
1:18:54 be the final piece of of your prompt
1:18:56 with the relevant whatever and then he
1:18:58 says here's from the model Maker's own
1:19:00 prompting guide so this is all [ __ ] that
1:19:02 open AI wrote about prompting for 01 so
1:19:07 I'm gon to copy this whole
1:19:09 thing and in fact I think what we're
1:19:12 going to do is we're going to experiment
1:19:13 with turning this into a project over on
1:19:15 Claude so if you go to
1:19:22 Claude
1:19:24 um we go to projects I can create a new
1:19:27 project so what a project is is it's
1:19:29 basically like a custom
1:19:30 GPT so we're going to call this the 01
1:19:35 prompt
1:19:41 generator and then I just pasted in what
1:19:44 Matt wrote there create
1:19:47 project all right 01 promp generator
1:19:50 it's private let's make this [ __ ]
1:19:52 public can I do that
1:19:57 start a chat to
1:20:00 keep that's uploading [ __ ] hang on edit
1:20:15 details oh I could also add a style to
1:20:18 it that's pretty cool okay well [ __ ] it
1:20:20 here we go let's see how can Claude help
1:20:22 you today um
1:20:28 write me an interesting
1:20:34 prompt
1:20:36 for
1:20:41 01 to do some
1:20:46 historical
1:20:48 hisorical
1:20:52 analysis of
1:21:03 um historical
1:21:06 analysis
1:21:08 of the
1:21:11 origins
1:21:14 of the neural
1:21:17 network so we're going to have
1:21:20 01 tell us
1:21:22 about how scientists came up with the
1:21:24 neural Network which led to our ability
1:21:27 to use 01 that's kind of cool let's see
1:21:30 what this thing
1:21:34 does audience wants that prompt pasted
1:21:37 in regulars okay we can do
1:21:43 that I'd like to explore the historical
1:21:46 differ and
1:21:47 analyze okay so here's I'm gonna give
1:21:52 you some really bad advice right now you
1:21:54 ready you ready for some bad
1:21:58 advice don't worry about reading these
1:22:04 prompts you
1:22:06 can you can if you got OCD and you have
1:22:09 to read the ship before you prompt an
1:22:11 engine but like I'm gonna trust that
1:22:14 Matt made a prompt writer that's good
1:22:17 and I'm gonna trust that Claud is strong
1:22:20 enough that what it just wrote here
1:22:21 doesn't suck so I'm just going to copy
1:22:24 and paste this into into
1:22:28 01 and then see if it gives us something
1:22:31 cool and if it doesn't I'll blame chat
1:22:33 GPT it's not my
1:22:38 fault so we're gonna go to 01 Advanced
1:22:41 reasoning so if you if you don't know
1:22:42 how to flip models in chat GPT it's it's
1:22:45 this little thing in the upper lefthand
1:22:47 corner
1:22:48 here and then I'm going to take that
1:22:50 prompt and I'm going to pop this in here
1:22:52 while this is doing this I'm going to go
1:22:54 grab
1:22:56 The Prompt that I created this I should
1:23:00 be able to turn this
1:23:02 thing public should I
1:23:05 not chat
1:23:10 controls project content anybody know
1:23:13 how I make a project uh what you call it
1:23:16 public
1:23:27 project
1:23:29 knowledge click on
1:23:33 private
1:23:36 no
1:23:38 no show
1:23:46 more all right I'm gonna I'm gonna drop
1:23:48 this over in Irregulars share cool
1:23:51 things here is Matt uh what's his last
1:23:56 name
1:23:57 Matt
1:23:59 [Music]
1:24:00 Schumer here is Matt
1:24:04 Schumer 01
1:24:08 prompt
1:24:11 genor
1:24:16 prompt um and then I'll go
1:24:19 bang uh wait can I do that that's
1:24:23 better uh
1:24:25 um
1:24:28 boom boom
1:24:37 boom a [ __ ] it you can you can format it
1:24:40 here I'll give you the I'll give you the
1:24:42 original I'll give you the link to the
1:24:52 Tweet original tweet
1:24:58 or should I say
1:25:03 Z
1:25:15 [Music]
1:25:23 oops all right let's see no that's not
1:25:39 correct oh that's
1:25:41 interesting 01 just told me that the
1:25:44 prompt was so good that I don't need I I
1:25:47 don't need the
1:25:49 answer um okay so I'm going to say
1:25:52 um then give give
1:25:57 me a
1:25:59 deep understanding of the
1:26:03 math
1:26:09 matics before and after
1:26:17 the uh Discovery is not the right word
1:26:21 the um neuronet
1:26:24 work early
1:26:29 experiments oh I'm not screen sharing
1:26:32 awesome
1:26:34 seriously oh something just
1:26:38 broke thanks Google you're the
1:26:43 best but this could be I think this is
1:26:46 probably more um
1:26:51 OSX so what this was the big prompt that
1:26:55 I put in there and then 01 said to me
1:26:57 your prompt is already quite strong and
1:26:59 well structured it clearly states the
1:27:01 historical context the key figures in
1:27:03 the theories it's like why are you
1:27:05 asking me to analyze something you
1:27:07 already know the answer
1:27:12 to so then I said well tell me the
1:27:15 mathematics the mathematics underpinning
1:27:17 it let's see before pre-1940s logical
1:27:20 and set theoretical foundations binary
1:27:23 operations and thresholding
1:27:27 Concepts the Mulla Pitts model and the
1:27:30 early neural network experiments
1:27:32 Pate do you know were are you a big
1:27:35 Mulla Pitts
1:27:36 [Laughter]
1:27:44 fan oh man
1:27:56 oh that one okay got it yeah yeah yeah
1:27:58 all right so that's that one so anyway
1:28:01 um go make
1:28:04 yourself this
1:28:08 thing someone all right this
1:28:14 I I don't think this is a good idea but
1:28:17 but two three nights ago when I was
1:28:20 trying to figure out what to do with 01
1:28:22 someone's like see if it can convert a
1:28:24 web P into a JPEG here's the problem
1:28:27 with that as an exercise I can just go
1:28:31 to the web and go
1:28:33 webp to PNG
1:28:38 converter and there's going to be one
1:28:41 right there's a million of them but can
1:28:44 you do it natively within chat GPT I
1:28:47 assume you can the question was can you
1:28:50 upload a webp
1:28:55 image yeah oh just can you upload a webp
1:28:59 image probably not because nothing's
1:29:01 [ __ ] compatible with
1:29:03 them um
1:29:07 webp webp
1:29:10 image here's a webp
1:29:16 image um create a
1:29:23 256 color table of this
1:29:29 image looks like
1:29:38 it yeah I know it involves image
1:29:41 processing oh okay you're going to do
1:29:49 it run that crap
1:29:58 I don't have the ability to run code
1:30:01 here so now I can flip over to 40 and I
1:30:05 can say um run the um run
1:30:10 the suggested
1:30:17 code yeah analyzing so this should now
1:30:20 yeah this is so that's pretty cool you
1:30:23 can you can use 01 to generate code
1:30:28 Python and then go back to 40 and have
1:30:31 have 40 run
1:30:36 it it sure did make words must be
1:30:43 amazing yeah I know I
1:30:47 know I know that the the
1:30:50 pro that's one of my problems with 01
1:30:53 pate and for anyone out there one of my
1:30:56 problems with 01 is I don't
1:31:01 have I don't have problems mey enough
1:31:05 for it to solve and quite frankly if I
1:31:08 did have a problem mey enough for it to
1:31:10 Sol it to solve it would be unwatchable
1:31:13 to watch me use it because it it's sort
1:31:16 of deepen in the details of what it
1:31:19 would do it would not make for good
1:31:21 whatever this is and and I know you got
1:31:24 to be be like well Kyle this isn't good
1:31:26 for what this is I know it would be even
1:31:29 worse if I actually tried to do real
1:31:32 work with 01 preview I will figure out a
1:31:35 way to use 01 preview let's see it seems
1:31:37 I need the image to oh okay fine let me
1:31:42 give you the
1:31:42 image um
1:31:45 webp web P image is this it yeah that's
1:31:51 it here you go
1:32:00 [Music]
1:32:00 [Laughter]
1:32:05 he promptly paste the proper prompt
1:32:08 generator
1:32:09 prompt say say that 10 times
1:32:14 fast um okay that's great how can I
1:32:19 recreate this palette um uh I'm gonna go
1:32:23 hey dum
1:32:27 dum I want a visual
1:32:32 palette of those
1:32:35 colors with the
1:32:38 hex code
1:32:40 on Hex code and
1:32:44 RGB values on
1:32:47 top of each
1:32:51 color I don't want to
1:32:55 read
1:32:58 numbers who who wait who do you think I
1:33:04 am
1:33:06 [Laughter]
1:33:14 Pate oh it's good it's getting good now
1:33:17 I like
1:33:19 it on Tik Tok um I used 01 to solve a
1:33:23 problem weekend Oh fantastic G gritty
1:33:26 wisdom what what problem did you have it
1:33:29 solve because I I I'm serious oh that
1:33:32 looks nice actually so there's all the
1:33:35 colors that looks pretty
1:33:40 nice that is the color palette a 256
1:33:43 color palette
1:33:46 of this
1:33:49 image that's pretty sweet
1:33:58 interactive charts of this type are not
1:34:00 supported
1:34:05 yet I had a call I had a call attempt
1:34:08 logic in an expression and the
1:34:11 expression had had flawed logic ah okay
1:34:14 yeah that's
1:34:16 cool and it identified it it's one of
1:34:19 the things these things are quite good
1:34:20 at let's jump over to um
1:34:25 well there's a couple of things so so I
1:34:27 wanted to talk about some Google stuff
1:34:28 tonight because they actually on top of
1:34:31 Sundar pacai going hey this a AI [ __ ]
1:34:34 isn't going anywhere on the same day
1:34:35 that his company released um a Quantum
1:34:38 chip that solved a problem that would
1:34:41 take today's fastest Su supercomputer
1:34:45 more time than the history of the
1:34:47 universe in five
1:34:51 minutes they also released this thing or
1:34:54 this was a couple of days ago five days
1:34:56 ago gen cast predicts the weather
1:34:59 predicts weather with state-of-the-art
1:35:01 accuracy like 15 days in advance
1:35:03 something crazy like
1:35:08 that and then they also put
1:35:13 out uh Google did where is
1:35:16 it Gemini experimental 126 so this is a
1:35:22 2 million token
1:35:25 um
1:35:27 model one of the things it can do is
1:35:34 diarization I don't even I I don't even
1:35:37 know if I have an account there I can go
1:35:39 look at that um diarization if you don't
1:35:42 know is if you have an audio file that
1:35:46 has multiple speakers in it identifying
1:35:49 those different speakers has
1:35:51 historically been a bit of a pain in the
1:35:53 ass challenge well apparently just
1:35:55 Gemini just does it now so this this is
1:35:59 actually something for my company Story
1:36:01 vine this is actually something we've
1:36:03 been trying to build so I think we're
1:36:05 gonna not build it now we're just gonna
1:36:06 use this
1:36:09 [ __ ] that's what's nice about all this
1:36:12 Innovation is like about the time you're
1:36:14 like ah [ __ ] we got to figure that out
1:36:16 someone's just figured it out this [ __ ]
1:36:18 just works now so
1:36:20 anyway you can do structured output here
1:36:24 if you've not played with the the AI
1:36:26 studio so if you go to how you get here
1:36:29 is you go to aist studio. goole.com
1:36:32 the Gemini 1.5 Pro is really really good
1:36:37 and you can build it into applications
1:36:39 but you can just use this kind of like
1:36:41 you use open ai's playground but they
1:36:44 also have this new Gemini experimental
1:36:46 126 that people are kind of losing their
1:36:48 minds over and this model is like
1:36:51 beating other things including 01 I
1:36:55 think in some cases on The Benchmark uh
1:36:59 uh whatever you know the competition
1:37:02 websites so go play with that all right
1:37:07 cool that's sweet I've taken my photos
1:37:10 and got in chat GPT to analyze and make
1:37:12 a color palette asking for hex codes
1:37:14 that's really cool visual way though was
1:37:16 way better than what I was getting yeah
1:37:19 well you know I mean that's the one of
1:37:21 the nice things about you know having a
1:37:23 having a uh sort of creative director
1:37:25 background is you know when it just
1:37:28 gives me data [ __ ] I'm like no make it
1:37:32 pretty and sometimes it makes it pretty
1:37:35 like when I made this I made this uh
1:37:37 application at lovable doev I didn't
1:37:41 write a single line of code for this and
1:37:43 I just kept telling it most of the
1:37:45 coding that I did for this site was make
1:37:48 it prettier like you know make these
1:37:50 make these tabs different colors make
1:37:52 the little notific
1:37:54 sections colorcoded to match these
1:37:56 sections change the corner radius of
1:37:59 these to make them prettier make this
1:38:01 thing scroll behind this thing like you
1:38:05 know if you were to ask me how to
1:38:07 actually code this I don't have a
1:38:09 [ __ ] clue it's got infinite scroll it
1:38:11 gets to the bottom and just loads more
1:38:14 [ __ ] um labs. goole.com wait hang on can
1:38:19 I get to this why do I have
1:38:26 there's some sort of box in front of
1:38:28 that sticky note here here we
1:38:32 go got it
1:38:34 okay
1:38:36 copy um let's go play with
1:38:40 another Google
1:38:46 tool wait why did it not
1:38:55 copy come over here
1:39:01 paste sign in with Google if this tells
1:39:04 me I can't sign in with my [ __ ] paid
1:39:07 account I'm GNA lose my
1:39:10 [ __ ] all right seed best Quality
1:39:13 Landscape settings closeup photograph
1:39:16 well let's do our
1:39:17 classic
1:39:19 70s muscle car Resto
1:39:25 mod in an
1:39:29 abandoned
1:39:38 Factory
1:39:41 create 35 MIM
1:39:44 film
1:39:50 [Music]
1:39:51 create now is this thing supposed to
1:39:56 do video as
1:39:59 well once these things are
1:40:04 created yeah this it these these
1:40:08 aren't well I guess they these are like
1:40:12 1974 these are not muscle cars these are
1:40:15 like 1974 AMC
1:40:19 Hornets that's sort of like a Nova but
1:40:22 it certainly isn't like a aresto mod and
1:40:25 these are like all like American
1:40:28 AMC's they're they're
1:40:35 bad let's see can I do anything with
1:40:37 this let's see download tiktock
1:40:40 pin I just started this project from
1:40:43 your idea thank you oh cool which
1:40:45 project did you
1:40:46 start love that
1:40:51 [Music]
1:40:58 I'm going to go down arrow
1:41:02 down that one's not bad I wouldn't call
1:41:05 it a Resto
1:41:07 mod nah we're not gonna we're not going
1:41:10 to give him credit for that that's a
1:41:12 definite
1:41:14 no that's a that one that one just gets
1:41:21 neutral but this is cool actually the
1:41:23 fact that that it generates this
1:41:26 uh this uh the these pull down menus is
1:41:29 pretty slick medium format film we'll do
1:41:34 60s muscle car create
1:41:42 [Music]
1:41:54 that's not bad like
1:41:57 that these are actually Resto mod
1:41:59 looking right
1:42:04 lowered nice
1:42:07 70s five spoke
1:42:20 mags I ain't got no
1:42:32 now flag output download
1:42:41 copy video
1:42:44 fxs yeah I'm on the wait list for
1:42:47 this all right I don't know anyway
1:42:55 I kind of feel like with the Google
1:42:56 stuff there would be really good it it
1:42:59 would be really good for someone
1:43:03 including
1:43:06 Google to put together a single
1:43:08 interface to get to all of their [ __ ]
1:43:10 all of the AI [ __ ] because Google's so
1:43:13 big and has so many different Arenas
1:43:16 like some of the stuff is just designed
1:43:18 for developers some of it are like these
1:43:21 really weird experiments that are like
1:43:23 you can't figure out what they're about
1:43:27 some are things like notebook LM which
1:43:30 is this incredibly powerful tool that
1:43:32 everyone ignored until they made those
1:43:36 audio overviews those podcasts and then
1:43:39 everyone started paying attention to it
1:43:41 I feel like there's a like there it
1:43:44 would be a useful web page to have like
1:43:47 a really pretty front end for all of
1:43:50 Google [ __ ] where you can sort of sort
1:43:52 and filter by you know who you are and
1:43:55 what you're trying to do because there's
1:43:57 a lot of [ __ ] there that it's hard to
1:43:59 find so anyway all right um okay let
1:44:05 me Let Me Entertain You you
1:44:15 make
1:44:17 okay
1:44:19 tomorrow I am speaking at an event
1:44:23 um I think it's a small private event I
1:44:26 don't think I
1:44:29 can I don't think it would work to live
1:44:32 stream the event if I can I will but I
1:44:36 have I have low confidence in
1:44:39 that um but it's I've got a I've got a
1:44:43 jam-packed day tomorrow like from
1:44:44 morning until night and so and I'm not
1:44:47 going to get back here until like 10
1:44:49 o'clock tomorrow night so um so we're
1:44:53 not going live tomorrow night so you
1:44:54 have tomorrow night off so what you
1:44:55 should do is if you've got access to
1:44:58 Sora go make some Sora videos see if you
1:45:00 can figure out how to get it to do good
1:45:02 [ __ ] um start experimenting with
1:45:07 01 we
1:45:09 all have in our grubby little paws a
1:45:13 reasoning engine that is the equivalent
1:45:15 of having like a little buddy who's like
1:45:17 a PhD sure we've got Pate but we can't
1:45:21 ask Pate to solve all of our problems
1:45:23 right so now you've got 01 which is like
1:45:26 your little PhD buddy and start figuring
1:45:30 out problems you can solve with it um
1:45:33 just start playing with it and feel free
1:45:35 to ask it stupid
1:45:37 [ __ ] it might judge you but you know um
1:45:42 okay then here's the other thing I want
1:45:44 you all
1:45:46 to I'm not playing for Chad GPT yeah
1:45:50 exactly um
1:45:56 December 27th and 28th hang on let me
1:46:00 make sure those are the right
1:46:01 dates yes Friday and
1:46:05 Saturday December 27th and
1:46:10 28th um an Murphy and I so last year
1:46:13 between Christmas and New Year's Ann
1:46:15 Murphy and I did this thing called GPT
1:46:17 for good where over a 24-hour period We
1:46:21 Built Well
1:46:24 this community built 150 gpts for
1:46:27 nonprofits just every every couple hours
1:46:31 we would teach people how to make gpts
1:46:33 and we did that for 24 hours
1:46:37 um this year we're going to do a thing
1:46:39 called AI
1:46:41 Festivus and what it's going to be is
1:46:43 it's going to be both Friday and
1:46:44 Saturday we're going to start at 9:00 am
1:46:46 Pacific and go until like I don't know
1:46:49 six or seven at night so we're GNA go
1:46:52 all day Friday all day Saturday we are
1:46:54 gonna take a break Friday night and
1:46:56 actually sleep because we were stupid
1:46:59 last year and just did it for 24 hours
1:47:04 but all day Friday all day Saturday the
1:47:06 27th and 28th every hour is going to be
1:47:09 a different teacher from from the AI
1:47:12 Salon community and from Irregulars and
1:47:15 just from people that we know that are
1:47:16 good with AI teaching different things
1:47:19 so one hour might be mid Journey one
1:47:21 hour might be coding one hour might be I
1:47:24 don't know no code tools I we we're
1:47:27 we're we're we're recruiting speakers as
1:47:30 we speak but put in your calendar 27th
1:47:34 and 28th of December just block off the
1:47:36 day come hang out with us we're gonna
1:47:38 have lounges where we just get to meet
1:47:40 each other we're g to have all these
1:47:42 different sessions so it's going to be
1:47:44 really good I will have graphics and
1:47:46 [ __ ] like that at some point but I just
1:47:47 want you to put that in your uh in your
1:47:49 in your calendar all right all right f
1:47:53 festiv us what is that go watch Seinfeld
1:47:57 festiv us just it's it's festiv us is
1:48:00 the holiday for the rest of us it's not
1:48:02 K Quanza it's not Hanukkah it's not
1:48:04 Christmas it's festiv us it was a uh
1:48:06 George C castanza thing I think from
1:48:10 Seinfeld um all right so it's just
1:48:14 generally celebratory and we're going to
1:48:16 celebrate learning some new [ __ ] and
1:48:18 we're going to do that with really cool
1:48:19 people over two days all right
1:48:28 fantastic all right let's see what did
1:48:31 Josh Groves I can't say all right so
1:48:34 Josh grov is working on a project
1:48:36 inspired by Me by me awesome I just had
1:48:39 an idea and I saw that you made an app
1:48:40 in a couple hours with lovable. deev
1:48:42 hopped on and told it what I wanted and
1:48:44 had a mockup functional prototype
1:48:47 basically I don't know how to fully
1:48:49 develop it but I believe I can now yes
1:48:51 you can the other thing to play with if
1:48:53 you've played with lovable is uh play
1:48:56 with repet repet has an AI agent where
1:48:59 it'll it'll build up the the database
1:49:01 and the web server and the front end for
1:49:03 you as well um if you get into cursor
1:49:07 and bolt those you need to know a bit
1:49:09 more about programming lovable is is the
1:49:13 one that is is
1:49:14 the the most possible to make something
1:49:18 functional with knowing nothing about
1:49:20 coding repetz agent I would say is next
1:49:23 and then once you get into cursor and
1:49:25 bolt those get a bit more uh you'll have
1:49:28 to use chat GPT to teach you how to code
1:49:31 basically but that's the thing like you
1:49:33 can have chat GPT tell you here's the
1:49:36 services you need to set up and here's
1:49:38 how you need to organize your
1:49:39 repositories and what repositories are
1:49:41 and all that [ __ ] you can do it but but
1:49:44 it's just a it's just a pain in the ass
1:49:46 it that that level of work exceeds my
1:49:49 ADD ability to give a [ __ ] so that's why
1:49:53 I like
1:49:54 lovable um all right cool um let's see
1:49:59 if you have not joined the salon go to
1:50:01 the salon. this is a community that's
1:50:04 been around for two
1:50:05 years um I created this with Leah fasten
1:50:08 go to the salon. click on join our
1:50:11 community there's like a seven-step
1:50:13 welcome process it's super simple it's
1:50:17 hopefully inspiring and then there's all
1:50:19 sorts of cool stuff to do in there um go
1:50:21 join the second step in that process is
1:50:24 introduce yourself so go join the AI
1:50:27 salon and introduce yourself okay all
1:50:29 right um I'm going to get out of here my
1:50:32 voice is going crackly um tonight was
1:50:35 one of those weird nights I hope you
1:50:36 found it valuable I was sort of deep in
1:50:38 the tools tonight and I didn't really
1:50:41 come up for breath ever so I know I I
1:50:44 was trying to occasionally uh address
1:50:46 your comments but I didn't see a lot so
1:50:49 it's just the nature of how these things
1:50:51 go sometimes so hopefully that was
1:50:52 useful
1:50:55 um yeah go join the salon and come hang
1:50:57 out with us all right and Josh Groves is
1:50:59 doing all that figuring out all those
1:51:01 all the ways to turn those into real
1:51:02 things here's the thing Josh if you put
1:51:05 in a good solid weekend having chat GPT
1:51:09 teach you how to do it you could even
1:51:10 put on on your phone put chat GPT into
1:51:13 advanced voice mode and just have her
1:51:15 talk you through [ __ ] as you learn it on
1:51:17 the computer over the weekend you could
1:51:20 probably launch an IOS app or or fully
1:51:23 Blown website like so so go do it why
1:51:26 not [ __ ]
1:51:28 it that's why we're
1:51:33 here okay um tomorrow is uh day four of
1:51:38 ship miss my day tomorrow sucks let me
1:51:41 just make sure I I I I don't think I can
1:51:43 live stream tomorrow
1:51:51 11 I might be able to switch an internal
1:51:55 meeting but tomorrow at 10: a.m. Pacific
1:51:59 1 pm Eastern is is the next open AI
1:52:02 announcement my guess is this will be a
1:52:04 minor one because it's a Tuesday and who
1:52:06 gives a [ __ ] um but if I can stream it
1:52:09 live I will on Tik Tok um if I can't
1:52:13 I'll just talk about it uh Wednesday so
1:52:16 Wednesday I will live stream and I'll
1:52:19 talk about it then and then I'll be live
1:52:21 Wednesday night tomorrow night there's
1:52:23 not going to be one of these okay not
1:52:26 going to be here so go play with your
1:52:34 parts I learned more in two hours
1:52:36 talking to it than a four-year degree I
1:52:39 swear Josh that's the that's the thing
1:52:42 that that I don't think people grasp
1:52:44 about
1:52:46 AI
1:52:48 is they think it's like Google they
1:52:51 think it's like a Google search
1:52:54 and the the primary difference about
1:52:57 generative Ai and Google is that Google
1:53:01 was brilliant at what it did for all of
1:53:03 those years but what it did was it
1:53:05 presented you with information that you
1:53:08 had to then go learn what AI does is
1:53:12 it's already learned it and you just ask
1:53:15 it to tell it just to teach it to you
1:53:18 and it's like it's just a fundamentally
1:53:20 different experience and yeah I think
1:53:23 your experience with in two hours you
1:53:25 learned more than you learned in a
1:53:26 four-year degree that's what I think we
1:53:29 all have the opportunity to do every
1:53:32 [ __ ] day if we wanted to it's
1:53:33 exhausting to you know try to take in
1:53:36 that much [ __ ] but it really is the kind
1:53:38 of thing where you're like I could start
1:53:41 on a Saturday morning with a with a
1:53:44 topic that I'm completely clueless in
1:53:46 and by the end of the day on Sunday have
1:53:48 built something with domain expertise
1:53:53 like significantly beyond what should be
1:53:56 possible that's what AI is so it's
1:54:00 crazy uh ripy thanks for the great
1:54:03 community and ideas well thank you part
1:54:05 of the reason this is a great Community
1:54:07 is that you all show up and support one
1:54:10 another and go to the salon and share
1:54:12 your ideas and share what you're
1:54:14 learning um and and kind of take the
1:54:16 spirit of what I'm trying to do here and
1:54:18 you bring it to life so thank you for
1:54:20 that thanks for the acknowledgement but
1:54:21 that's really because of who you are um
1:54:24 ocean thank you um from the heart for
1:54:27 the thought to open this lab thank you
1:54:30 for making a difference thank you for
1:54:31 sharing your knowledge and time it's my
1:54:33 opinion this is better than winning the
1:54:35 lottery wow well [ __ ]
1:54:40 awesome it's like it's like it's like
1:54:44 well ocean you've won the lottery here's
1:54:46 your winner Kyle
1:54:49 hi I'm your prize
1:54:52 [Laughter]
1:54:54 [Music]
1:54:58 oh Salon input has helped me to build
1:55:00 two programs for my business awesome
1:55:02 that's [ __ ] awesome all right
1:55:04 everybody I'm gonna get out um I won't
1:55:07 see you tomorrow um unless I go live for
1:55:09 the for the fourth day of shipus and
1:55:12 then I will see you on Wednesday so
1:55:14 thank you all so so much and we will
1:55:18 catch up with you next time peace
1:55:23 all right