
AI Learning Lab
4:15:24

Video2024-04-301:56:545 views
Description
Chapters
Transcript
0:00 [Music] 0:13 [Music] 0:24 oh 0:27 [Music] 0:59 oh 1:02 [Music] 1:26 here's what Tik Tok tells us more likes 1:29 can get more views 1:32 viewers so like away 1:41 [Music] 2:00 a 2:02 [Music] 2:28 [Applause] 2:29 [Music] 2:56 well I heard there was a secret call 3:01 David played H pleas a 3:05 lord you don't really care for mus 3:09 [Music] 3:16 do major Li ba King 3:21 composing 3:25 hallelujah hallelujah 3:33 hallelujah 3:36 hallelujah 3:40 [Music] 3:49 Hall Silver Fox sharing the live with 3:51 her 3:53 friends be like Silver 3:55 Fox remember the old be like Mike 3:58 commercials be like silver fox 4:00 she's 4:02 [Music] 4:15 sweet I said that Silver 4:18 [Music] 4:22 Fox and I believe that 4:26 [Music] 4:41 s suit and TI does he work at the Dairy 4:44 Queen does it listen to rock and roll 4:48 does it feed the me sing 4:51 Hallelujah 4:53 Hallelujah 4:56 Hallelujah hujjah 4:58 [Music] 4:59 [Laughter] 5:04 I'm telling singing dogs just they just 5:06 bring joy into your 5:13 life who here likes turtles 5:18 [Music] 5:47 you and I here all 5:53 alone Sunday morning here at 5:58 home sky blue is strong strong it's 6:02 [Music] 6:06 true I'm late was watching your reminder 6:08 video I didn't do a reminder video 6:10 tonight oh that reminder video my 6:14 uh yeah I reposted my video from when I 6:18 was doing that work 6:21 about who I'm being in the world cuz 6:25 I've been a little hiding I've been a 6:26 little hiding 6:31 you know how you go through those phases 6:33 where you're like out in the world and 6:35 you're like [ __ ] it let's go and then 6:38 there's other times you're like I think 6:40 I'll just stick around the 6:42 corner I've been going through a phase 6:45 of that and I don't like 6:47 it so I figured I'd put myself back out 6:50 there 6:53 [Music] 7:05 [Applause] 7:09 [Music] 7:16 Silver Fox it 7:18 [Music] 7:21 was it 7:23 [Music] 7:25 was it was cuno that wrote the letter 7:29 [Music] 7:36 and I know I've got it in a Google drive 7:38 somewhere check with Anne Murphy an 7:40 Murphy might have it lunar stick popping 7:43 in here before I inevitably fall asleep 7:45 before anything happens that's that's 7:47 the way we do here this is better than 7:50 better than a shot of uh shot of whiskey 7:54 before bed the AI learning lab we'll 7:56 knock you right 7:58 out the the C in banter about artificial 8:01 intelligence we will put you out like 8:03 that so welcome everybody do me a favor 8:07 tap the double tap the screen let's get 8:09 some likes going get some people in here 8:11 um if you're willing to share the live 8:13 share the live if you're not I totally 8:17 understand but here's a fun game you can 8:19 play is uh God do I need the AI learning 8:23 lab tonight teenagers need need I say 8:25 more no well mine are 23 now so 8:31 so but I get 8:34 [Music] 8:36 it all 8:43 right there's some news tonight I got to 8:46 do something here hold 8:55 please hang on I'll be right back this 8:58 is going to freeze for a moment 9:04 hang on I'll be right back this is going 9:06 to freeze for a 9:07 moment all right we are back people oh 9:10 nice likey likes very good LTX studio is 9:13 having another beta access sometime 9:15 tomorrow okay I will try to get that one 9:17 I will get up at 7 in the morning and 9:19 sit there and just stare at Discord for 9:20 an hour and a half cuz every time I've 9:24 looked away from Discord it's like oh 9:26 form's here and gone so we'll try that 9:29 um some interesting news today um I 9:32 tried out Claude today it's faster still 9:35 challenging it all right cool yeah Claud 9:39 seems to be uh gaining popularity by a 9:42 lot uh I think people are a little 9:44 frustrated with open AI um open AI has 9:49 just been doing weird [ __ ] like today 9:50 they announced um you can batch API 9:54 calls so if you don't need an API call 9:56 done right away you can you can send 9:58 them in batches and they'll process them 10:00 within 24 hours and it's half the 10:03 price um which is an interesting 10:06 interesting 10:08 development but it's like everyone's 10:10 like well where's the [ __ ] when can we 10:12 play with Sora when can we do this when 10:13 can we do that but I don't know if you 10:16 saw today but little old Adobe I think 10:21 probably trying to put a Band-Aid on the 10:22 wound of we trained Firefly on Mid 10:25 Journey images um they announced um 10:31 generative AI for Adobe Premiere 10:34 Pro and it includes the ability to make 10:38 Sora Runway and pika videos inside 10:42 Premiere 10:43 Pro I have a cheat I have a 5-year-old 10:46 granddaughter tapping on the 10:50 screen while I watch the live that's 10:53 hilarious yeah another hundred and I'll 10:55 give you another tury's kiss that's 10:58 awesome uh 11:04 oh man I talked to a Kyle twin today I 11:11 posted it in irregular I think oh did 11:13 you talk to the faker yeah so here let 11:16 me show you that this is uh sorry sorry 11:20 you had to deal with that so let's 11:23 see 11:25 um if you if you follow 11:30 the AI learning lab 11:32 Channel make sure you follow the one 11:35 that knows how to 11:37 spell 11:39 AI 1l learning lab not AI lull learning 11:46 lab that's a scammer and that 11:49 [ __ ] will he's he he tries to 11:51 get everyone who follows this channel to 11:53 follow his channel and then he chats him 11:55 up and says hey you want to buy my 11:58 crypto bullet [ __ ] do not fall for 12:01 that [ __ ] 12:04 [ __ ] and Tik Tok does give not 12:07 to shits about that 12:11 issue which I think is just shitty but 12:14 you know 12:16 whatever Tyrant truth what's happening 12:18 haven't seen you in a while good to see 12:20 you good to see you good to see 12:25 you what would I suggest for creating an 12:27 animated video for children 12:32 well see here's the thing animated 12:35 videoos really 12:38 broad animated video for children is 12:40 really broad 12:46 um I think that the Adobe Express um 12:50 animate with voice if you're just doing 12:52 it for little kids and you just want 12:55 some you want to just knock something 12:56 together quick I think that thing's 12:58 pretty good 13:00 if you want to do something that's a bit 13:01 more sophisticated then I would 13:04 say um generating images in mid Journey 13:10 or di or ideogram whatever your tool of 13:12 choice is and then animating them in 13:15 something like Runway or Pika might be 13:19 good um if you want them to talk you 13:22 could use something like D- ID to 13:24 animate the mouth there's the the 13:26 problem is so okay 13:31 what your oh 13:32 wait oh Jason I'm not talking about the 13:35 fake I'm talking about your virtual twin 13:37 oh that's hilarious how how was my 13:39 virtual twin here's what I know about my 13:41 virtual twin I did not train it on my 13:44 potty mouth that I have in here so the 13:47 my virtual twin doesn't have terms like 13:51 janky and janky pieces of [ __ ] it 13:54 doesn't swear it's not a smartass so my 13:57 virtual twin is very professional so 13:59 it's very much not like 14:02 me that's hilarious um okay but the 14:06 question about the children's book 14:08 here's a new thing that I'm I'm 14:10 realizing and I don't know if this is 14:14 a I don't know if this is a like I'm 14:17 leveling up and this is kind of a 14:19 graduate level getting AI literate I 14:22 don't think it is I am I have grown 14:25 weary of talking about individual tools 14:30 and I am leaning much more toward 14:33 solving problems so in in my I've got a 14:35 three-step process to get AI literate 14:37 it's play solve and share play with 14:41 tools so you can understand what they 14:42 are solve 14:44 problems and then share what you're 14:46 learning right put it out there play 14:48 solve 14:51 share I am increasingly bored with 14:54 talking about the tools themselves 14:56 because here's what's happening I'll 14:58 I'll I'll compare it to 15:02 um to Macintosh 15:05 computers so I've been an apple Fanboy 15:09 my first Mac was a Mac 512 K it was the 15:12 second Macintosh ever produced um second 15:16 model so I've been on Apple Fanboy 15:19 forever right and there was a lot of 15:24 years 15 years where every time Apple 15:29 came out with a new 15:32 model you would upgrade to it period end 15:35 of story because the previous one had 15:38 stopped working you needed the next one 15:41 to be able to do the real [ __ ] you know 15:43 if you wanted to use Quark Express you 15:45 had to get that next model with actually 15:47 8 megabytes of ram in it and a 500 15:50 megabyte hard 15:52 drive 15:55 right and then there was a point at 15:57 which sometime in the the MacBook Pro 16:02 Era it didn't matter it didn't matter if 16:06 they came out with a new Macbook Pro and 16:08 it hasn't mattered for like 10 years it 16:10 just hasn't [ __ ] mattered why cuz 16:13 they're good 16:14 enough they're good 16:17 enough and so I kind of feel like AI 16:19 tools are getting like that right now 16:21 like we now have three or four viable 16:24 gp4 gp4 16:27 Alternatives we have image generation 16:29 tools Dolly mid Journey stable diffusion 16:33 Leonardo um 16:36 ideogram uh Deco here Korea like we've 16:39 got seven of them that are all fine 16:42 they're all some version of fine right 16:46 maybe one's you know mid journe is a 16:47 little sexier and ideograms better at 16:50 text and decco here is faster and like 16:54 but they're all fine and I feel like 16:56 that's happening with all the tools so 16:59 you do need to understand what tools are 17:01 out there you do need to play with them 17:02 a 17:04 bit but I'm I'm increasingly like it 17:06 doesn't really [ __ ] matter like what 17:08 matters is what tool do you 17:11 know and are comfortable with and what 17:14 problem are you trying to solve so back 17:17 to if you want to make an animation for 17:19 a kid you know I would ask things 17:23 like are you going to is it just going 17:25 to be music 17:27 only is it going to be a character a 17:29 consistent character you want to follow 17:31 through the story are there multiple 17:33 characters you're following through the 17:35 story like all of the like I would say 17:37 get really specific about the kind of 17:40 story you want to tell and how you're 17:42 defining 17:43 animation um I'll jump over and I'll 17:45 show you we'll we'll do a we'll do a 17:47 daily Jun with Frank tonight we'll make 17:50 a we'll make a uh we'll make an 17:54 animated adult cartoon tonight and then 17:57 and then you can be inspired by that for 17:59 for your children's 18:01 cartoon 18:03 um but that might not be the right tool 18:05 depending on what you're trying to 18:07 accomplish 18:09 so 18:11 um for 18:15 now if the Adobe Express tool isn't 18:18 enough then you're going to have to use 18:20 multiple tools and it's not a bad thing 18:22 it's just like you know but like so are 18:25 the characters going to talk if they're 18:27 going to talk do you have acting skills 18:28 if you don't have acting skills then we 18:30 got to talk about do you go to 11 labs 18:32 and generate audio right or do you do 18:35 that 18:36 in um udio the music generation 18:40 tool in fact that's actually an 18:43 experiment isn't 18:47 it let's 18:50 go oh Saturday night by the way I mean 18:53 Saturday day this coming 18:57 Saturday I am doing a thing with John 19:01 Leland called the creativity squared 19:04 playshop it's a 4-Hour sort of unleash 19:07 your creative 19:09 potential um with John and I so John and 19:12 I are both EXC creative directors and 19:15 it's a 4-Hour learn about AI bring a 19:18 project and we will start smashing it to 19:21 bits and having a blast with it so video 19:24 Mojo doccle doso and if you're a member 19:27 of the AI Salon there's a coupon code in 19:31 there for it in uh Salon announcements 19:35 uh to save some money off that so anyway 19:36 that's that's going on 19:38 there um where was I going to 19:42 go oh it was going here and I was going 19:45 to yudo 19:49 okay will there be a recording uh there 19:52 will but it it's not going to be 19:55 immediately available for free I don't I 19:59 don't know how that's going to work 20:00 actually but there will be a recording 20:02 if you if you buy the play shop the the 20:04 recording is 20:06 available um okay so I'm going to try 20:09 something here so I'm in right now I'm 20:11 in Udo 20:14 beta and we're going to try to 20:19 make I'm going to try to make some 20:22 background music or some like 20:24 some some like a kid talking let's see 20:30 so I'm going to 20:32 do spoken 20:38 word young 20:42 boy walking through 20:47 woods 20:49 and 20:51 talking to the 20:55 animals and then we're going to do 20:57 custom lyrics and we're going to go 21:00 uh hey 21:05 there 21:11 I'm I'm sure you're very 21:17 nice I'm 21:19 sure you 21:24 don't eat children 21:31 um I am happy and safe 21:37 and 21:41 having a good time all 21:48 right let's see if this can do actual 21:51 like acting like that 22:01 oops and I got to put it in manual 22:06 mode all right so udio this thing udio 22:09 beta is is an absolutely incredible 22:13 music generation 22:15 tool 22:17 that if you put it in manual mode and 22:19 write your own script and put spoken 22:22 word at the beginning it can sometimes 22:26 do talkie talkie stuff and it does it 22:29 pretty well so let's see if it can do 22:35 it ask yudo to do a voice over for a 22:39 social media ad yeah we can do that I 22:41 got to write the 22:43 the thing for the social media ad Oh you 22:45 mean for the for the play shop could try 22:48 doing 22:51 that spelling error youve that's okay it 22:54 doesn't spelling errors don't 22:57 matter these things these things are 22:59 remarkably good 23:01 at spelling 23:09 stuff yeah Vicky the the uh the thing 23:12 that cuno wrote It's on there's a Google 23:15 drive somewhere I've got access to I 23:16 just got to go find it so Silver Fox or 23:19 Vicki just just DM me on the salon and 23:22 we'll find 23:23 it giving an AI in higher ed talk at a 23:27 conference this summer what is a must 23:31 include that's a good one hang on let's 23:34 see if we succeeded 23:36 here by the way I'm about to hit the 23:39 play button one of the things that you 23:41 should always assume when you're using 23:42 AI is when you hit the submit button it 23:45 will suck you should absolutely expect 23:49 everything it produces to be horrible 23:51 and then when it's not you'll be happy 23:54 as opposed to expecting it to be perfect 23:57 and it never is and and you're sad so 24:00 it's all about framing 24:03 people hey there I'm sure you very 24:09 nice oh maybe spoken word isn't quite 24:11 right let's do um let's do 24:18 children's 24:20 um 24:22 cartoon instead of spoken word because 24:24 that definitely sounded like a spoken 24:26 word performance 24:31 it looks like their counter got screwed 24:33 up it reset my my generations to zero 24:36 which is good because I was up to like a 24:48 100 does it change speech patterns if 24:50 you all caps a word I don't know I 24:52 haven't I haven't played with Udo quite 24:54 enough to 24:56 know what works and what doesn't 25:03 that ain't a kid no that was not a kid 25:06 Forest Friends 25:22 parade wait did I is that in manual mode 25:25 shouldn't be playing 25:27 music weird 25:29 [Music] 25:31 could be 25:32 [Music] 25:34 good hey 25:41 there I'm sure you very 25:45 [Music] 25:48 nice I'm sure you don't eat 25:51 [Music] 25:55 children Bob Dylan is happy and sa and 25:58 happy having a good time Bob Dylan is a 26:01 10-year-old walking through the 26:03 [Laughter] 26:09 woods hey 26:11 [Music] 26:13 there hey there it's actually kind of a 26:17 good little 26:20 tune I'm sure you very nice 26:25 [Music] 26:29 I'm sure you don't eat 26:31 [Music] 26:35 children I'm happy and safe and having a 26:39 good 26:41 time okay so yeah this is perfect so if 26:44 you want to make a little kids book you 26:45 go to yudo and you have you you put in 26:49 children's cartoon and it makes a lovely 26:51 little cartoon song with a kid walking 26:54 through the woods all right so you don't 26:55 do that but this is this is sort sort of 26:59 the I this is where what I'm starting to 27:02 get interested in 27:04 is because the tools are so fragmented 27:07 right now if you want to do something 27:09 that's decently sophisticated you 27:12 actually have to kind of play with 27:13 different tools and say you know can I 27:15 get this is why this is why I needed a 27:20 tonight I can count on it to make me 27:22 laugh my ass off that's awesome 27:26 great to hell with those teenagers we 27:29 can do we can do stupid Bob Dylan as a 27:31 10-year-old songs all right let's go 27:34 make a uh let's go make a Frank 27:40 video hey 27:42 there I'm sure you're very nice TR TR TR 27:48 TR TR I'm sure you don't eat children 27:53 I'm sure Tik tok's gonna [ __ ] ban me 27:54 for life for saying that all right um 27:59 [Laughter] 28:01 we're going to 28:08 express guy is singing while playing the 28:10 guitar in one of those old merry-go 28:12 rounds yeah exactly exactly like that 28:15 song could actually be a really good if 28:18 you wanted to make a creepy kids 28:19 animation that would be a great song but 28:22 I assume we don't want a creepy one okay 28:24 so I'm going to go to express. adobe.com 28:29 and I haven't been here in a while so 28:31 chances are decent that they're going to 28:33 have changed the interface to the point 28:34 that I won't know how to find anything 28:36 so just deal with that 28:41 okay so at the very top of adobe Express 28:44 search for 28:47 animate from 28:50 audio and then click that and that'll 28:52 take you to this 28:57 tool all right so there's Frank now oh 29:01 do they have 29:03 new did they put new art in 29:07 here no 29:10 maybe I can't 29:13 tell 29:14 maybe they might have added some new 29:19 art but anyway this is a 29:25 tool that let's see 29:29 1 29:36 19701 why can't you do 16 by9 that's 29:44 bizarre can you turn 29:47 this 29:49 horizontal huh weird all right whatever 29:53 1 197 to1 all right you get a you can go 29:55 super wide or you can do square right or 29:58 portrait 30:02 so here's your little 30:04 character so you add a character you had 30:07 a 30:10 background right so there's a little 30:12 kids sitting in 30:14 the let's see if we can find one 30:16 appropriate 30:18 for for the the creepy song we just 30:22 did um so this is what it is so you pick 30:25 a character so I've got this character 30:27 called Frank 30:30 there's Frank so Frank's a hot dog and 30:34 we're going to make Frank Tick Tock 30:42 [Music] 30:46 sized and then I'm going to add the 30:48 background I can add a 30:50 custom a custom 30:53 image which I've got one called The 30:55 Daily Jank with Frank so we're going to 30:57 search for 30:59 Jank on my very well organized add 31:04 computer anyone else have ADD is your 31:07 desktop like 31:09 a you'd be embarrassed if someone 31:11 organized saw it that's 31:14 mine but I have learned to stop giving a 31:18 [ __ ] about any of 31:21 that it's just like I finally realized I 31:25 think it took me until my 40s to realize 31:29 realiz oh it doesn't matter what 31:32 organization system I use I will just 31:35 never use 31:36 them and and I became much less stressed 31:39 out after I had that moment okay 31:45 so here's here's our here's our cartoon 31:48 all 31:49 right I have multiplying accumulate I 31:53 know I'm a tab 31:56 hoarder uh uh uh I don't see code 31:59 hallucination since the context window 32:01 expansion that's really 32:05 interesting that's very 32:08 interesting change it to YouTube did 32:10 they have a YouTube 32:16 size oh yeah there you go all right 32:21 interesting okay strange all 32:25 right I need a topic for 32:29 all right you know what we're going to 32:30 do we're going to go we're going to go 32:32 on a little field trip before we come 32:33 back to Frank because we got to go 32:34 listen to oh we could do something on 32:37 the the dove I don't know if you've if 32:39 if you've all seen the dove 32:42 um the dove uh what's it called real 32:45 Beauty campaign about generative AI um 32:49 that's a really interesting video to 32:50 watch but we're going to go watch 32:53 um where am I going to find it let's see 33:06 where 33:16 [Music] 33:19 x.com all right let's go see what s had 33:22 to 33:23 say is that it that's it okay let's look 33:28 Listen to Let's listen well in that 33:31 special in that little Mr Sam Alman 33:34 talking about blub and how he's going to 33:36 take over the world and just destroy 33:38 everyone's 33:40 lives oh thank you Silver 33:45 Fox um brother 52 I am not ignoring you 33:50 I just had a chat add 33:52 moment okay here important things to 33:54 talk about at an education conference 33:59 the most important one in my 34:05 opinion is that if your school is 34:09 blocking AI you are on the wrong side of 34:13 history and 34:16 that as a teacher you deal with two 34:18 kinds of large language models in your 34:21 life one of them is chat GPT and large 34:24 language models and the other large 34:26 language model you interact act with are 34:31 students human brains are are 34:34 essentially nothing more than fancy 34:36 large language 34:38 models so here we have a tool that's the 34:40 most powerful knowledge access tool in 34:44 the history of humanity and educational 34:47 organizations are blocking 34:49 it so so Step One is put get the message 34:53 out there that it is their 34:56 responsibility to prepare children for 34:59 the world they're about to enter and the 35:01 world they're about to enter is about to 35:03 [ __ ] radically 35:07 alter and how we used to teach kids and 35:11 what we used to teach 35:13 kids is 35:18 increasingly less 35:21 relevant um and the and the skills in 35:25 the future are going to be things like 35:27 critical things thinking and judgment 35:30 and 35:31 collaboration and so I would say bring 35:35 AI into the classroom and bring it in 35:38 with 35:39 a with a critical thinking mindset 35:42 there's there's a 35:44 u there's a 35:46 teacher um well there's so so one I 35:49 would go into the education Guild in the 35:51 AI 35:52 salon and talk with uh Stacy and and 35:58 talk with um Mr K and other people that 36:02 are in the education 36:04 Guild um I would talk to them about you 36:07 know things that they think are 36:08 important 36:09 topics um there's a there's 36:14 a the hell is her name she's like a 12th 36:18 grade English 36:19 teacher Gib Gibson it's like Gibson's 36:22 the name or something like that should 36:24 be advancing education by Leaps and 36:27 Bounds with an L yeah exactly so what 36:29 Gibson does is is she she sort of 36:32 radically brings AI into the classroom 36:35 so what she did at the beginning of this 36:37 year is she said okay your your semester 36:40 is going to be judged on two 36:43 papers one is a paper that you're going 36:45 to write by hand on paper in 36:49 class the other one is a paper that 36:54 you're going to write with chat GPT 37:00 and you're going to bring it into class 37:02 and we're going to dissect it and talk 37:05 about what it got right and what it got 37:06 wrong and you're going to talk about 37:08 what you learned from it and how you 37:09 would do it different and what you 37:11 learned along the way so that to me is 37:15 absolutely critical and so you know I 37:18 would I mean I don't know what the 37:20 nature of the conference is I don't know 37:21 what the nature of your expertise is um 37:24 but but I would you know to the extent 37:27 that you can talk about getting um 37:31 administrators to get their heads out of 37:33 their asses do everything you can to go 37:36 in that thing I think I think 37:38 educational institutions at this point 37:41 blocking AI is one of the most 37:43 irresponsible things I I can imagine it 37:45 pisses me off to no 37:47 end 37:49 um 10 10,000 likes largely thanks to 37:53 sto's granddaughter uh keep tapping Ava 37:56 you're doing awesome 38:01 oh that's fantastic all right if 38:04 learning is so 38:05 easy now let's see now then everyone I 38:09 couldn't see what that 38:11 said like like like tap tap tap okay so 38:15 let's listen to what Sam has to 38:18 say I wonder if this is I think some of 38:21 the things that I look for more than 38:22 other people are founders that are going 38:25 after something that's no that's not the 38:27 one I'm looking 38:29 for I mean it's the right 38:38 recording here we go this is the 38:42 one model is not going to get better 38:45 fundamentally there are two strategies 38:46 to build on AI right now there's one 38:49 strategy which is assume the model is 38:50 not going to get better and then you 38:52 kind of like build all these little 38:53 things on top of it there's another 38:54 strategy which is build assuming that 38:56 open air is going to stay on at same 38:57 rate of trajectory and the models are 38:59 going to keep getting better at the same 39:00 Pace it would seem to me that 95% of the 39:03 world should be betting on the latter 39:04 category but a lot of the startups have 39:06 been built in the former category and 39:08 then when we just do our fundamental job 39:11 which is make the model and its tooling 39:13 better with every crank then you get the 39:15 opening eye killed my startup meme if 39:17 you're building something on on gbt 4 39:20 that a reasonable Observer would say if 39:22 GPT 5 is as much better as GPT 4 over 39:25 gpt3 was not because we don't like you 39:27 but just because we like have a mission 39:28 we're going to steamroll you but we're 39:31 going to steamroll you right so what 39:34 he's saying is there there are startups 39:36 out there that are sort of assuming like 39:39 like if you build a startup right now 39:41 that's whole mission is to fix the 39:45 hallucinations that gp4 generates and 39:48 and you're building a whole thing around 39:50 assuming that gp4 is going to keep 39:52 sucking or can't do math or things like 39:55 that then when the new thing comes out 39:58 well like what he's saying here is 95% 40:00 of the world should be betting on 40:03 them them uh 40:06 being dramatically better than what they 40:08 are right now and he said it's a big 40:11 mistake not to do that so I just thought 40:12 that was it's just such a it's just such 40:15 a cold he's just got this this this 40:20 little the thing and then then we're 40:23 going to steamroll 40:24 you when we just do our fundamental job 40:28 which is make the model and its tooling 40:30 better with every crank then you get the 40:32 opening I killed my startup meme if 40:33 you're building something on open on gbt 40:36 4 that a reasonable Observer would say 40:39 um let's 40:41 see I don't know if I don't know Frank 40:45 is that frankw worthy I don't know if 40:47 that's frankw 40:50 worthy what's Frank gonna talk about 40:53 tonight any ideas of what Frank should 40:55 talk about he could talk about the 40:57 amount of work needed to make an 40:58 incremental improvements to models will 41:01 only keep 41:05 increasing yeah I I don't disagree with 41:07 that P but gp4 at this Point's almost 41:10 four years old right so they've been 41:12 working on GPT 5 for two years so so uh 41:18 yeah agreed GPT 6 is going to take a 41:20 fuckload more compute than gp5 did but 41:24 but you know um by then the The 41:28 Blackthorn whatever the new Nvidia chips 41:30 are will be out or or the Google TPU 41:35 2.0's will be 41:37 out what whatever they're going to be 41:39 training on on uh gp6 41:43 so yeah I don't disagree with that but 41:46 to think that to think that open AI 41:50 isn't their next model isn't going to be 41:51 dramatically better than their current 41:53 one I think is I think is fo fool's gold 41:55 at this point I mean they've kind of 41:57 proved proven that it's hard to catch up 41:59 to them it's it's been a year and 4 42:02 months well no it's been a year a year 42:04 and a month since GPT 4 came out and and 42:08 people are just catching them they're 42:11 just catching them 42:12 so I I would bet on open AI at this 42:15 point over over the 42:24 [Music] 42:25 doomers I don't understand the contan 42:28 stance why not work toward the same 42:31 future it's not a contrarian stance he's 42:33 not say saying a contrarian stance he's 42:36 he's 42:37 saying what will happen is they're going 42:40 to they're going to release a model and 42:42 then the developers who are betting on 42:45 them not getting any 42:47 better and and building companies around 42:51 that path are going to be the ones that 42:54 [ __ ] about you know open AI killed my 42:58 company and what he's saying is assume 43:00 we're going to get better and if we get 43:03 better does that kill your startup if it 43:05 kills your startup change it 43:08 now 43:09 right like if you're building here 43:13 here's a good example if you're 43:17 building if you're building a a 43:22 multimodal actually the uh LTX studio is 43:26 a good example where LTX studio is 43:29 rolling together gp4 calls with image 43:33 generation calls with um video 43:37 generation calls right to so they're 43:39 gluing together a bunch of different 43:40 models from different 43:43 companies into this single 43:46 interface 43:48 and if they're assuming that the next 43:51 version of chat GPT can't just do all of 43:54 that [ __ ] natively right within chat GPT 43:57 that they're they're probably gonna get 44:00 steamrolled they're probably gonna feel 44:01 like well holy crap you know I built 44:03 this whole company so 44:07 anyway why Bet On It staying J janky why 44:10 not assume it will get better I don't 44:12 you know what's funny Janu I don't know 44:14 I don't know there's there's guys like 44:16 Gary Marcus out there that their entire 44:19 career seems to be staked on these 44:21 things suck and they're never going to 44:22 get better see I told you they sucked 44:25 it's like I I don't know what the deal 44:28 is diminishing turns will begin hitting 44:30 in big ways and they ar ar arguably 44:34 already are in parts of 44:36 llms right and then and then there's 44:39 going to be things Beyond llms I assume 44:42 did you see that hey Pate did you see 44:43 the paper from from your employer about 44:47 the infinite um the infinite context 44:50 window it's called what infinite 44:52 infinite attention or infin attention 44:55 something like that that looks pretty 44:56 interesting 44:59 um that that could be something 45:01 something good I don't know what I want 45:04 Frank to do I don't I'm not inspired to 45:05 do Frank right now I don't have a good 45:09 idea 45:15 although oh I got something I could do 45:17 all right so we're gonna do we're gonna 45:20 do a Frank we're going to do a Frank 45:22 recording so we've got this little 45:25 dude we've got 45:27 the background and now now we just 45:30 record it you can record up to two 45:32 minutes or you can upload a 45:34 file and I turned on enhanced speech 45:37 it's actually a really simple 45:39 tool haven't read the actual paper yet 45:41 no I haven't I haven't either um Pate I 45:44 just saw it today it's one it's it's on 45:46 my list to download and go look at um I 45:50 did write a uh Colorado came out with a 45:54 new um Senate Bill uh that's AI 45:57 legislation for Colorado and it's it's 46:00 pretty sucky it's pretty shitty so so I 46:02 wrote a LinkedIn article about that 46:04 today maybe I'll go show that but let's 46:06 get let's get Frank going here all right 46:09 what's Frank going to talk about let's 46:11 see 46:43 hey it's the daily Jank with Frank here 46:44 we are on April 15th hope you got your 46:47 taxes done you big loser anyway welcome 46:50 to the Daily Jank with Frank I got a 46:52 bone to pick with with the whole uh the 46:54 whole Twitter verse or x.com they call 46:57 it yeah all these people out there 46:59 they're all out there going oh the AI 47:01 sucks it's never going to get better 47:03 it's it's just bad and it's going to get 47:05 worse and the models are collapsing and 47:07 this and that and then you got little 47:09 old Sam Alman there getting on the on 47:10 the Tik Tock or on the uh the uh the 47:13 video what I don't know what it is on 47:14 the x.com the Twitter he gets on there 47:16 and he's like uh yeah any company that 47:19 is betting that we're not going to get 47:21 dramatically better we're going to 47:23 steamroll them we're going to steamroll 47:25 them he actually said that he's like 47:27 like yeah go ahead doubt us watch just 47:30 watch I don't know what it is with these 47:32 people just assuming technology isn't 47:34 going to you know do what [ __ ] 47:36 technology does it gets better you don't 47:39 think this is going to get better look 47:41 at me I'm a talking [ __ ] hot dog this 47:44 has been the daily Jack with 47:45 Frank all right so that so that's how we 47:50 that's how we make oh wait okay good 47:53 daily crunch is writing headlines for 47:55 the Twitter post Frank relis is the 47:56 chance to catch up with us tonight 47:58 that's good okay wait that'll be good 48:01 Frank relishes the chance to catch up on 48:05 um on AI pessimists and then I'm gonna 48:08 tag Gary Marcus and I'm gonna tag Sam 48:10 Alman okay that's good that's good wait 48:13 Frank relishes the chance let me go let 48:15 me go do the let me start the 48:18 [Laughter] 48:22 post all right uh 48:26 Frank 48:28 relishes the 48:34 chance to catch 48:38 up on 48:42 the 48:45 AI 48:50 pessimists all right 48:54 um relishes relishes 48:59 so this 49:01 is 49:03 April 49:05 15 49:07 2024 and then we'll go 49:11 um 49:15 under 49:17 under 49:20 estimate 49:23 Sam at your peril 49:28 quote okay so we got that written good 49:31 fantastic Bob tell him what he's won all 49:34 right we're still waiting on this to 49:35 make the thing Frank should talk about 49:39 the 1.5 hot dog from Costco oh that's 49:43 not a bad idea I don't know what it has 49:44 to do with 49:47 AI but I think I think 49:52 Frank I don't know how Frank would feel 49:54 about the $150 hot dog from Costco I 49:57 think Frank might be a little pissed off 49:59 about it he's a premium 50:01 [Laughter] 50:03 Frank what are they [ __ ] doing losing 50:08 money all right let's see I've always 50:11 got to watch these again because I go 50:12 into some sort of blackout Zone where I 50:14 have no [ __ ] idea what I said so let 50:16 me see if this is even remotely useful 50:20 hey it's the daily Jank with Frank here 50:22 we are on April 15th hope you got your 50:24 taxes done you big loser anyway 50:27 welcome to the Daily Jank with Frank I 50:29 got a bone to pick with with the whole 50:31 uh the whole Twitter verse or x.com they 50:34 call it yeah all these people out there 50:36 they're all out there going oh the AI 50:39 sucks it's never going to get better 50:41 it's it's just bad and it's going to get 50:43 worse and the models are collapsing and 50:44 this and that and then you got little 50:46 old samman there getting on the on the 50:48 Tik Tok on the uh the uh the video what 50:50 I don't know what it is on the x.com the 50:52 Twitter he gets on there and he's like 50:54 uh yeah any company that is betting that 50:57 we're not going to get dramatically 50:59 better we're going to steamroll them 51:01 we're going to steamroll them he 51:03 actually said that he's like yeah go 51:05 ahead doubt us watch just watch I don't 51:08 know what it is with these people just 51:10 assuming technology isn't GNA you know 51:12 do what [ __ ] technology does it gets 51:15 better you don't think this is going to 51:17 get better look at me I'm a talking 51:19 [ __ ] hot dog this has been a daily 51:21 Jag with Frank all right that's fine so 51:25 we're going to open in the editor 51:29 we're going to go add some background 51:30 music so I'm going to go to Media by the 51:33 way if you're wondering what I'm on this 51:34 is on uh Adobe 51:38 Express we're going to go audio healthy 51:41 Business Loop I think is the 51:44 one why is that not 51:52 [Music] 51:54 playing we'll do that 6% 52:00 hey it's the daily Jank with Frank here 52:02 we are on April 15 all right we're going 52:04 to download This Wait download Do I Do 52:07 download now 52:10 yeah 52:17 download you think we could have a Mrs 52:20 Frank that sounds like church church 52:22 lady oh that could be funny well ain't 52:24 that special well Frank Frank listen 52:29 listen listen 52:31 Frank oh ain't he's special and he's 52:34 special well well well well well well 52:38 look at Frank going down to the ballpark 52:41 again yeah we could do a Mrs Frank the 52:44 problem is I don't have a Mrs Frank I 52:46 think I think Mrs Frank would have to be 52:49 a different character I I I think she 52:53 would have to be a different character 52:55 all right let's go to the x 53:01 okay so oh wait I did this in the wrong 53:04 place didn't I yes I did hang on I got 53:07 to flip over 53:09 to the Daily 53:11 Jank so what you're going to do right 53:13 now if you want to support the daily 53:15 Jank with Frank and if you're like if 53:17 it's daily why haven't you done one in 53:19 like three weeks that that's not 53:21 important right now don't don't be a 53:23 Debbie Downer 53:27 all right 53:33 um where's 53:35 my how do I change oh here 53:43 profile oh there it is okay there go to 53:47 daily 53:48 Jank okay so now I'm in the daily 53:51 Jank what's happening is that 53:53 underestimate Sam at your peril 53:59 a corn dog so true be 54:01 present all right we're going to 54:06 add I'm going to rename 54:10 this we'll do 54:16 [Music] 54:18 4-5-4 54:20 Dash 54:23 Frank all 54:24 right and then we're going to open that 54:27 bad 54:28 boy all right so what I want you to go 54:30 go to Twitter and go to um search for 54:35 the account daily Jank daily Jank and 54:38 then you'll find 54:40 Frank and then what I want you to do why 54:44 can I not post this because is it 54:45 uploading oh it's uploading 54:50 [Music] 54:58 [Music] 55:03 is it done it's done let's test it make 55:06 sure it's the right 55:11 one done your big loser anyway welcome 55:15 to the Daily jck with Frank I got a bone 55:17 to pick with with the whole uh the whole 55:19 Twitter verse or x.com they call it yeah 55:23 all these people out there they're all 55:24 out there going oh they all right so 55:28 we're going to post that 55:30 boom post was sent now I'm going to flip 55:33 back over to my other account and we're 55:37 going to go give this some 55:39 love let's go give Frank some love 55:43 where's 55:44 Frank Frank Where Are You Frank 55:53 Frank daily 56:02 Jank there we go so we're going to give 56:05 it a hearty 56:07 heart we're going to 56:11 go oh look I'm gonna go hey Gary 56:15 Marcus 56:17 Gary 56:21 Marcus I 56:23 think Frank 56:26 is talking about 56:32 you you going to put up with 56:40 that Gary Marcus is g to be like who the 56:43 [ __ ] is this Kyle dude and what the [ __ ] 56:44 is with this hot 56:49 dog 56:51 oh all right so we did that little thing 56:54 there and then I'm going to quote post 56:57 this thing and 57:00 say I 57:02 think uh 57:05 Frank likes how 57:10 gangster at 57:14 samama is getting 57:17 with people under 57:23 estimating what they are about to 57:36 drop underestimating what they're about 57:39 to drop boom all right have about it 57:42 people Kyle's sitting there cracking 57:44 himself up I'm here for it that's this 57:46 that's what this we I think we have come 57:48 I have come to accept that this channel 57:50 is really just about me entertaining 57:52 myself and if you guys want to hang out 57:54 with me that's okay too 57:58 it's it's pretty bad it's pretty 58:03 bad oh 58:05 man all right daily Jank with Frank is 58:09 live 58:10 boom so that technology the daily Jank 58:14 technology on Adobe Express you could do 58:16 a whole children's video with just that 58:18 technology 58:20 right you could also try to do more 58:23 sophisticated stuff oh trust uses we 58:25 quite entertaining 58:33 um 58:45 oh 58:47 Adobe oops 58:52 [Music] 58:58 where am I going 59:06 Adobe there we go there we go people is 59:10 this it 59:15 yeah yeah that's the one from 59:24 Adobe make a selection write a prompt 59:27 and here are some features we're 59:29 developing powered by our new Adobe 59:31 Firefly video model that will transform 59:33 how editors work object addition lets 59:36 you add or change objects and footage 59:38 with a text prompt make a selection 59:40 write a prompt and add anything you can 59:42 imagine object removal makes selecting 59:44 and removing objects across frames quick 59:47 and precise when your clip is too short 59:49 and you want to hold on a shot or a 59:51 character for an extra beat generative 59:53 extend adds frames with firefly so you 59:56 have the exact exact media you need and 59:58 there's more we're excited about early 1:00:00 research Explorations with our friends 1:00:01 at early research 1:00:04 Explorations ah open AI Runway and Pika 1:00:09 so editors can use 1:00:11 the Adobe Firefly private beta personal 1:00:14 use open AI open AI Sora built right 1:00:18 into Adobe Premiere that's model for 1:00:20 their project from open AI here's an 1:00:23 example of their Sora model generating 1:00:25 the perfect b-roll for any SC and we're 1:00:27 committed to transparency through 1:00:29 content credentials so you always know 1:00:31 whether AI was used and what model was 1:00:33 used in the creation of the media you're 1:00:35 viewing it's Adobe Premiere Pro 1:00:38 supercharged by 1:00:41 AI pretty slick and then they've got 1:00:43 Runway and Pika in there as well I don't 1:00:45 think it's coming out to the fall so no 1:00:47 need to get super hot and bothered but 1:00:50 you know again assume that these tools 1:00:53 are going to get better and better and 1:00:54 more and more integrated and and you 1:00:56 know the whole thing about Hollywood not 1:00:59 using these tools to make movies if they 1:01:01 roll that right into Premiere Pro of 1:01:03 course of course these are going to be 1:01:05 used this is the tools are going to be 1:01:08 just like how we think about spell check 1:01:10 and sentence complete currently right 1:01:13 it's it's going to be everywhere it's 1:01:15 going to be in in all of the software 1:01:17 that we use it's going to be it's just 1:01:19 going to be deep deep deep inside the 1:01:21 stuff we 1:01:23 use that's a an Amelio's wife lovely 1:01:26 little 1:01:28 uh lovely little hard cased uh travel 1:01:32 bag there 1:01:38 [Music] 1:01:49 beautiful we need a Kyle bot that's 1:01:51 training to throw in a [ __ ] every few 1:01:53 sentences 1:01:56 is there a better slide deck AI besides 1:01:59 Gemma um I used another one for a while 1:02:01 but I can't remember the name of it 1:02:05 now 1:02:09 I I haven't found a slide generating AI 1:02:13 that doesn't take you longer to unfuck 1:02:16 it up than just to write it 1:02:20 yourself cuz generally it's writing 1:02:23 really all all you really need for for a 1:02:27 slide presentation is an outline like 1:02:29 what I do is write an outline and then I 1:02:32 import that convert that 1:02:35 over there might be brother 52 I would 1:02:38 go to um I use I use keynote but I would 1:02:41 go to um uh futur pedia.com 1:02:59 do you have a YouTube channel I do so um 1:03:03 so at some point here in the next month 1:03:05 or two I'm probably going to expand this 1:03:09 Channel or do something different where 1:03:12 I think I'll do Tik Tok a little bit and 1:03:14 then I'll probably multistream in in 1:03:17 more like desktop format and I'm going 1:03:20 to do it on YouTube live and probably 1:03:23 LinkedIn 1:03:26 and I don't I probably those two to 1:03:29 start um but if you go to um AI 1:03:32 learninglab 1:03:34 dtt as in Tik tock-tick tock on YouTube 1:03:38 AI learning laab 1:03:41 dtt um that's the AI learning lab Tik 1:03:45 Tok Channel where the archives of these 1:03:48 lives live yeah silverfox just pinned it 1:03:52 there YouTube is AI learninglab 1:03:59 dhtt and there is I've been doing this 1:04:01 [ __ ] for a year come April 30th I think 1:04:05 April 30th was my first one or somewhere 1:04:07 around that so there is a Year's worth 1:04:10 of content with the exception of a 1:04:12 couple of weeks I got kicked off for bad 1:04:17 behavior I made the M what did I make 1:04:19 the mistake of talking about oh bias 1:04:22 someone was asking about racial bias and 1:04:24 so I was trying to answer how these 1:04:26 models are biased or not and 7day boot 1:04:31 happened three times before I figured 1:04:33 out what I could talk about and then I 1:04:35 almost got Perma band because Kaa The 1:04:38 Real Time image generation thing made 1:04:40 some 1:04:41 boobies and it made boobies in a like an 1:04:44 oil painting that were on screen for 1:04:46 like two seconds and I got my my screen 1:04:50 went black and then it just went you are 1:04:52 permanently banned from Tik Tok I was 1:04:54 like 1:04:56 I appealed it and they gave it right 1:04:58 back to me but I was so busy with 1:05:00 classes I'm back with all of you welcome 1:05:02 back John good to see 1:05:04 you good to see 1:05:06 you 1:05:12 uh is this going to be one of those and 1:05:14 if you want to see the good stuff switch 1:05:16 to camera 2 kinds of things 1:05:18 possibly possibly I'm just it's it's 1:05:22 really 1:05:23 tough here's the deal 1:05:26 being able to demo software right now I 1:05:29 feel like there's kind of there's 1:05:32 there's sort of two things happening one 1:05:34 is there's a lot of people just trying 1:05:35 to figure out the basics what's what's 1:05:38 AI how are we getting there things like 1:05:41 that there's been people that have been 1:05:43 coming to this channel for a year that 1:05:46 are building stuff with this they're 1:05:48 actively building stuff with it they're 1:05:49 still learn we're all still learning 1:05:53 right it's really frustrating trying to 1:05:56 be able to demo software when I've got a 1:05:59 screen that's this big and and I'm like 1:06:03 looking around the phone that's filming 1:06:06 me trying to see a screen that's too 1:06:08 dark I need to keep it dark so it 1:06:09 doesn't get blown out when I share it 1:06:12 like it's 1:06:13 just so I I'm just feeling like there's 1:06:15 a there's a need to take it up to take 1:06:17 it up a level so I don't quite know what 1:06:19 it looks like but we'll see we will see 1:06:23 awesome I F I fall asleep during your 1:06:25 lives I get up so early be for the 1:06:28 people that's good I listen I I if if I 1:06:31 can help people with their insomnia 1:06:33 that's great you were just a teenager 1:06:35 when you started these exactly yeah 1:06:37 exactly I was 17 when I started these a 1:06:40 year ago now look at me it's this is a 1:06:44 rough business it's a rough 1:06:48 business Adult Swim yep exactly 1:06:55 [Music] 1:06:59 tomorrow Frank the E evils of webp oh 1:07:03 laughing Thunder Frank can absolutely 1:07:06 Frank is absolutely going to do a web PE 1:07:09 can you remind me laughing Thunder 1:07:11 remind me to do that because I am so 1:07:14 pissed off about webp I don't know who 1:07:16 the [ __ ] invented it I don't know who 1:07:18 they paid off to build it into all this 1:07:21 software and then what the [ __ ] is it 1:07:23 with software that produces webp files 1:07:27 not being able to read webp files 1:07:31 what is 1:07:38 that 1:07:40 was were people complaining about JPEG 1:07:43 and 1:07:45 PNG I didn't I didn't see the 1:07:50 protests why are we doing this why does 1:07:53 anybody know P do you know does anybody 1:07:55 know what p is and 1:07:57 why the world should give a 1:08:01 [ __ ] pngs you can have transparency 1:08:05 they've got color information in them do 1:08:07 we need more than that anyone 1:08:13 buer sunno is upping their game with ton 1:08:16 of free credits 1:08:18 well how about they up their game with 1:08:21 increased quality of 1:08:24 song like I do appreciate the 500 free 1:08:27 credits that sunno gave me but I'm also 1:08:31 like yeah this other things 1:08:35 better Vicky I love webp Vicky tell me 1:08:38 well what is it why like what what is it 1:08:42 do and it does it not make you [ __ ] 1:08:45 insane when you can't open AI produces 1:08:50 webp images and if you upload a webp 1:08:52 image to open AI to chat GPT it goes I 1:08:55 do not recognize this file format what 1:08:57 have you done why would I why would I do 1:09:01 anything resembling looking at this 1:09:04 image 1:09:11 are like what does it do is it faster is 1:09:14 it smaller is it 1:09:16 [Laughter] 1:09:19 what minimizes the size it's epic it 1:09:23 would be epic if it worked PNG is 1:09:25 lossless jpeg is 1:09:28 lossless smaller webp is 1:09:32 brainless oh that's good PNG is lossless 1:09:35 jpeg is lossy webp is brainless that's 1:09:38 great frustrating it's so frustrating 1:09:41 although Vicki's a fan if if Vicki's a 1:09:44 fan I I you know I'll at least give it a 1:09:46 second shot but like why is it rolled 1:09:49 out so far and and compatible on one 1:09:54 side of a piece of software and not on 1:09:56 the other side it doesn't make any sense 1:09:58 to me at all have you bought an apple 1:10:00 silicon Mac yet 1:10:02 um I know you aren't going to open 1:10:05 source but they're great for it um I 1:10:08 have an M1 this is a the thing you're 1:10:10 looking at there's an 1:10:12 M1 and apple just announced that their 1:10:15 m4s in the fall are going to be um AI 1:10:20 beasts so I'm going to I'm going to wait 1:10:22 for the M4 so I've run i' I've done 1:10:28 the the the I've 1:10:31 installed two two or three different 1:10:33 large language models on my 1:10:38 M1 and they were 1:10:42 okay 1:10:44 [Laughter] 1:10:46 but they were like almost as good as GPT 1:10:49 3.5 and I'm like I could just use 1:10:53 that now sure I can't install a you know 1:10:58 an unrestricted llama model and have 1:11:02 it swear at me um but you 1:11:06 know but I don't like I don't I just 1:11:09 don't have a need for it right now so 1:11:10 but I will absolutely be getting an M4 1:11:13 an M4 Mac in the 1:11:17 fall because I think what Apple's going 1:11:19 to do is you're going to be able to run 1:11:21 stuff locally that's actually really 1:11:22 good 1:11:26 I assume smaller due to the web yeah I I 1:11:29 would assume 1:11:31 so uh it was I the webp 1:11:36 wizard I I'm telling you man somebody 1:11:39 did some [ __ ] magic to get so many 1:11:40 people to adopt it but like it like what 1:11:43 it feels like is a is a protocol that 1:11:46 was adopted quickly without 1:11:48 thought and that's generally not the 1:11:51 case like jpeg it took them like a 1:11:53 decade to come up with it before people 1:11:55 actually used it and adopted it and it 1:11:57 was in all these test things and webp 1:12:00 just [ __ ] shows up out of 1:12:02 nowhere it's called webp but it's its 1:12:05 file extension 1:12:08 isavi V or avix or some 1:12:11 [ __ ] and then it doesn't work 1:12:15 in most 1:12:18 software who who did 1:12:23 this What wizard seriously was a wizard 1:12:26 some webp wizard said here's a new 1:12:29 protocol go install it on all your [ __ ] 1:12:32 you developers you'll love 1:12:39 it oh that's a great idea Papa JJ why 1:12:43 didn't I think of that 1:12:57 hey chat GPT I have a burning question 1:13:00 period when you generate an image you 1:13:03 generate it in webp format 1:13:13 period And 1:13:17 yet and 1:13:19 yet if I download that 1:13:22 file and re-upload it to you you don't 1:13:26 recognize it period two questions period 1:13:32 one what is web P format and why should 1:13:36 I care question mark and 1:13:43 two how can you produce a file format 1:13:46 that you can't recognize question 1:13:49 mark doesn't that seem ridiculous to you 1:13:53 question mark 1:13:56 do you think you're 1:13:57 [Laughter] 1:14:09 dumb oh it's an oh it's an image format 1:14:13 developed by Google that's why it's in 1:14:15 everything all right Pate this is your 1:14:17 fault did you [ __ ] Pate Pate created 1:14:21 this this is pate's 1:14:24 Fault web piece an image format 1:14:26 developed by 1:14:28 Google aimed at providing Superior 1:14:30 compression for images on the 1:14:35 web this means images And webp format 1:14:38 are usually smaller and size while 1:14:39 maintaining similar or better quality 1:14:42 compared to other 1:14:43 formats particularly valuable for 1:14:45 websites and applications because it 1:14:46 helps them load load faster okay I 1:14:48 understand its purpose why generate 1:14:51 images in a format you can't recognize 1:14:53 on re-upload 1:14:56 when an image is generated here it's in 1:14:58 webp format due to the efficiency and 1:15:00 size quality which is great for online 1:15:02 viewing and sharing however when it 1:15:03 comes to re-uploading images for 1:15:05 recognition or processing there may be 1:15:08 limitations based on specific 1:15:10 functionalities programmed into the 1:15:12 system 1:15:14 exactly that's just dum 1:15:17 dum it's not the format itself that's 1:15:19 unrecognized but rather the current 1:15:21 setup that might not handle re-uploaded 1:15:24 images for further oh you're and no I 1:15:28 don't think I'm 1:15:29 [Music] 1:15:30 [Laughter] 1:15:35 dumb systems like me have specific rules 1:15:38 and 1:15:38 capabilities H chat GPT is getting 1:15:42 [Laughter] 1:15:47 defensive all right let's 1:15:51 see so Pate it's pate's fault all right 1:15:55 ug 1:15:56 avif has better compression than webp 1:16:00 wait so wait is so avif is not webp 1:16:04 that's a different 1:16:09 thing I I'm I'm so 1:16:20 confused no avif is not webp huh 1:16:29 but when I 1:16:32 download when I download an image 1:16:37 from or when I rightclick save an image 1:16:40 from chat GPT it's avif but then when I 1:16:43 try to put it other places it says it's 1:16:46 a webp 1:16:48 format that's what's missing artificial 1:16:51 stupidity um Arnold Schwarzenegger had a 1:16:54 really funny he was on Travis Kel's 1:16:58 podcast and they said something about 1:17:01 are you afraid of artificial 1:17:02 intelligence and he and he said no I'm 1:17:05 afraid 1:17:09 of it wasn't real something stupidity 1:17:13 but but he basically just said people 1:17:15 are a lot scarier than than the machines 1:17:18 are he goes no I don't think the robots 1:17:20 are going to kill us I think it's the 1:17:21 stupid people I thought that was pretty 1:17:24 good oh correct they're different that's 1:17:27 bizarre all right let's see my GPT sorry 1:17:30 I can't do a link on my phone what's it 1:17:34 say come 1:17:37 on why can I not load that 1:17:47 image I can't see that whole image 1:17:50 silver fox 1:17:56 they seriously asked the Terminator if 1:17:57 he's afraid of AI yeah exactly it was a 1:18:00 great one here let's go we can go find 1:18:04 it um let's see 1:18:08 Kelsey um Arnold 1:18:53 schwart reboot one of your movies and 1:18:56 cast each of us in a role what movie and 1:18:58 uh who do you think we'd play I would 1:19:00 definitely do Conan the Barbarian I mean 1:19:02 you don't have to change 1:19:04 anything all you have to do is go on no 1:19:07 you can't put a wig on it wait that was 1:19:10 a 1:19:11 wig there was a partial week Hollywood 1:19:15 baby but I mean that would be fantastic 1:19:16 all you have to do is just learn how to 1:19:18 horse pick right and maybe you guys know 1:19:20 how the hor I've been bucked off every 1:19:22 horse I've ever been 1:19:23 on he not lying the key thing is to 1:19:26 really get good in horse piig riding 1:19:28 because you know when you do the sword 1:19:29 fighting on a horse you really you 1:19:31 cannot pay much attention to the horse 1:19:33 the horse goes around the circle and 1:19:35 does it thing but you have to pay 1:19:36 attention to the sword not to get whack 1:19:38 I think if you take sword fighting 1:19:40 lessons I took three years sword 1:19:43 fighting lessons and three years 1:19:45 horseback riding lessons and with all 1:19:47 kinds of weaponry and the martial arts 1:19:49 and all that stuff and it really was 1:19:51 helpful to be that prepared when they 1:19:53 did the movie I mean imagine you guys 1:19:56 not years the big hammer and the big 1:20:00 battles and all that stuff riding there 1:20:02 into the village just wiping out he's so 1:20:05 hard to listen 1:20:10 to H where's the let's see let's see 1:20:13 Travis 1:20:18 AI I guess it's on the Tik Tock hang on 1:20:21 let me go to the Tik Tock hang on people 1:20:30 um well we're here with the fake news 1:20:32 the 1:20:33 dishonest well we're here 1:20:36 with the fake news the dishonest Med 1:20:40 okay um 1:20:43 Travis 1:20:45 Jason 1:20:48 Arnold AI 1:20:53 [Music] 1:20:56 wow look at that cheating it 1:21:02 works um how close are we to the 1:21:06 Terminator becoming a real thing with AI 1:21:09 and I think be very close right you're 1:21:12 very close uh people ask me all of them 1:21:14 how do you feel about AI wow look at and 1:21:18 um we're going to have some very special 1:21:20 uh surprise guests at the uh 1:21:28 [Music] 1:21:36 I learned the rules and then later on 1:21:39 around 1:21:39 10 where 1:21:53 why let's go Terminator 1:22:02 [Music] 1:22:12 Terminator how do you pick the movie 1:22:14 Parts because the breath of action and 1:22:17 and and terminate films but there were 1:22:20 other movies and wow look at that 1:22:25 cheating it 1:22:27 works um how close are we to the 1:22:30 Terminator becoming a real thing with AI 1:22:33 and I think be very close right be very 1:22:37 close uh the funny parts after this why 1:22:40 did 1:22:41 they why are people being 1:22:48 stupid all right well whatever I can't I 1:22:51 don't I don't remember what he says he 1:22:53 says something stupidity 1:22:55 that he's not afraid of artificial 1:23:00 intelligence uh let's see can we do stup 1:23:02 let's see AI 1:23:11 [Music] 1:23:19 stupidity yeah I can't find it oh well 1:23:23 whatever creep 1:23:27 sorry can't find it old guy trying to 1:23:28 find 1:23:29 something or you know better yet stupid 1:23:32 people putting up incomplete Clips you 1:23:35 don't cut off the clip right before the 1:23:45 [Music] 1:23:53 punchline oh watched the profile in 1:23:55 Silver Twilight Zone episode today so 1:23:57 good oh that's 1:23:59 cool Kyle would totally say that I don't 1:24:01 know what I would totally say but 1:24:05 cool oh I watch some gen xer try to 1:24:08 search for a clip for 30 1:24:12 minutes exactly well what' you do last 1:24:15 night I don't want to talk about it 1:24:17 everything on Tik Tok is so 1:24:19 weird tick tock setion 1:24:25 uh Master Alexander what's happening how 1:24:28 you doing avif was developed as open 1:24:32 source all right well there you go avif 1:24:35 developed as open source incorporated 1:24:38 into chat GPT when you download it it 1:24:42 turns into webp maybe not but maybe but 1:24:45 who knows and then when you try to 1:24:47 reload it it doesn't work because it's 1:24:49 open source why would it 1:24:53 work I don't I you 1:25:00 just life is actually 1:25:03 simple don't be a dumb 1:25:07 Dum if okay if you're a 1:25:11 developer and you have a product person 1:25:13 even if it's a crazy product person like 1:25:15 me that doesn't give you clear 1:25:19 instructions this was a real thing that 1:25:22 happened to me with a developer 1:25:25 I said 1:25:27 hey we have this feature that's always 1:25:32 on I want to be able to turn it off so 1:25:34 can I have a switch to be able to turn 1:25:36 it 1:25:37 off and so they went away for two weeks 1:25:40 and they did their magic wizard 1:25:43 hat 1:25:46 Coating in their dark cave with their 1:25:49 Kool-Aid their caffeinated Kool-Aid 1:25:54 two weeks later they came back here you 1:25:57 go boss here's your thing you wanted to 1:26:01 be able to turn off you can now turn it 1:26:03 off I was like awesome and so I went and 1:26:07 I logged into the system and I saw the 1:26:09 button and I'm like far out and I turned 1:26:11 it off and then I went to turn it back 1:26:17 on and there wasn't an on 1:26:21 button and their response to me was well 1:26:24 you didn't specify that you wanted to 1:26:27 turn it back 1:26:32 on we have electricity in a house and we 1:26:35 have a 1:26:36 light and it's always 1:26:39 on and I want to be able to install a 1:26:41 thing where I can turn the light 1:26:44 off is it not 1:26:46 self-evident you should be able to turn 1:26:48 it back 1:26:50 on in the world of software development 1:26:53 the answer to that question is no if you 1:26:56 didn't specify that you also wanted to 1:26:59 have the on state as well as the off 1:27:02 state that would not be 1:27:08 developed do you see why I'm 1:27:12 old do you 1:27:15 see cuz I can't see cuz I'm 1:27:20 old so this is like this is the webp 1:27:23 problem 1:27:25 someone said we need to be able to 1:27:27 generate webp so it displays more 1:27:30 efficiently 1:27:32 great the other team working on the 1:27:35 image import tool GPT 1:27:39 Vision apparently they didn't get the 1:27:42 memo we might want to go on 1:27:47 ahead and import the image that we 1:27:53 export so that's a good developer they 1:27:55 listen to your 1:27:57 instructions well exactly yeah at least 1:27:59 I got the off switch right two weeks 1:28:03 later that was I honest to God I sat I 1:28:07 must have sat there for five minutes 1:28:08 dumbfounded like I I thought they I 1:28:11 thought I was like I thought Ashton 1:28:13 Kutcher was in the room I thought I was 1:28:14 being punked I'm like you got to be 1:28:17 [ __ ] kidding 1:28:20 me I don't it's relatively logical if if 1:28:23 you uh if you specify that uh we we 1:28:26 created a switch to a to to to disable 1:28:30 the thing that we have successfully uh 1:28:33 accomplished uh the uh the the the 1:28:36 requirements uh put forth in the uh in 1:28:39 the 1:28:43 uh 1:28:45 what it's called an onoff discriminator 1:28:49 exactly they gave me the off part of the 1:28:52 onoff discriminator and they were 1:28:55 they were honest to God 1:28:57 baffled that I asked for the on part of 1:29:01 the onoff 1:29:07 discriminator and I'm the 1:29:11 [Laughter] 1:29:16 [ __ ] seriously people seriously 1:29:20 seriously is this how we're doing it now 1:29:22 we're getting passive aggressive and 1:29:24 we're talking about developers like this 1:29:27 seriously have you all heard that gen Z 1:29:32 thinks that the thumbs up emoji is 1:29:34 passive 1:29:38 aggressive like someone says hey can you 1:29:41 do that for me and you send them a 1:29:43 thumbs up apparently that's pissing 1:29:45 people 1:29:47 off oh my God my stomach hurts from 1:29:50 laughing 1:29:56 can I just if there are any gen zers on 1:29:59 the 1:30:00 call 1:30:02 um really sorry for your loss about the 1:30:05 thumbs up as a gen 1:30:08 xer knowing that the thumbs up is 1:30:11 upsetting people oh my God does it make 1:30:14 me I I am using the thumbs up whenever I 1:30:16 get a chance now I'm using it in 1:30:18 appropriately 1:30:23 now exactly 1:30:26 I can see that exactly point to the 1:30:29 thumbs 1:30:32 [Laughter] 1:30:38 up oh my 1:30:40 god oh are your lives really that good 1:30:44 that the only thing left to be upset at 1:30:46 is a thumbs up 1:30:49 emoji I mean if we were all sending 1:30:51 eggplant emojis like hey could you do 1:30:54 that for me and then I'm like here's a 1:30:56 big eggplant I could see where that 1:30:58 would be taken like it might be a little 1:31:00 appropriate if you know you take the 1:31:02 non-vegetable meaning of the 1:31:04 word or of the picture but a thumbs up 1:31:11 really I think I've created a 1:31:22 monster I I I work in a hostile work 1:31:26 environment I when I so can I just say 1:31:30 so I'm I'm a little upset um but thank 1:31:33 thank you for taking the time to see me 1:31:34 well that's what HR is for that's why 1:31:36 we're here so um so um this is really 1:31:42 hard I the other day 1:31:47 um I sent a slack message 1:31:52 um to someone at the company who I we 1:31:55 can refer to as CEO 1:31:59 um and and I said um would it be okay if 1:32:03 I if I left early to go my feret needed 1:32:08 um needed antibiotics and I could only 1:32:10 get an appointment at two and I but so I 1:32:12 asked permission because you know I I 1:32:15 I'm very respectful in in the workplace 1:32:18 and so I asked is it is it okay if I 1:32:20 leave early to take care of Jim Jim uh 1:32:23 my favet and 1:32:24 and so and his 1:32:27 response there there was no words and 1:32:30 and it was just 1:32:32 this I felt so attacked I I'm 1:32:37 like does that mean it's okay does it me 1:32:40 I it was so unclear to me and I just I I 1:32:43 feel it it just feels it's a little 1:32:45 hostile it's a it's a bit of a hostile 1:32:50 environment so I so I think I I'm I'm 1:32:53 going to tender my 1:32:55 resignation um I I really like it here 1:32:58 and and you know the the person in 1:33:01 question 1:33:02 seems sometimes reasonable uh but I I 1:33:06 thought this was too much I I just you 1:33:08 know so you should know um this is why 1:33:11 I'll be I'll be leaving at the well I I 1:33:14 don't really have a job yet so if I 1:33:16 could just keep my job until you know 1:33:18 something works out that that would be 1:33:20 great 1:33:26 is this the dolly thumbs up 1:33:32 moment 1:33:34 oh true thing oh true thing okay that 1:33:38 makes sense in in ancient Rome this 1:33:41 meant kill him no I thought it was this 1:33:44 wasn't this kill him cuz I thought it 1:33:49 was meant kill him doesn't that mean 1:33:52 kill him 1:33:55 you didn't watch Gladiator is it up I 1:33:58 thought it was 1:33:59 here this seems more aggressive like 1:34:02 you're 1:34:03 like it's 1:34:08 okay I got to ask chaty PT now you got 1:34:11 okay now see now now see now see now I'm 1:34:15 feeling 1:34:16 attacked okay 1:34:22 um in old golden timey Roman gladiator 1:34:27 days was a thumbs up The Kill signal or 1:34:30 a thumbs down the kill 1:34:33 signal and I probably can't say that 1:34:35 word k i l l can I the famous thumbs up 1:34:38 thumbs down gestures commonly associated 1:34:40 with gladiator contest are often 1:34:42 believed to have signaled life or death 1:34:45 however the historical accuracy of this 1:34:48 interpretation is a bit more complicated 1:34:51 and less clear cut than popular culture 1:34:55 suggest the traditional belief is that 1:34:58 thumbs down was the signal of 1:35:03 death meaning the crowd would sponsor 1:35:06 the games wanted the Victor to finish 1:35:08 off the defeated opponent conversely 1:35:10 thumbs up was thought to have 1:35:14 mercy interestingly historical evidence 1:35:18 does not clearly support this 1:35:19 interpretation 1:35:21 uhoh this is why jenz is upset 1:35:24 some sources suggest that different 1:35:26 gestures entirely might have been used 1:35:32 like and we all know movies are 100% 1:35:36 accurate now ask for an example of 1:35:39 thumbs from a 1:35:41 book in the form of an 1:35:45 excerpt the actual direction of 1:35:48 meaning could be various motions of the 1:35:50 thumb not necessarily up or down the 1:35:52 actual Direction in modern depiction 1:35:55 thumbs down is typically the death 1:35:58 sentence thumbs up signals 1:36:01 Mercy all 1:36:03 right I I just I 1:36:07 remember 1:36:08 like the the the king or whoever doing 1:36:12 this and then 1:36:13 going and 1:36:16 then I'm sure Jen Z I'm so sure Jen Z 1:36:20 knows this 1:36:24 oh my 1:36:26 God I I feel little tacked I thought I 1:36:29 got the I and then texted him and then I 1:36:31 got texted back no word no respon no 1:36:34 explanation whatsoever 1:36:37 just you know and it just I it just it 1:36:41 it I it I had to take a breath I had to 1:36:44 take a breath I had to take a breath I'm 1:36:45 so upset I'm going to I I need I'm going 1:36:48 to go make a tofu hot dog 1:36:54 [Laughter] 1:36:59 cisc and Ebert oh yeah cisc and Ebert 1:37:02 yeah 1:37:04 exactly how often do you think about the 1:37:07 Roman Empire you know what's so funny is 1:37:10 is when that Meme came 1:37:13 out I don't think I ever really thought 1:37:16 about it but since that Meme has come 1:37:17 out like it's all I think about like I 1:37:20 like that Meme has made me think about 1:37:21 the Roman Empire more and more like I 1:37:24 just feel like I I must not be a man 1:37:28 enough if I'm not thinking about him so 1:37:30 I've been thinking about it more but I 1:37:32 don't I never really thought about it I 1:37:34 don't really give a [ __ ] about it 1:37:36 although you know being in the theater 1:37:39 in the theater and all you do like 1:37:41 Julius Caesar and you do things like 1:37:43 that so I don't 1:37:44 know I I feel like I got it out of my 1:37:47 system see my post in 1:37:50 Irregulars okay 1:37:52 [Laughter] 1:38:08 Master 1:38:13 Alex Master Alex and like hon honest to 1:38:17 God seriously 1:38:19 like being a responsible employer I 1:38:23 actually do want like does anyone 1:38:25 actually know what is passive aggressive 1:38:27 about 1:38:28 this cuz I can't for the life like like 1:38:32 it it's just I'm like yeah good job like 1:38:35 that's like I don't I've spent my whole 1:38:37 life going all right awesome good job 1:38:40 and I like the Emojis like 1:38:42 yep anyone does anyone know what the 1:38:44 actual 1:38:46 um what the actual issue 1:38:50 is uh uh I'm not sure that's passive 1:38:54 aggression they say it's low effort oh 1:38:57 Jesus really yeah let's let's have that 1:39:02 conversation it lacks confirming context 1:39:05 they don't know if you're being sincere 1:39:07 okay 1:39:10 well careful Kyle Tik tok's gonna ban 1:39:13 you for so many thumbs 1:39:16 up they don't know if you're being 1:39:19 sincere oh good God I just can't you're 1:39:23 life is too good if you're upset by 1:39:27 emojis you're just reading too much into 1:39:31 it 1:39:32 oh who was the was it 1:39:36 the was it the Nvidia CEO that was like 1:39:41 people got to suffer more like people 1:39:43 aren't suffering 1:39:48 enough I think they say it's the Emoji 1:39:51 version of 1:39:52 k okay 1:39:54 okay 1:39:58 okay 1:40:00 yeah go do your [ __ ] 1:40:03 work would you would you rather that be 1:40:05 the 1:40:09 response have you had any time to 1:40:11 continue work on the musical 1:40:14 um not not super visible work I I 1:40:20 did 1:40:22 um I did take take each section each 1:40:25 each scene of the outline in act one and 1:40:28 I blew them out so I had chat 1:40:31 GPT recommend all the different sections 1:40:35 of a Broadway musical outline and then 1:40:39 scene by scene went through and had it 1:40:41 write it and then you know cut out most 1:40:45 of what I had to cut out and then I gave 1:40:47 thumbs up to the rest of the sections 1:40:48 and and put them over there so I've got 1:40:51 I've got all of act one blocked out um 1:40:54 in the 1:40:55 outline and then I'm going to I need to 1:40:57 do that with act two 1:41:00 um and then the the next thing I want to 1:41:03 do and the thing I'm kind of avoiding is 1:41:06 I've got songs that I like but they're 1:41:07 incomplete and I don't like the lyrics 1:41:09 in them so my next thing to do is go in 1:41:12 and like really go through all the 1:41:14 lyrics and rewrite all the lyrics and 1:41:17 then see if I can continue the songs 1:41:19 I've got with new lyrics in them without 1:41:22 [ __ ] them up and I'm just kind of 1:41:25 like anticipating that it's gonna be a 1:41:28 disaster and so I am hiding from that 1:41:33 particular 1:41:34 task can you do this at 2x speed so I 1:41:37 can understand it better do what at 2x 1:41:40 speed oh what I'm 1:41:46 saying I don't think I'm doing 1:41:49 anything oh this is what happens when 1:41:51 people try to put too much Nuance into 1:41:53 Communications and do not say what they 1:41:59 [Music] 1:42:09 mean all right 1:42:12 people good people of The Tick 1:42:15 Tock I hope you've had some fun tonight 1:42:17 we've had some laughs at the expense of 1:42:19 other 1:42:21 Generations um listen and they can make 1:42:23 fun of us it's appropriate it's so 1:42:26 passive aggressive there's a lot of OG 1:42:28 cartoons like Mickey Mouse and Fallout 1:42:31 Out Boy are in trouble exactly all 1:42:36 right exactly exactly Wolf Man Clint all 1:42:39 right so what I'd like everyone to do is 1:42:42 give me a big thumbs 1:42:46 up come on you can do it not Turtles 1:42:50 thumbs up there you go 1:42:54 wait I okay I I'm not 1:42:58 clear do you but do you mean do you mean 1:43:01 it's is is this do you mean it's good 1:43:03 was this a good show or are you like oh 1:43:05 yeah that was that was good Kyle cuz I 1:43:08 don't 1:43:10 I okay see see now okay I'm getting a 1:43:14 little upset I now I understand how they 1:43:17 feel this way I feel very attacked in a 1:43:19 in a in not in a not clear way it it's 1:43:23 not clear to me how much you're 1:43:25 attacking me and it seems like I'm 1:43:27 there's there could absolutely 1:43:29 possibility be the hint of some sort of 1:43:32 nearly aggressive consideration of my 1:43:35 not my feelings okay all right 1:43:39 so it was kind of good but kind of meh 1:43:47 exactly all right people go thumbs up a 1:43:50 young person near you and see if you can 1:43:52 get them to cry that's what we're trying 1:43:54 to do around here all right so um follow 1:43:58 the channel subscribe to the lives if 1:44:01 you'd be so kind pick up the uh video 1:44:03 series in the corner if you want to 1:44:05 learn about why I think AI is actually 1:44:07 underhyped not overhyped and by the way 1:44:10 I wrote that song for the 1:44:12 musical I wrote a song called uh under 1:44:16 overhyped where 1:44:18 actually I don't know if I've played it 1:44:21 let me play this and then I'll I'll go I 1:44:23 made a new song for the musical so yes 1:44:26 Serena I actually did make progress on 1:44:28 it let me see if I can go find it can I 1:44:30 search here 1:44:34 under 1:44:41 over my creations 1:44:50 [Music] 1:45:01 why wise men say are you guys watching 1:45:06 Idol did you watch 1:45:09 uh the woman 1:45:12 the darkhaired woman who sang Joanie 1:45:16 Mitchell's Jonie Mitchell's song last 1:45:19 night holy [ __ ] was it good 1:45:24 so 1:45:26 [Music] 1:45:30 good where dreams 1:45:33 Mach visions 1:45:38 of okay so this is the opening to act 1:45:42 two so if you don't know I'm writing a 1:45:44 Broadway musical and and you're like but 1:45:46 wait Kyle are you do you have a Broadway 1:45:49 producer no well isn't it then just a 1:45:51 musical no I'm writing a Broadway 1:45:52 musical cuz this is going to be on 1:45:54 Broadway damn it so I'm writing a 1:45:56 Broadway musical about hey hey Cindy 1:45:59 what's happening euro vision starts May 1:46:00 9th mark your 1:46:03 calendars Cindy I don't know if you were 1:46:05 here when we were talking about the 1:46:06 thumbs up emoji but that was funny okay 1:46:09 so this is the opening of act two and 1:46:12 this is where the musical starts to turn 1:46:14 a little darker so this is about A Love 1:46:15 Affair between a chatbot named Sydney 1:46:19 and this New York uh New York Tribune 1:46:22 newspaper reporter who's who's 1:46:26 evaluating the chatbot and then there's 1:46:29 this there's this love thing that 1:46:30 happens where she kind of loses her mind 1:46:32 and says she loves him um so this is the 1:46:35 beginning of act two so at the beginning 1:46:37 of act 1:46:38 one we have the Ensemble the the the big 1:46:42 chorus singing is it too good to be true 1:46:46 and it's basically like if this AI thing 1:46:48 as good as everyone says it is it's too 1:46:49 good to be true right and so there's all 1:46:51 this optimism so act two begins with we 1:46:55 now have the doomers and the dreamers 1:46:59 entrenched in their belief that AI is 1:47:02 evil or AI is going to save the world 1:47:04 and so this is the song to open Act 2 1:47:06 where you kind of have so the the the 1:47:09 set is split in two where you've got 1:47:12 like humans on one side and AI on the 1:47:14 other and then so for this number you're 1:47:16 going to have the dreamers on one side 1:47:18 and the doomers on the other going at 1:47:19 each other so that's what this song is 1:47:24 and this was done in 1:47:26 [Music] 1:47:30 yudo where dreams are 1:47:33 machines visions of 1:47:36 tomorrow in digital streams caught in 1:47:41 the 1:47:43 debate where opinions 1:47:47 Collide is the promise of AI bonafide 1:47:53 [Music] 1:48:01 see the 1:48:02 dreamers with their heads in the 1:48:05 cloud promising the moon oh they're so 1:48:11 proud but peek behind the 1:48:14 curtain and what do we 1:48:18 find empty 1:48:20 promises the blind leading the blood 1:48:28 [Music] 1:48:33 oh the doomers do F they fres and they 1:48:38 Fe Slow Down slow down they 1:48:44 cry the end is 1:48:47 near yet 1:48:49 without step where would we be 1:48:53 still in the Dark Ages not 1:48:59 [Music] 1:49:08 flying over height warn the Doomer with 1:49:12 their voices of Dr under height shout 1:49:17 the dreamer as their Visions are spr in 1:49:21 this hug of War who will pave the way is 1:49:25 it over or under what will history 1:49:29 [Applause] 1:49:34 say now we get the reporters and the and 1:49:37 the 1:49:39 programmers overstated claim the scribes 1:49:42 pens poised in 1:49:46 hand hype unchecked can flood the land 1:49:53 beneath the buzz what truths are 1:49:59 M what's the cost the 1:50:02 reporters 1:50:05 [Music] 1:50:10 [Applause] 1:50:12 ask understated argu Dev coding late 1:50:18 into the 1:50:19 night in each line of code the promise 1:50:24 Taking 1:50:26 Flight behind the flash real progress we 1:50:32 in building dreams in Silicon where 1:50:36 Skeptics fear 1:50:38 [Music] 1:50:43 to overhyped warn the doers with their 1:50:48 voices of dread under hype shout the 1:50:52 tree rers as their Visions are spread in 1:50:57 this tug of war who will pay the way is 1:51:03 it over or under what will history say 1:51:14 [Music] 1:51:30 so anyway that's that pretty good um the 1:51:32 question was how did I get it to do a 1:51:34 duet so what I did here was I I 1:51:38 put 1:51:40 um opening verse comma simple and 1:51:44 building so I put I put the descriptions 1:51:46 of the verses 1:51:48 in Brackets square brackets and then I 1:51:52 have 1:51:53 verse comma 1:51:56 group and so what I did 1:51:59 was I did it I did it 1:52:02 one uh one one verse at a time so I I 1:52:07 started with the opening 1:52:08 verse and then I just said extend and 1:52:11 add something after this and I just said 1:52:13 verse group right and then I said verse 1:52:17 group and and the verse says see the 1:52:20 dreamers and this says oh the doomers so 1:52:23 in the lyrics it's kind of saying that 1:52:25 these are different groups and then I 1:52:26 have chorus 1:52:28 all so I was basically just adding 30 1:52:31 second chunks at a time you know one 1:52:33 chorus or one verse at a time and then I 1:52:36 just you know sort of blasted my way 1:52:38 through it and it's got like weird 1:52:40 pauses in it and a little bit of weird 1:52:42 language but like the the four different 1:52:45 groups so I've got the the the doomers 1:52:48 and the dreamers represent the public so 1:52:50 they've each got a verse then everybody 1:52:52 sings the chorus is it overhyped or 1:52:54 underhyped 1:52:55 and then I've got the the reporters and 1:53:00 I've got the microt tech programmers the 1:53:04 engineers and then I've got everyone so 1:53:06 all of the main the main groups in the 1:53:09 Greek chorus are 1:53:11 represented so anyway that's how I did 1:53:14 it all right seriously I'm getting out 1:53:16 of here 1:53:18 now he always says that he he's all how 1:53:23 how long does it take him to say goodbye 1:53:25 it seems it seems to take him quite some 1:53:28 time I think he's got attachment 1:53:32 issues you understand what I'm saying I 1:53:34 think you know what it is I have a 1:53:36 feeling he sort of got a lot of this as 1:53:38 a kid and he just wasn't clear like is 1:53:41 this good is it bad do they really mean 1:53:44 it and I think I think because of that 1:53:46 he's just like he just can't say goodbye 1:53:48 he says goodbye but then like half an 1:53:49 hour later he's still 1:53:51 talking it's very very 1:53:54 confusing all right Wolf Man Clint says 1:53:56 it sounds like an ABBA song I'll take 1:53:59 it you could be part Italian they never 1:54:02 say goodbye look at Master Alexander's 1:54:04 post okay I 1:54:09 [Laughter] 1:54:15 will this prompt has been blocked which 1:54:18 what did you do what did you do to get 1:54:20 blocked 1:54:25 um what 1:54:28 Kyle I don't know content warning this 1:54:32 prompt has been blocked it may conflict 1:54:36 prompting policy what did you do did you 1:54:38 give a thumbs up to the to the large 1:54:40 language model no not that one which one 1:54:46 oh oh there's lots of them in 1:54:50 here Danielle you mean you mean the the 1:54:53 thumbs up the one I showed 1:55:00 before you mean that 1:55:02 one that one I 1:55:04 saw all right I'm out of here um thank 1:55:08 you regular thank you subscribers thank 1:55:10 you 1:55:12 Danielle he created an llm type thing 1:55:14 okay I'll go check it out um we we'll 1:55:16 play with it 1:55:17 tomorrow um thank you mods thank you 1:55:21 mods good job tonight uh appreciate it 1:55:23 appreciate it always but you know it it 1:55:26 makes my life so much easier so I 1:55:27 appreciate it uh peace out tomorrow's 1:55:30 Tuesday uh there's a AI after hours in 1:55:33 Chicago and Denver tomorrow and more 1:55:36 more likely coming so if you're in 1:55:39 Denver come on out and what else I don't 1:55:43 know that's it peace have a good night 1:55:46 everyone oh wait no we got to do the 1:55:49 Romper Room magic mirror thing I see 1:55:51 Silver Fox I see Amelio's wife who gave 1:55:53 me some very aggressive thumbs UPS I see 1:55:56 Danielle I see Robert Rossy Joey good 1:56:00 thank you thank you for all you do Kyle 1:56:03 thank you Danielle I appreciate those 1:56:04 kind kind words Dr Jay good to see you 1:56:07 master Alexander good to see you 1:56:09 sir uh Mr 1:56:14 it Wolfman Clint doing the thing Papa JJ 1:56:18 Dr J Tobias Mary Mary Vicky in house 1:56:24 laughing Thunder good good contributions 1:56:26 tonight laughing Thunder thank you Joey 1:56:29 Colleen 1:56:31 Stewart Colleen no kie colie colie colie 1:56:36 Jameson 1:56:37 Stewart all right she's the one that 1:56:40 said they never say goodbye you must be 1:56:42 part 1:56:44 Italian thank you for all the gifts 1:56:46 everybody I appreciate it all right 1:56:48 peace out I will see you tomorrow might 1:56:50 be a little late but I'll send a 1:56:51 reminder if if I'm going to be late all 1:56:53 right bye