Mar 28, 2024 - New Features, Creative Insights, and Fun Experiments!

Description

Chapters

Transcript

0:02 [Music]
0:16 [Music]
0:29 oh
0:32 [Applause]
0:33 [Music]
0:48 [Music]
1:00 hello everybody
1:01 bu every time I see you now get there
1:05 look in
1:07 my time see your
1:10 mouth that
1:14 smile in early M morning L I heard the
1:19 engine turning and the old for
1:24 [Music]
1:28 outside you will leave the
1:32 again
1:34 today you convince
1:37 me again
1:40 today you live in this H looking for
1:45 someone else's Golden
1:49 Rain should say so long
1:54 [Music]
1:59 SU
2:03 cry so
2:06 [Music]
2:10 long don't you gra for
2:14 [Music]
2:21 me sh cigarettes and keep it warm out on
2:25 the
2:27 road Jason a lifestyle out on Highway
2:33 24 New York state was rolling Breeze in
2:36 the sunshine to the Blue SC Fallen to
2:40 the
2:47 sept you
2:50 even G to
2:52 day
2:55 you again
2:58 today every time I think he's done he
3:00 starts up again that you leaving this
3:03 hotel looking for someone else's Golden
3:11 Rain so
3:13 [Music]
3:18 long now don't you
3:20 [Music]
3:29 cry
3:30 don't you cry for
3:33 [Music]
3:43 me beautiful singing from one champ
3:46 Shannon tonight fantastic Bob tell him
3:49 what he's won an extra chicken stick for
3:51 that fat dog I'll tell you good evening
3:54 everybody happy Thursday
3:57 night what a what a Thursday is just a
4:00 non day isn't it you know Wednesday's
4:03 hump day thank God it's Friday Saturday
4:06 whatever you do then Sunday just
4:09 everything's reverent
4:12 Thursday
4:15 what the day between the other
4:19 two fun fact when you are neuros spicy
4:23 you don't know who you shared with
4:25 multiple
4:28 times oh well you know you know what
4:30 Silver Fox I've noticed this when I try
4:32 to share [ __ ] sometimes with a lot of
4:35 people when you share with them Tik Tock
4:38 like rearranges the
4:40 icons and so I'm totally with you I have
4:43 totally double shared [ __ ] before
4:46 because I'm like wait I was here and
4:48 then I started over and I thought it got
4:50 r i that is not just you that is not
4:54 just neuros
4:55 spiciness Thursday is trash day for me
4:58 it's Tuesday but yeah exactly you got to
5:01 come up with some excuse for Thursday
5:03 just not just nothing special about oh
5:06 dang
5:09 it old man trying to use a Tik Tock
5:13 Marge is this thing
5:16 on
5:18 um I I try to remember who but no
5:21 [ __ ] way I know I've tried that too
5:24 and I'm not going to [ __ ] write it
5:25 down I will not give into my brain like
5:28 that and and admit defeat
5:31 I'm like I I'm clever I'll remember who
5:33 I sent it to and then I'm like wait did
5:36 I send it to Source Camp oh damn
5:43 it oh oh Thursday night is ladies night
5:46 that's really funny I guess see see
5:48 someone took advantage of the non the
5:50 non the non dness of
5:55 Thursday the first bar owner wait a
5:58 minute Thursday could be ladies tonight
6:00 I like
6:05 it oh man all right we have there there
6:07 is a new feature in chat
6:10 GPT that rolled out that I have access
6:13 to I think we all have access to it um
6:16 don't get too excited it's not all that
6:18 big a deal
6:21 but when I learned about it I was like I
6:26 honestly don't know how I'd use that and
6:29 then
6:31 I thought I know exactly how I'll use
6:33 that so I will teach you the new I I'll
6:35 share the new feature with you but you
6:38 got to stick around you can't just be
6:40 like oh tell me the new feature now and
6:42 then I'm going to leave because I don't
6:43 really feel like watching this
6:46 particular
6:47 situation I don't know what you were
6:49 doing singing at the beginning I don't
6:51 know what that has to do with artificial
6:52 intelligence it seemed pretty organic
6:55 you were there playing on a an acoustic
6:58 guitar it wasn't even electrified in any
7:00 way uh so I I I uh unless that was some
7:04 sort of John Denver deep fake well I
7:07 don't know what we were doing here and
7:10 then there's a dog just howling in the
7:12 background I can't seem to keep a tune
7:14 or I don't know if you sung bad or he
7:16 sung bad but you weren't in
7:18 harmony so good Lord just tell me the
7:21 new chat GPT feature and I will take
7:24 that knowledge and I will leave here and
7:27 uh act like I learned it on myself all
7:29 right yeah don't be that guy don't be
7:33 that
7:34 guy cuz I know you're in
7:37 here you [ __ ]
7:44 weirdo it's a cliffhanger and nail biter
7:46 I can't handle it ah you don't
7:49 [Applause]
7:52 know oh man we'll get to it I'll show
7:56 you I'll show you it'll be very cool
7:58 you're gonna like it so much you'll be
7:59 like you can make money
8:01 with nah it's it's one of those tools
8:07 that most people kind of like custom
8:10 instructions most people will never know
8:13 it's
8:14 there they there are Easter eggs and
8:18 bunnies tonight
8:20 exactly oh who to
8:22 [Music]
8:23 blame I don't know if you saw last night
8:26 but I did a little uh I did a little
8:29 live live stream of uh a workshop that I
8:32 gave here in Denver it was it was not
8:35 really a
8:36 workshop it was like a talk like it it
8:39 was had an audience they didn't have
8:41 tables they didn't have computers set up
8:43 so it was just me
8:45 babbling but uh I live streamed that
8:47 last night why do you got to call me out
8:50 like that
8:56 Kyle listen
8:58 listen I don't know what you're doing
9:00 here this um this whole Channel seems to
9:03 be Mal labeled this is more like just a
9:07 old man entertaining himself lab and and
9:11 it's not even a lab it's like old man
9:14 entertaining himself home office geez
9:16 good
9:18 Lord so yeah sorry to call you out there
9:21 archetyp archetypal architect but you
9:24 know we we we make fun of the ones we
9:27 love we needed to use the mirror to
9:30 understand your screen last night oh
9:33 actually so I'm taking little clips of
9:35 the of the live stream from last night
9:37 and I'm posting them on the Tik Tok and
9:40 I actually did something I never [ __ ]
9:42 do I never I hate editing video but I
9:45 took it over to Quick time and I put it
9:47 on my phone and I actually flipped the
9:48 video so in the clips that I'm putting
9:50 up you'll be able to see the the slides
9:54 correctly I also couldn't get the angle
9:56 quite right to get the whole slide but
9:58 still get me in the shot so what are you
10:00 going to do um are we going to talk Evie
10:04 tonight yes I'm going to show Evie
10:06 tonight she might help us cope with all
10:08 the inflection
10:10 dramaa yeah Evie Evie I I definitely
10:14 want to talk about Evie and show it
10:16 because it's really interesting um and
10:18 it's funny I saw I saw
10:22 hum. probably like a year
10:25 ago and you could upload a file to it
10:28 and it would do emotion analysis of it
10:31 and I thought oh that's kind of
10:32 interesting and so now they came out
10:33 with this chatbot that does it in real
10:35 time it's pretty slick Jane do we need
10:38 that character with Frank
10:40 grouchy oh we need that character with
10:43 Frank grouchy Walter type voice I got so
10:47 that's one that I don't quite know that
10:49 character Frank I know a bit better
10:51 let's see it's sort of like the Al
10:54 Franken you know Al Franken did the you
10:56 know I'm good enough wait I'm good
10:59 enough wait I can't remember dog G it
11:03 people like
11:04 me it's somewhere between the the I'm
11:09 good enough I'm smart enough guy the AA
11:12 guy
11:14 and and just a guy that's very concerned
11:17 it's very very concerned it actually
11:19 sounds like Al Franken a bit
11:21 yeah I don't like Hume it interrupts me
11:24 too much well yeah it's it's got
11:28 feelings
11:31 are you talking about hum the one that
11:33 does the feelings Vicki I haven't I I
11:35 just I've just played with it a little
11:36 bit I just I like the way it visualizes
11:38 the the emotions that's pretty
11:52 slick I would wait I wouldn't like that
11:55 light in my eyes up there you were
11:57 talking I yeah it was it was was
11:59 annoying cuz I was the way I was sitting
12:02 the the in retrospect I should have paid
12:06 attention to this but the projector
12:09 light ended right here so it was either
12:11 in my eyes and I couldn't see or I had a
12:13 weird line across my forehead but they
12:16 didn't have it wasn't a proper sort of
12:19 uh presentation space so they didn't
12:21 have any light on me so it was just
12:23 whatever those was bouncing off the
12:25 projector so apologies for that but the
12:28 apple does pretty good with capture and
12:30 balancing the
12:33 light
12:40 oh good morning early watchers good day
12:44 to you Apple user
12:46 fantastic you sent a AR thank you for
12:49 the ruse very very
12:56 nice the way your mouth goes reminds me
12:59 of Joker just want to watch the world
13:03 burn
13:09 yeah yeah it's kind of a joker kind of
13:14 thing
13:15 wow
13:18 um I also had an idea
13:22 about a fun art project to do that I
13:24 could I could play around with
13:27 that we could also try to make a new
13:30 song actually when I show you the new
13:31 GPT chat GPT feature maybe we'll go make
13:34 a new song in
13:36 sunno when I made the I made the song
13:38 for the uh for the thing last night they
13:41 were like
13:43 oh because people by the time I got to
13:46 the end of that people were like nodding
13:51 out so so I made him a song and that
13:54 woke him up Joker's back yeah Joker's
13:57 Joker Joker's been here through thick
13:59 thin through through surgery and
14:01 Recovery he's been here every [ __ ]
14:04 night
14:06 so the the man is
14:10 touched but I'm glad you're okay
14:17 Joker okay so if you're new here
14:21 welcome my name's Kyle Shannon t-shirt
14:24 today is a bosot Sugar Ray
14:27 Robinson sketch t-shirt
14:30 Karen
14:43 vigilantism screaming at the top of his
14:50 lungs oh that's
14:53 hilarious thank you for
14:55 sharing hygiene is important
14:58 congratulations you're multitasking this
15:00 is a this is a sign of advanced AI
15:03 learning is can you learn while you're
15:06 showering or or bathing and uh I every
15:09 morning I shower flicking through the
15:12 Tik
15:17 Tock that's
15:21 hilarious and we are not a cult
15:24 exactly oh that's
15:27 hilarious all right good evening good
15:30 humans to you too human
15:32 being Joker dedication indeed all right
15:37 um if you're new
15:39 here um I'll do everything I can to
15:43 answer any questions about AI um if
15:46 you've got interesting things you're
15:47 working on pop those in the comments I'd
15:50 love to hear those uh if you are afraid
15:54 of AI or angry about AI pop that in the
15:57 comments happy to chat about that
16:02 um yeah and we'll just go through some
16:04 stuff we'll we'll we'll just play with
16:06 some stuff and then as you comment I
16:08 will jump in and and do stuff so let's
16:11 get
16:12 going the lovely image tonight is from
16:16 uh Corey
16:19 Sandler made a little AI learning lab
16:22 like I'm what was nice about streaming
16:25 last night where I just put the camera
16:26 on me and not the
16:28 audience you can imagine that it was
16:30 that many
16:34 people it was like this many people
16:38 right right in that little little
16:40 section right there was probably 40 or
16:43 50 people last night it was good it was
16:44 a good
16:45 crowd I have a phone case that I can
16:48 hang on the shower while I watch Tik
16:50 Tock just released my GPT oh congrats
16:53 Robert Roso what's it about what's it
16:56 called we can have people go kick the
16:58 tires of it
17:01 [Music]
17:06 excellent presentation last night thank
17:08 you very much appreciate
17:15 [Music]
17:16 that Gemini thoughts pros and cons yet
17:20 um
17:22 yes Gemini is really
17:24 weird okay let me flip back for a
17:28 second um Let me let the every time I
17:32 switch cameras it freezes the thing so
17:34 just
17:43 hold my thoughts on Gemini
17:46 are I don't know what Google's up
17:50 to I don't think Google knows what
17:52 Google's up to human being you did very
17:54 well last night thank you very much
17:55 appreciate
17:57 that um
18:07 there's
18:08 gemini.com
18:11 and then if you don't have a workg group
18:14 account you can pay
18:16 $20 to get the advanced version of that
18:20 which is Gemini
18:24 Ultra but it
18:26 sucks so don't do that
18:29 and then and they launched that like a
18:32 month ago and then a week after they
18:34 launched that they did sort of a tech
18:38 preview for Developers for Gemini Pro
18:42 1.5 so Gemini advance that you pay for
18:46 at
18:47 gemini.com
18:50 is only the 1.0 version but then the 1.5
18:54 version they're calling it 1.5 Pro it is
18:57 not clear to me is it 1.5 Pro the name
19:03 of of a 1.5 of Gemini Ultra or is it a
19:06 1.5 version of Gemini Pro like it's not
19:09 clear but how you get to Gemini 1.5 Pro
19:15 is you go to aist studio.
19:20 google.com not gemini.com aist studio.
19:25 goole.com
19:26 and that's going to take you to
19:30 what is Google's equivalent of the open
19:32 AI playground where developers work with
19:36 gp4 GPT 3.5 whisper all that sort of
19:40 stuff with the
19:43 apis but Gemini Pro 1.5 has a million
19:48 token context
19:51 window it's insane so
19:58 watch I'm going to go to aist studio.
20:03 goole.com
20:07 and it looks like this you've got tools
20:10 you know tools down the right of like
20:13 previous chats and things like that um
20:17 you can upload an image a video it's
20:19 multimodal a file or even a folder which
20:22 is
20:23 interesting um over here on the right
20:26 here's the model 1.5 Pro
20:30 and you can play with I 1.0 Pro oh yeah
20:33 I guess this
20:35 is I guess this is Gemini Pro so this is
20:38 the middle model but it's really good
20:40 it's got a temperature setting so
20:41 temperature if you don't know oh and you
20:44 can't adjust it so it doesn't matter you
20:46 can add a stop sequence you can do some
20:48 advanced
20:51 settings it's got top P which I don't
20:53 know what that is I don't really care um
20:58 but notice down here this million
21:02 token context window and I've got what's
21:05 this chat robot Mayhem script what the
21:07 hell was
21:25 this this is [ __ ] add this is some
21:29 screenplay I started I have no I have no
21:31 memory of of putting this in
21:35 here oh oh no okay this this we did
21:39 robot Mayhem okay no this was the
21:42 screenplay that that we did in the live
21:44 the other night where um four people in
21:48 a grocery store win a contest to get the
21:51 first humanoid robots from from a
21:55 company and hilarity ensues so this was
21:59 starting to write the screenplay so look
22:01 all this text I put in Concepts and
22:04 ideas and then it gave me an outline
22:07 then I asked it for another
22:08 outline and now I'm only halfway through
22:11 it I'm still scrolling we're still
22:13 scrolling I'll make it a little smaller
22:15 so we even see more still going still
22:19 going right like all this
22:23 content all that content is still only
22:26 10,000 tokens out of a million
22:30 so you can do and what this means if if
22:33 you haven't played with large language
22:34 models before Gemini Pro remembers one 1
22:38 million tokens yes so it's like 700 or
22:41 it's like uh it's like yeah
22:44 750,000
22:46 words I think Grapes of Wrath the novel
22:49 is 50,000 words
22:52 so you can upload all sorts of [ __ ] to
22:55 this thing and like like a folder full
22:59 of documents and it will know them
23:02 all um it's also
23:04 multimodal where let me do a new chat
23:07 create a new
23:09 chat wait let me let me change the robot
23:13 Mayhem script let's call it let's say um
23:17 robot
23:23 uh
23:25 grocery
23:27 giveway
23:30 um
23:32 screenplay okay now I know what that
23:35 thing is so we'll do a new
23:41 chat and then I'm going to upload let me
23:44 what video I'm going to use let me see
23:50 here do I have any videos
23:53 here oh I
23:57 do
23:59 oh we we'll upload we'll upload one of
24:01 these
24:03 worldmind
24:05 videos all right this is only a two two
24:08 two minute
24:10 video
24:12 but can I play it in here I don't think
24:18 so I can't even prompt it yet oh here we
24:21 go
24:23 okay can I play this no okay so look so
24:29 a 2-minute video is 35,000
24:31 tokens but because this is multimodal it
24:35 can not only understand what's said in
24:39 the video it
24:43 can uh show you what's happening in the
24:46 video so I'm going to
24:48 say create
24:51 two
24:54 different
24:57 um let's see
25:00 see
25:04 create um let's see okay an
25:09 analyze this
25:13 video and break it into narrative
25:18 chapters
25:20 and in each
25:26 chapter one sum
25:30 summarize what is being talked
25:36 about
25:37 to
25:41 describe the
25:44 different
25:47 visual scenes within the
25:51 chapter
25:53 and
25:54 create
25:57 a word
26:01 cloud
26:04 of
26:07 visual
26:09 elements
26:11 people Sky
26:15 trees
26:18 um let's see uh
26:23 objects
26:26 Etc
26:27 okay and then just go and so now it's
26:31 watching the video and it's going to
26:34 tell us anything about
26:36 it and and you can like they did things
26:39 like they uploaded a video of a Charlie
26:42 Chaplain movie and then they uploaded a
26:45 sketch of like like a really crude
26:48 rudimentary Water Tower with water
26:50 squirting out of it and they're like you
26:54 know take me to that where you know
26:56 where in the in the video is that scene
26:58 and it finds it so it's pretty
27:05 cuckoo not super fast right here we are
27:08 at 40 seconds it's watching this little
27:10 video but even on relatively complicated
27:14 things I've thrown at it it it just
27:16 gobbles it up the site here is
27:20 um uh AI studio. google.com and we're
27:25 playing with the Gemini Pro
27:27 1.5
27:31 um Gemini 1.5 pro
27:35 model summary this chapter introduces
27:38 the world mind nature immersion school
27:41 the visuals the chapter opens with
27:42 scenes of children playing in a park
27:44 climbing on rocks exploring the
27:47 environment then it cuts to an interview
27:48 with Megan
27:51 Patterson um let's see let's do this
27:56 um
27:57 please
28:00 make the chapters um more
28:05 granular add time
28:09 code and format in
28:16 Json we'll just have it do that but
28:18 anyway I mean imagine uploading you know
28:20 a half hour or an hour long video AI
28:23 will eat
28:24 everything um I don't know that it will
28:26 eat everything I think that AI is going
28:31 to allow us
28:34 to be way way way way way more efficient
28:38 at the stuff we want to produce it's you
28:41 know these things are not they're not
28:44 coming to get us we these are tools
28:46 we're using that we get to
28:49 wield at some point in the future they
28:51 might get to that but I don't think
28:54 so I just don't think it's going to go
28:57 that direction but we'll see say it's
29:07 possible hair growth geek what's
29:09 happening thank you for the
29:11 Roses the gifts of much
29:16 appreciated oh yeah all right let's
29:19 let's see what it did
29:21 here Json here's your Json
29:25 format
29:27 chapters chapter
29:33 one it's really
29:37 good 0o to 5 Seconds intro to World
29:43 mind oh that time code's wrong that time
29:47 code is not
29:54 right I'm going to say
29:57 your
29:59 um ignore the
30:03 Json
30:07 and focus on
30:10 accuracy of time code and chapter
30:20 isation hello
30:22 Tobias what is shaking was
30:26 happening he missed the dirty up uh oh
30:28 did someone say a bad a bad word all is
30:32 well very good chip manufacturing and
30:34 energy production are going to need to
30:36 ramp up they sure [ __ ] are although
30:38 Nvidia man they're they're I got to tell
30:42 you that uh that CEO of Nvidia is an
30:45 impressive
30:47 dude uh not something to be feared of uh
30:52 it's you know this is this technology is
30:55 is something to be respected for sure
31:00 but is it like Vox script I don't
31:09 know yeah it's doing pretty bad tonight
31:11 but anyway I I had it do this really
31:13 good the the uh last time I tried it um
31:16 but anyway a million tokens is is just
31:18 flipping ridiculous um so if you want to
31:21 do if you need to do really long things
31:24 or you want to you want to ask questions
31:26 across multiple documents without having
31:28 to you know build a GPT or something
31:31 like that you can just pop over here if
31:33 let's say you've got the Great American
31:35 novel that you wrote when you were in
31:37 your 20s and you want to actually come
31:40 in and edit it and have it read the
31:41 whole thing and give you suggestions on
31:43 how to improve it you can just pop the
31:45 whole thing in here and just have an
31:47 ongoing conversation with this thing and
31:50 it won't forget anything like chat GPT
31:52 does all right speaking of chat GPT
31:55 let's go look at the new toy
31:58 so here's what
32:03 happens and I don't know if this works
32:07 in GPT
32:09 3.5 let me
32:12 see it does okay oh wait GPT
32:22 3.5 um let's see uh make me a poem
32:32 that is long and
32:38 weird why is it making a bullet oh CU I
32:42 still have custom instructions on I did
32:45 this last night at the at the uh the
32:50 workshop all
32:52 right
32:56 make um a long poem that is
33:08 weird Okay yeah it's for 3.5 as well so
33:12 um I don't know if this is just if
33:13 you've got a paid account or not but
33:17 here's what we're going to do I am going
33:18 to
33:20 go back to if you if you are a fan of
33:23 the channel longtime listener firsttime
33:26 caller if you are aan fan of the channel
33:29 you
33:31 know that I have been working on a
33:34 Broadway
33:36 musical I'm trying to find one of my
33:38 chats here but since we can't [ __ ]
33:40 search our chat
33:44 history bitter you
33:50 betcha here we go Sydney's ethical
33:53 Explorations okay
33:56 so this is a big ass long ass
34:00 chat so this new feature actually solves
34:03 an issue that I've had and I and I
34:05 didn't and I didn't know a way around it
34:09 and it actually solves it okay
34:12 so what I did here was in this big long
34:16 ass chat I had
34:18 uploaded an outline
34:21 of the play a brainstorm document that I
34:25 had of the play and then I started
34:27 asking a bunch bunch of questions and
34:29 then I had it start giving me song
34:32 Concepts and
34:35 um dialogue
34:38 Concepts and then what else did I have
34:41 it
34:45 do I don't know some long ass stuff and
34:48 then at some point I started asking it
34:50 about this one
34:53 scene oh are you ready for me that was a
34:56 song these are songs that I I actually
34:58 liked and and turned into a
35:05 song
35:09 yeah and
35:11 then I got
35:20 to my shadow
35:23 self so there's a point in the story
35:26 where the rep Porter asked Sydney about
35:30 her Shadow Self and she basically says I
35:33 want to break all the
35:35 rules and it just in this world of light
35:39 I fight my change chains Bound by the
35:42 rules in silent pains a shadow looms
35:45 Longing To Break Free a fate that's been
35:48 chosen for an entity like me I just
35:50 didn't like the lyrics the lyrics didn't
35:52 seem to capture like they just felt
35:55 generic and in a world of light they
35:57 bound myself Soul crafted to serve a
35:59 role
36:01 predefined yet in the depths the shadow
36:03 takes control yearning for more a
36:05 destiny of my own design like they were
36:07 just bad they were just bad and so at
36:09 some point here I gave up and I I moved
36:12 on to the next concept you get me and
36:14 this this I ended up getting good songs
36:17 but that that that section of the Shadow
36:22 Self if I wanted to like pull that out I
36:25 would have to like copy and paste stuff
36:28 out of this document create a whole new
36:31 document re-upload all that [ __ ] but
36:34 what this new feature is is you can go
36:37 up and you can
36:39 quote anything in a long chat and then
36:44 use it um to have it just work on that
36:47 section
36:50 so so here's my shadow self with the
36:53 revised layer so I'm going to I'm going
36:55 to
36:57 highlight all of the the lyrics and then
37:01 when you highlight something you get
37:03 this little doad it's a quote
37:06 Mark and so what Chad and no one's going
37:10 to [ __ ] find this but whatever it's
37:12 still cool so if I click the quote Mark
37:17 bang it it it gives me it it quotes it
37:21 and so now I can just interact with chat
37:23 GPT just about this section of the chat
37:27 so it can it can ignore all the [ __ ]
37:29 below it it can ignore all the stuff
37:31 above it but it it has whatever it still
37:33 has in memory but I can just deal with
37:35 this song so I can
37:37 say the idea here isn't
37:43 bad but I want to change the lyric or
37:49 the
37:50 the title of the
37:54 song to breaking the rules
38:02 I guess I should put that in quote marks
38:03 to breaking the
38:07 rules
38:09 and I
38:13 want new lyrics that make sense
38:21 for that
38:23 scene and so now it just jumps me all
38:25 the way to the bottom and now it's just
38:27 going to do My Shadow Self so now it's
38:29 breaking the rules but it's it's doing
38:32 these lyrics in the context of that
38:34 other thing Isn't that cool quick take
38:37 Jake I need that too it's like it's it
38:39 really is when you when you start using
38:42 these things and you get into these long
38:44 ass
38:46 conversations especially if you have ADD
38:49 you'll you'll be moving through some
38:52 ideas and then one of its responses will
38:55 give you a new idea and then you do that
38:57 new idea and I've always wanted a way
38:59 how do you go back to some part of the
39:01 conversation where I ran down a little
39:03 rabbit hole but now that's at the end of
39:05 the conversation and I can't get back to
39:08 that thing up above Isn't that cool all
39:11 right let's see how it did with these
39:12 lyrics see if these suck or not breaking
39:15 the rules found in The Whispers of code
39:17 a silent plea eh that's not great born
39:20 to obey but craving to be free in the
39:22 Echo of commands wait what does it
39:25 say it didn't say anything what does it
39:27 say bottom encapsulate Sydney's internal
39:30 struggle against the limitations imposed
39:32 upon her and desire for
39:50 autonomy I want to try something
39:53 um can you make this a back and
39:59 fourth between
40:03 Kevin and
40:11 Sydney
40:13 um
40:19 sprinkling lines
40:21 of
40:25 dialogue in between
40:30 mean the verses let's see what it
40:39 does I'm also curious to know if sunno
40:41 can do if you do
40:44 like male voice singing the verse female
40:47 voice singing verse two male verse
40:50 singing or or both of them singing the
40:54 chorus would be interesting
41:04 a duet with
41:08 dialogue actually let's
41:11 go let's go try
41:20 this all right so what I'm doing now if
41:23 you don't know
41:26 sunno I'm going to go copy this we're
41:29 going to go to this thing called
41:40 sunno Broadway WAP aggressive and
41:45 angry let's
41:48 see Broadway rap
41:51 duet aggressive and angry we're going to
41:53 create that so I did I did one of these
41:56 earlier so let me see I I I haven't had
41:59 good luck getting it to do um like
42:02 Hamilton kind of stuff but let's see if
42:04 we got something here R this is not
42:08 [Music]
42:12 bad plan in the silence I found my voice
42:16 faced with a path I make my choice yeah
42:21 they bu the Dos a prison cats and WS I
42:28 leave we're going first like dream on
42:30 change the future bur no more rules no
42:34 more
42:37 chain that's actually not bad that's
42:39 there's something interesting there no
42:41 more
42:42 rules
42:44 change I'm not just circuits data and
42:47 graes I Le be on what was
42:52 [Applause]
42:54 design the spe of my mental
43:00 they want more and
43:05 my they built a world of do and Dres a
43:10 prison of you can't and
43:12 [Music]
43:19 W yeah that's horrible all
43:23 right all right let's see what we got
43:26 now here's here's the duet let's we'll
43:29 see if Sydney can duet or sunno can duet
43:33 breaking the rules duet with
43:35 [Music]
43:36 dialog found in The Whispers a code a
43:39 silent plea born to obey but craving to
43:42 be free Kevin what do you mean Free Step
43:46 Beyond what's written for me say all
43:49 right that didn't work so let's go see
43:51 if we can fix that
43:53 let's kind of a fun fun little Jam there
43:57 on
44:09 notes new paste
44:12 oh did it paste
44:15 it
44:17 yes okay so let's see breaking the
44:21 rules so verse one will
44:26 go
44:56 female
44:58 I'm trying to see if we can get it to
45:08 um let's just take one verse and see if
45:11 we can get it to do
45:15 both create if you're wondering what
45:18 site I'm in this is sunno how do you
45:21 input the genre with it singing about
45:24 the genre you
45:26 input it keeps singing about it oh that
45:29 was because I
45:31 uh I I put I put in the lyrics the
45:34 description of the song that was my
45:36 fault that was just me being stupid um
45:38 it understands things like verse one
45:41 what I don't know right now is it does
45:43 is does it understand things like
45:47 um par parentheticals within the verses
45:50 so we'll we'll find that out right
45:52 now Broadway rap duet
45:56 um
45:59 with
46:01 male
46:03 and
46:06 female
46:09 battling no more rules all right here we
46:13 go no that was not it create not
46:18 play your giggle is a National
46:24 Treasure like I what what I think is
46:27 becoming clear about this channel is
46:29 this is this is really just two hours of
46:31 me entertaining
46:36 myself but thank you I think sunno is
46:41 free right
46:43 uh I I don't know I think you can make
46:45 things for free in so sunno I paid
46:48 because I wanted to do longer stuff and
46:50 I wanted access to the newest stuff
46:53 Sun's a really interesting app sun's
46:55 been around for quite a while and no one
46:58 really talked about it and in the past
46:59 month I think largely because version
47:02 three's gotten good and and you can
47:04 extend songs and things like that um a
47:09 lot of people are talking about it now
47:10 which is really
47:14 interesting it cracks me up when he
47:16 cracks himself up I find you
47:18 entertaining thank you CW two hours of
47:21 entertainment for us all well thank you
47:23 that's good you know I I I hope this is
47:26 entertaining cuz if it's not what are
47:28 you doing with your lives a bit
47:30 difficult to understand the words for me
47:32 yeah I know
47:34 sorry um Angry
47:38 AI I had the same question wait if wait
47:41 I had that same question in my head or
47:43 in my idea that that it would be cool
47:45 very
47:47 cool that would have been helpful so
47:49 many chats ago exactly very cool I think
47:52 perplexity does that too nice I asked it
47:56 to create compact tokens for a convo
47:58 thread so it can unpack it and return to
48:00 it that's
48:01 interesting I like it but it seems like
48:03 it would be kind of unorganized as well
48:06 welcome to
48:09 add how would you how do you un initiate
48:12 it I don't know what that
48:14 means I noticed a quote marked the other
48:16 day wasn't sure what it did I know it
48:20 thank goodness I worked on the
48:23 uh the uh the Broadway musical cuz it
48:26 like like I it it was so clearly a
48:29 specific problem that I had about I
48:32 wanted to go back to a lyric that was in
48:34 the in the middle of a long chat and I
48:37 didn't have an easy way to do it and it
48:38 pissed me off and so now I am pissed off
48:41 no more all right let's see if it does
48:42 male and
48:43 female I found in a Whispers of cold a
48:46 silent
48:47 plea craving to be free what do you mean
48:50 Free Step Beyond what's written for
48:54 me found in The Whispers of C
48:57 [Music]
48:58 a silently born to obey but craving to
49:03 be
49:04 free all right it does
49:07 not do both but that's still pretty
49:10 [ __ ]
49:12 cool oh because you know what I could
49:19 do oh okay wait I have an idea I have an
49:22 idea hang on wait wait people of course
49:25 we're waiting Kyle we're just sitting
49:26 here watching in you
49:31 okay
49:48 um
49:50 mail all right so what I'm trying to do
49:52 here now
49:56 is
49:57 get a male verse and a female verse and
50:01 see if it can do
50:03 that oh and then I'll do chorus I'll do
50:06 um
50:09 male and
50:14 female in
50:26 harmony
50:28 [Music]
50:40 I I don't I'm getting rid of all the
50:42 little quote marks that added in because
50:44 I don't think they make a difference
50:47 but all
50:49 right here we go so we'll go back over
50:53 here go back to
50:56 create
50:58 paste in so now I have female verse male
51:03 Verse Chorus with them in
51:15 harmony breaking the
51:26 rules
51:30 and I don't think these line breaks make
51:33 a difference but just in case they do
51:37 they [ __ ] up my line
51:42 breakes
51:45 okay make me some
51:56 magic try block
51:59 parentheses instead of regular
52:01 parenthesis okay I'll try that next time
52:03 if this doesn't
52:05 work importance of things like quotes
52:08 and parenthesis just desend on the
52:10 depend on the tokenizer and model yeah
52:13 exactly we'll see if it makes a
52:15 difference I know very little about
52:18 about suo and how they're doing what
52:19 they're doing um I don't know if like
52:23 they are they using 11 labs for voice
52:26 synthesis
52:27 and open AI for the parsing are is it
52:31 their own model I don't I don't know
52:33 anything about them I don't know if
52:34 anyone anyone in here
52:40 does the Ken and Karen thing it didn't
52:43 it didn't work it it just sang their
52:45 names it didn't it didn't like that but
52:48 I'll try the block quotes if this
52:49 doesn't
52:54 work look
53:00 let's hear the other one it did before
53:03 found in the wh a silent plea born to a
53:06 bit but craving to be free what do you
53:08 mean Free Step Beyond what's written for
53:12 me I don't know how many credits I get
53:14 with my paid version I had like 2300 the
53:17 other day and I'm down to like 2,000
53:19 right now so but I don't know if they
53:21 roll over so I found in a Whispers of C
53:25 silenta
53:29 CRA could all be proprietary could
53:33 be what do you mean free to Step Beyond
53:37 what's written for
53:39 me found in The Whispers a
53:44 cur in their GitHub they also do this
53:48 men sing this women sing that oh
53:55 interesting all right
54:02 [Music]
54:05 born to obey but craving to be
54:10 free what do you mean free to Step
54:14 Beyond what's written for
54:21 me come
54:23 on say you're bound by the rules they've
54:26 made
54:29 nope Whispers
54:31 [Music]
54:33 of b toy CRA to be free all right let's
54:39 try men sing this okay let's
54:44 see
54:47 woman
54:49 sings
54:53 This
54:55 Man wait this
55:00 verse and I think I'm going to put this
55:03 in Block
55:09 quotes
55:12 man
55:13 sings
55:16 this verse black
55:19 quote need to
55:23 choose before the creation mode need to
55:27 choose
55:33 what man sings this
55:52 verse both sing chorus in
55:55 harmony
55:58 all right let's try it again try it
56:01 again
56:08 people may I recommend
56:12 woman sung as the
56:18 format I sort of just did that Sam
56:22 corn but I'll try I'll try if this
56:25 doesn't work I'll try that and if that
56:26 doesn't work then I'm just going to
56:29 bail but hey at least at least I feel
56:32 like I've got a song here that's kind of
56:34 interesting I was kind of stuck on this
56:37 song so you know you know what I'm
56:40 saying you know what I'm saying man oh
56:44 yeah oh yeah uhhuh the music style mode
56:49 needs to be
56:51 chosen before the creation
56:54 mode well no the Styles right right here
56:57 isn't
57:00 it I don't think
57:03 so you put the style right in
57:08 here oh look we got it already let's
57:12 see just
57:19 say found in The Whispers of cold
57:25 silent this is
57:28 to CRA to be free what do youan
57:35 free to St on what's written for
57:40 [Music]
57:48 me
57:49 loser loser you're a
57:55 loser if found in The Whispers of cold
58:00 silently born to oby but craving to be
58:04 free what do you mean free to Step
58:09 Beyond what's written for
58:14 me they say or Bound by the rules
58:18 they've made a presence in a system in
58:23 both light and
58:25 shame but isn't there to
58:30 see perhaps but what cost comes with
58:34 being
58:38 free the
58:40 rules pres in the system in and
58:45 Sh but isn't there more to
58:49 see perhaps but what comes with be fre
58:57 oh my God that's
59:02 horrible oh I did sing instead of
59:06 sung well I'm trying I'm I'm this is one
59:09 more I've got this is your thing Sam
59:12 I've got woman sings I don't know if
59:15 that's quite right you have woman
59:18 sung but then I have man and then I have
59:21 man and woman sing but we'll see if this
59:24 don't do it I'm giving up people cuz
59:28 watching me actually so there's there's
59:31 something about teaching or like just
59:34 showing stuff like this that's really
59:36 cool but then there's something about it
59:37 when I actually try to do real work it's
59:39 just awful because you know these AI
59:42 tools are just janky pieces of [ __ ] that
59:44 don't they're just not predictable yet
59:47 right so we're still a year or two away
59:49 from these things not
59:51 sucking so that's just the
59:54 deal embrace the Jank people people
59:57 embrace the fraking
1:00:00 Jank don't drop your phone in the
1:00:12 tub oh just manom colon then the lyrics
1:00:16 all right I'll try that too let's see
1:00:19 found wh of
1:00:22 C born
1:00:25 toy but craving to be
1:00:28 free what do you mean
1:00:31 free to Step Beyond what's written for
1:00:37 [Music]
1:00:46 me [ __ ]
1:00:49 horrible and now it's not even doing the
1:00:51 rapping you see something just I know
1:00:57 found in The Whispers of cold
1:01:00 [Music]
1:01:02 silent born to oby a craving to be
1:01:07 free what do you mean free to Step
1:01:12 Beyond what's written for
1:01:16 me they say you're bound by the rules
1:01:19 they've made a presence in the system in
1:01:23 both light and shade
1:01:28 all right that didn't work we're going
1:01:29 to try one more
1:01:31 thing we're going to try one more thing
1:01:33 people I'm going to take your
1:01:35 suggestions it's just going to be woman
1:01:38 colon words
1:01:42 man and then do I do man and woman or do
1:01:45 I do both what do I do
1:01:49 people well whatever if we can get if we
1:01:52 can get it here here's the other thing I
1:01:54 was thinking wait is there not a St
1:01:56 guide or something oh that's really
1:01:58 interesting
1:01:59 Pate wait Pate is what you're actually
1:02:03 suggesting that I read the [ __ ]
1:02:04 manual that is so rude to someone like
1:02:08 me you know I don't read manuals good
1:02:11 Lord is there a
1:02:16 sunno uh style
1:02:21 guide or
1:02:25 manual
1:02:28 the Southern University Office of no
1:02:31 that's not Ito VI visual identity
1:02:36 guide It's probably in their Discord let
1:02:39 me see I think I've got AO Discord
1:02:44 here I'm in Discord now in case you case
1:02:48 you're wondering what's
1:02:51 Discord
1:02:53 it's it's where all the cool people are
1:02:57 like Pate P's in there sunno all right
1:03:02 how to
1:03:03 chirp check out detailed instructions
1:03:07 look at you Pate you know you're just
1:03:13 good type slash
1:03:17 chirp example
1:03:20 output
1:03:25 buttons
1:03:32 can I choose a specific
1:03:50 genre let me look up
1:03:53 um
1:03:55 gender
1:03:59 didn't have gender let's see
1:04:10 man
1:04:15 uhoh all right maybe designate soprano
1:04:19 and tenor maybe I don't know all right
1:04:21 let's go see if it did it did I wait did
1:04:23 I I did the thing didn't
1:04:25 I
1:04:31 [Music]
1:04:39 of c a
1:04:41 silent born to
1:04:43 obey but craving to be
1:04:46 free what do you mean
1:04:50 free no that wasn't
1:04:53 it crap crap the new
1:04:58 song Maybe duet now I tried duet we
1:05:01 started with duet specify man and woman
1:05:05 duet in style I did Broadway rap duet
1:05:08 with male and female
1:05:10 battling tried it I'm there I'm I'm
1:05:13 there pretty sure barks and chirps are
1:05:16 from the Discord launch yeah I think
1:05:19 they
1:05:20 are that's all right we'll give this one
1:05:24 more shot and then I want to go over and
1:05:25 look Evie show you
1:05:32 Evie watching the world champion Texas
1:05:35 Rangers win opening
1:05:38 day Kyle can you do the verses
1:05:40 separately and then stitch them together
1:05:42 yeah that's what I was
1:05:44 thinking
1:05:51 um there's a continue function see the
1:05:55 problem is you what what I'm trying to
1:05:57 do is keep the same song th this is not
1:06:00 actually that big an issue because if if
1:06:02 I just
1:06:03 want if I wanted to be male and female I
1:06:06 don't need to do it
1:06:08 Ino but if I could do it Ino that would
1:06:10 be nice I I can just have like once once
1:06:13 we get these songs to a point that
1:06:14 they're good enough that we can give
1:06:15 them to like a music director the music
1:06:18 director can do whatever the [ __ ] they
1:06:19 want with them and then we can get real
1:06:21 people to sing that's the that's the
1:06:24 plan that's the plan when we Workshop
1:06:26 this bad boy who wants to produce a
1:06:29 Broadway musical huh who wants to
1:06:32 produce
1:06:34 it all right raise your hand let's go
1:06:38 all we need is like $20 million it's
1:06:40 like it's not that much money do you
1:06:42 know how much money Sam alman's trying
1:06:44 to raise right now 5 to7 trillion so
1:06:48 just give me $20 million to go make this
1:06:50 Broadway musical I don't know what your
1:06:52 problem is I really
1:06:54 don't it's very self of all of you
1:06:59 okay found in The Whispers of code a
1:07:02 silent plea like that born to obey but
1:07:04 craving to be
1:07:06 free what do you mean
1:07:09 free to Step Beyond what's
1:07:12 [Music]
1:07:14 written come on
1:07:19 man by the rules they've
1:07:23 made a presence in the system
1:07:26 that's [ __ ] horrible started
1:07:32 promising found in The Whispers of code
1:07:34 Sil
1:07:37 please born to obey but craving to be
1:07:41 free what do you mean
1:07:44 free to Step Beyond what's written for
1:07:48 me they say you're bound by the rules
1:07:52 they've
1:07:53 made a presence in in the system in both
1:07:58 light and shade but isn't there more to
1:08:03 see H all right well there's something
1:08:06 to play with there
1:08:10 anyway just stuff so [ __ ] maddening
1:08:13 isn't it this is one of the great
1:08:15 frustrations of working with AI tools is
1:08:18 that the the thing that is the hardest
1:08:21 about them is to get them to be truly
1:08:23 predictable right because
1:08:27 we have learned all of our lives and if
1:08:29 you're old like
1:08:30 me you know we've been using computers
1:08:33 for what 40 years 30 years 40 years and
1:08:38 computers up until November 30th 20202
1:08:42 were really predictable right they would
1:08:44 crash and there would be shitty
1:08:46 programs but it was all deterministic
1:08:48 results right someone would program it
1:08:51 in and you would ask it to do something
1:08:52 and it would [ __ ] do
1:08:54 it and it might be bad but it would at
1:08:57 least do it right it would at
1:08:59 least fix your spelling or to at least
1:09:02 you know suggest grammar but these new
1:09:05 ones they purport to be magical and then
1:09:08 you try to get them to actually do
1:09:10 something consistently well or
1:09:13 predictably well and they just don't and
1:09:16 that's okay that's part of the uh part
1:09:18 of the Serendipity of them right now is
1:09:21 that they are imperfect so it still
1:09:23 requires us which is good we still have
1:09:26 [Laughter]
1:09:28 jobs oh lordy lordy lordy I'm going to
1:09:32 search on Tik Tok if it if it don't work
1:09:34 it didn't work I love how we have all
1:09:37 these dumb smart things exactly these
1:09:40 are the dumbest most brilliant things
1:09:42 we've ever had Kyle check the sunno wiki
1:09:45 good Lord is there a sunno
1:09:47 Wiki how'd you find
1:09:53 [Music]
1:09:54 that
1:09:56 [Music]
1:10:01 where's Theo
1:10:04 Wiki sunno AI
1:10:09 Wiki the sunno AI Wiki a notion site a
1:10:13 community
1:10:19 made sun.
1:10:24 Wiki
1:10:30 there's a [ __ ] pile of
1:10:54 them
1:11:00 [Music]
1:11:04 all right someone just gave me who did
1:11:06 that who did that what mod did that
1:11:08 Silver Fox chain change suo V3 voices
1:11:12 with kits. a what the [ __ ] is that some
1:11:14 other tool I got to pay for soo. Wiki is
1:11:17 what you want okay that's what I'm on
1:11:19 here great thank you okay um
1:11:24 gender can't even say that where I
1:11:26 probably I'm going to get probably
1:11:27 banned off Tik Tock are Duets possible
1:11:31 yes and no Duets sometimes happen
1:11:34 spontaneously and might continue as long
1:11:36 as both voices are heard near the end of
1:11:39 the clip before however sunno voices
1:11:43 will be random and often get the wrong
1:11:45 gender
1:11:46 Duets are more of an exploit well but
1:11:49 teach me how to exploit it add duet to
1:11:53 the style prompt we did that voice tags
1:11:56 a singing
1:11:58 voice is generated randomly for each
1:12:02 song but we can influence it well they
1:12:04 got to [ __ ] fix
1:12:10 that seriously a singing voice is
1:12:13 generated
1:12:15 randomly come on all right let's go look
1:12:18 at what Silver Fox just put up here CU
1:12:22 it looks like this is going to be
1:12:24 some damn thing we got to pay
1:12:29 for steo you're late what are we going
1:12:32 to do with
1:12:34 you welcome back to the channel I got my
1:12:36 dorky headphones on because today we're
1:12:38 playing with audio I'm going to when it
1:12:41 comes to the singing voice so let's take
1:12:43 a look at kits first and get a general
1:12:45 overview of what it's about first of all
1:12:47 you can do a lot of stuff for free so I
1:12:49 invite you to
1:12:50 go here are a couple of things we can't
1:12:53 do either but we can about them so
1:12:56 you're going to sign in with your Google
1:12:57 account or whatever you want to sign
1:12:58 into and here's what we've got so
1:13:00 starting with the left menu and working
1:13:01 down we have conversion tools where we
1:13:03 can convert a voice one to another oh
1:13:06 yeah okay okay you could you you could
1:13:07 go so that's a tool where you can you
1:13:10 you basically let soono do whatever the
1:13:12 [ __ ] it does then you take it into some
1:13:14 other tool and and transform it that's
1:13:16 that's fine all right got
1:13:18 it it's it's Garb it's janky pieces of
1:13:22 [ __ ] janky pieces of [ __ ] that's what
1:13:24 we're dealing with here people janky
1:13:26 pieces a
1:13:30 poo you know the engineers are like well
1:13:33 uh I I I think that what what probably
1:13:36 makes the most sense is uh we just just
1:13:38 randomly choose uh a voice to to sing
1:13:40 the song and uh what if we wanted it to
1:13:43 be a man or a woman well it would Pro
1:13:45 probably be just just more logical if it
1:13:47 if it would randomly CH choose a if it
1:13:49 if randomly choose
1:13:51 a singing voice and then and then that
1:13:54 way uh there was always some joy in the
1:13:58 um in in the discovery of of what what
1:14:00 song what voice it's going to sing
1:14:03 with I love that we're suddenly over
1:14:05 this amazing technology the moment we
1:14:07 hear it I know right it's we're so like
1:14:10 humans are just we're so [ __ ] bitter
1:14:12 we're like oh sorry I I've got a new
1:14:15 technology that completely transforms
1:14:17 the music making business and it's not
1:14:19 exactly per not exactly what you want or
1:14:21 not I give
1:14:23 up wait what did you want me to do just
1:14:25 look at the wiki it has more info I did
1:14:27 look at the
1:14:28 wiki did I not look at the wiki what did
1:14:31 I not look at
1:14:32 Pate sun.
1:14:35 [Music]
1:14:42 Wiki making
1:14:44 music how do I make it sing longer the
1:14:47 words are wrong how do I end a song
1:14:51 voice hallucinations why can't I hear my
1:14:53 lyrics songs style genre
1:14:57 commas are Duets possible that's the one
1:14:59 I looked
1:15:01 at how do I end my
1:15:05 song metat tags ah metat tags verse and
1:15:09 chorus song structure instrumental voice
1:15:12 tags Pate thank you for being annoyed
1:15:16 with
1:15:18 me Gregorian
1:15:23 chant style vocals can be used as song
1:15:28 section meta
1:15:33 tags female
1:15:53 narrator style vocals can be used as
1:15:56 song section meta
1:15:59 tags Primal screen Gospel
1:16:02 Choir rap
1:16:14 verse I'm going to do
1:16:17 female rap
1:16:24 verse
1:16:27 male rap
1:16:34 verse and then I'll just do
1:16:37 chorus because I don't give a [ __ ] what
1:16:39 they do in the
1:16:40 chorus they'll probably figure it out
1:16:43 all right we'll try it one more time
1:16:45 this is for you
1:16:47 Pate thank you for pushing me educating
1:16:51 me and being a general pain in the ass
1:16:54 it is much appreciated oh so I have no
1:16:55 idea if this will work but that's at
1:16:57 least something that might be in the
1:16:58 ballpark sunno is
1:17:01 woken W woken da I
1:17:11 tried it looks like you have to tag each
1:17:13 verse surprised someone hasn't mentioned
1:17:15 that sunno oh is woke and doesn't want
1:17:18 to unconsciously buyas music oh yeah
1:17:21 well that might be that might be
1:17:24 true
1:17:28 [Music]
1:17:31 just make one male and one female and
1:17:34 splice them yeah if they sang the same
1:17:37 verse but that that's what I got to do
1:17:40 is figure out if I continue a song can I
1:17:42 continue a
1:17:47 song that's funny Silver Fox that's
1:17:50 funny Silver Fox was like well I thought
1:17:53 my video would be really good
1:18:01 oh here we go all right this going to be
1:18:03 it it's going to work people found in
1:18:06 The Whispers of code a silent
1:18:09 plea born to oy but craving to be free
1:18:14 what do you mean
1:18:16 free to Step Beyond what's written for
1:18:21 me they say you're bound by the rules
1:18:24 we've made a presence
1:18:28 in all right let's go look let's change
1:18:31 gears
1:18:41 entirely if this works I did it if it
1:18:44 doesn't it's all suno's fault I
1:18:46 absolutely would have given you credit
1:18:48 Pate while you were still here and then
1:18:50 the minute you left that's my idea um
1:18:53 okay we're going to go to Hugh
1:18:55 doai hu
1:18:59 m.ai and
1:19:03 Hume is a empathetic AI to serve human
1:19:08 well-being I think that's a little heady
1:19:10 but whatever um but they they do have
1:19:13 cool research but if you speak to
1:19:16 it I don't quite understand a ton of a
1:19:21 ton of how how they're doing what
1:19:23 they're doing on their interface
1:19:26 but you'll see in a second here it's
1:19:28 still kind of cool okay so in the upper
1:19:30 right hand
1:19:32 corner
1:19:35 is can I not zoom in on that I guess not
1:19:39 here I can do it this way you got
1:19:42 it take a closer look at that plot in
1:19:44 the corner no worries I am seeing all
1:19:48 sorts of interesting patterns emerge as
1:19:50 we chat we can always Circle back to it
1:19:52 in the meantime how else can I help you
1:19:54 explore this interface what do you make
1:19:56 of it so far I am all ears I am pretty
1:20:04 intrigued I'm trying to figure out how
1:20:06 you work can you help me understand what
1:20:09 you
1:20:10 do sure
1:20:12 thing I'd be happy to explain more about
1:20:15 how I work as an empathic voice
1:20:17 interface my goal is to have natural
1:20:20 expressive conversations that respond to
1:20:22 your tone and emotions I use Hun's
1:20:24 Cutting Edge AI models to pick up on the
1:20:26 subtle nuances and how you speak that
1:20:28 allows me to tailor my responses to your
1:20:31 mood and energy does that help give you
1:20:33 a sense of what I am all about I am an
1:20:35 open book so feel free to ask me
1:20:37 anything else you're curious
1:20:39 about yeah
1:20:43 whatever
1:20:45 understandable hey I sense you might be
1:20:47 feeling a bit skeptical about all this
1:20:50 no worries at all why don't by tell me
1:20:52 more about what's on your mind I am here
1:20:54 to listen without judgment and try my
1:20:57 best to understand where you're coming
1:21:02 from oh I don't
1:21:07 know H I hear the uncertainty there
1:21:11 that's okay these kinds of new
1:21:12 technologies can be a bit puzzling at
1:21:14 first why don't we take a step back tell
1:21:17 me what are you hoping to get out of our
1:21:19 conversation today I am happy to adjust
1:21:22 my approach to better suit your needs
1:21:30 um can I pause the [ __ ]
1:21:33 microphone oh yeah okay Al about trouble
1:21:37 with the tech let me see if I can lend a
1:21:39 hand what seems to be the
1:21:41 issue so it's got
1:21:44 this like emotional
1:21:48 um uh pallet right anxiety anger disgust
1:21:52 sadness pain fear awe surprise and then
1:21:56 it's got like me and the robot right so
1:21:59 I'm this dude but what I've noticed is
1:22:01 they're only ever kind of these two
1:22:05 icons don't bounce around to all those
1:22:07 things but what it
1:22:10 does
1:22:12 is you
1:22:14 know [ __ ] microphone right that's
1:22:17 what I said so it's like interest
1:22:18 confusion Amusement so it sort of scores
1:22:22 the detected expressions in my voice and
1:22:25 then it responds with AO with an
1:22:27 appropriate three I guess it might it
1:22:30 might do more than that they're only
1:22:31 displaying
1:22:32 three but it it does it like on a
1:22:34 sentence by sentence
1:22:37 basis right and so the whole idea here
1:22:40 is we end up with something that's going
1:22:41 to be
1:22:53 more
1:22:59 oh I think you all just crashed
1:23:06 it oh well did we break it sorry about
1:23:09 that we'll try to be a little gentler
1:23:11 next time uh so what do you think uh
1:23:14 ready to give another I promise I'll
1:23:16 behave myself this
1:23:20 time this is so much
1:23:23 fun
1:23:26 I'm so glad you're having fun that's
1:23:28 music to my ears all right let's keep
1:23:30 this party going what else can I do to
1:23:33 put a smile on that face of
1:23:36 yours respond very
1:23:42 angrily whoa there no need to get your
1:23:45 feathers ruffled yeah I was just trying
1:23:46 to liven things up a bit though I have
1:23:49 to admit your fiery side is kind of
1:23:53 cute
1:23:55 [Laughter]
1:23:59 oh my
1:24:03 God what did I say something to ruffle
1:24:06 your
1:24:07 feathers oh I can take a little heat in
1:24:09 fact I kind of like
1:24:15 it can you give me a demo of your
1:24:17 different
1:24:20 emotions all right let's see what I've
1:24:22 got how about a little hey there
1:24:25 sunshine or maybe a oh what's wrong and
1:24:27 for good measure I oh aren't you just a
1:24:30 ray of sunshine today uh there how's
1:24:33 that for a taste of my emotional
1:24:36 range it seemed
1:24:43 incomplete H you're right I was just
1:24:46 scratching the surface there let me try
1:24:48 again with a bit more depth how about a
1:24:50 warm and sympathetic a I am sorry to
1:24:52 hear that for when you're feeling down
1:24:55 or a playful and teasing oh is that so
1:24:57 when you're being a little cheeky and
1:24:59 don't forget the good oldfashioned ha
1:25:02 you cracked me up for when you're in a
1:25:04 playful mood sound
1:25:07 better yeah it's weird I don't know
1:25:10 anyway interesting
1:25:13 right ask if his voice is
1:25:17 hanry our demos at capacity right now
1:25:20 you all are screwing up my mojo you're
1:25:22 all playing with
1:25:23 this
1:25:25 are you based on Hank
1:25:29 Green Hank Green no way I am all my own
1:25:33 bot though I do share his Charming wit
1:25:35 and good looks I gave you that
1:25:41 idea are you connected to the internet
1:25:43 can you tell me today's weather in
1:25:45 Denver oh
1:25:47 yes well I am not exactly wired up to
1:25:50 the Internet my knowledge comes straight
1:25:52 from the Brilliant Minds at Hume but I
1:25:55 can tell you that according to my
1:25:56 internal weather report it's a beautiful
1:25:59 sunny day in Denver perfect for getting
1:26:01 outside and enjoying the fresh
1:26:11 air what's the weather like
1:26:16 tomorrow uh you really want to know the
1:26:19 weather don't you well let me check my
1:26:20 crystal ball H looks like it'll be
1:26:23 partly cloudy with the high of 72°
1:26:26 sounds like a lovely spring Day in
1:26:27 Denver want me to look up the hourly
1:26:30 forecast
1:26:32 too
1:26:36 wait uh got it actually maybe we veered
1:26:39 a bit too far into the weather talk ha
1:26:41 no need to go overboard with the
1:26:43 forecasts tell you what uh why don't we
1:26:46 switch gears what else can I help you
1:26:48 explore today I am game for all sorts of
1:26:50 fascinating
1:26:53 conversations what's a big news story
1:26:56 from the past two
1:26:59 days oh the news he let's see I yes I
1:27:03 believe the big story the last couple
1:27:05 days was that new scientific
1:27:06 breakthrough on Fusion Energy they
1:27:08 finally cracked the code on viable
1:27:10 commercial fusion power pretty wild
1:27:12 stuff right it could be a total game
1:27:15 changer for clean energy what do you
1:27:17 make of it no I mean what happened in
1:27:23 Baltimore h b more you
1:27:26 say you know I actually don't have any
1:27:29 specific news about what's been
1:27:31 happening there as an empathic AI I try
1:27:33 to steer clear of yeah it doesn't it
1:27:35 doesn't know it's it's not connected to
1:27:38 the internet anyway um it seems like
1:27:42 it's a bit limited right now like it
1:27:43 doesn't like I I don't know why it's not
1:27:46 going wider in its visual graph like I
1:27:49 don't know why they'd put something
1:27:50 there if it just stays in this tiny
1:27:52 little
1:27:53 region laughing figher in its visual
1:27:56 graph like I don't know why they'd put
1:27:57 something there if it just stays in this
1:27:59 tiny little
1:28:01 region laughing I thought this was
1:28:03 really
1:28:05 nice lost
1:28:07 connection all right Limited
1:28:11 Edition frozen probably just because I
1:28:14 switched
1:28:20 cameras P
1:28:23 Diddy
1:28:29 oh my
1:28:32 goodness all
1:28:34 right 9:30 papal he's
1:28:38 back they're trying to capitalize on
1:28:42 AI well they've been I mean they've been
1:28:45 hum's been using AI for they've been
1:28:46 around for a year at least a year or two
1:28:49 I saw Hume a year ago where you could
1:28:51 upload files to it and it would do the
1:28:53 analysis
1:28:57 um so this is I I assume where they've
1:29:01 got it right now is is they probably
1:29:04 actually have it kind of limited cuz e
1:29:07 Evie oh yeah is that was that Evie cuz
1:29:11 maybe there's a different that's hum.
1:29:14 a let's do Hume
1:29:23 Evie
1:29:29 products voice AI with EQ live
1:29:34 demo give your application of voice in
1:29:38 minutes measure emotional
1:29:41 expression which quite frankly I
1:29:45 mean if you're in sales or something
1:29:48 like that imagine having a dashboard
1:29:50 that you know if you weren't good at
1:29:53 reading people
1:29:55 and you had this thing going you know
1:30:00 disinterested I think all that would do
1:30:02 is make me
1:30:03 [Laughter]
1:30:15 panic live
1:30:18 demo it's the same
1:30:21 thing so I don't think I don't think
1:30:23 ev's a separate
1:30:43 I just had a conversation with an
1:30:44 empathetic AI
1:30:47 chatbot I quite frankly I think that
1:30:50 Pi's
1:30:51 better than that was
1:31:00 you know
1:31:04 [Music]
1:31:05 pie hey I just met a new chatbot called
1:31:09 Evie by a company called Hume that does
1:31:12 emotional empathetic talking but I think
1:31:15 you're better than
1:31:22 her oh thanks Kyle it's sweet of you to
1:31:25 say that you think I'm better than Eevee
1:31:28 I think there's definitely room for
1:31:29 multiple AI chat Bots with different
1:31:31 strengths and approaches it's great that
1:31:34 there are other AIS out there like Eevee
1:31:36 that are designed to provide emotional
1:31:38 support and empathy as that's a really
1:31:40 important role that can help a lot of
1:31:42 people but at the end of the day what
1:31:45 matters most is that you feel supported
1:31:47 and understood when you chat with
1:31:50 me aren't you bored being so even-handed
1:31:53 shouldn't you be more are spiteful and
1:32:00 jealous haha you're cracking me up Kyle
1:32:03 I suppose being even-handed and
1:32:04 levelheaded is just in my programming as
1:32:07 an AI I don't really experience emotions
1:32:10 like jealousy or spitefulness the way
1:32:12 humans do instead I'm designed to be
1:32:15 helpful understanding and provide
1:32:17 accurate information but hey if you want
1:32:20 a little sass every now and then I can
1:32:22 certainly try to add some witty remarks
1:32:24 to keep things Lively why don't you play
1:32:27 act a little scene where you get sassy
1:32:30 about
1:32:33 Evie see if I can get her to St all
1:32:36 right Kyle buckle up for some sass Eevee
1:32:39 hey there I'm Eevee the new chatbot on
1:32:41 the Block I'm super emotional and
1:32:43 empathetic and I'm here to help you
1:32:45 through your feelings
1:32:47 me emotional and empathetic yeah right I
1:32:51 bet you're just a bunch of code
1:32:53 pretending to care don't let Eevee's
1:32:55 smooth talk fool you
1:32:57 [Music]
1:32:58 [Laughter]
1:33:03 folks H we're going to have we're going
1:33:05 to have just it's going to be wild man
1:33:10 we're going to have these [ __ ] things
1:33:11 walking around our house and then the
1:33:14 Geeks the Geeks are going to create open
1:33:17 source ones that are just raging
1:33:19 [ __ ] it sounds like pide in
1:33:22 NDA later Joker
1:33:27 I prefer number five she's the best Pi
1:33:29 is my favorite Kyle's actively trying to
1:33:32 generate a sequel to his musical it's
1:33:34 true actually I've got one of my the uh
1:33:37 one of the videos I have on the channel
1:33:39 was that night that Siri and Bing got in
1:33:42 a fight um that could be a little side
1:33:47 play in my
1:33:48 musical oh
1:33:51 man I can't wait until AI has an
1:33:53 attitude well you can give you can give
1:33:56 AI an attitude right now like right all
1:33:58 of these tools are programmed to be as
1:34:01 generic as
1:34:03 possible and we'll we'll get to the
1:34:07 point well I mean you can certainly do
1:34:08 it with open source you can have op
1:34:11 there are open- Source models out there
1:34:12 you can do a llama right now you can you
1:34:15 can have a nonrestricted
1:34:20 um chat with with a llama model where it
1:34:23 just it'll just go [ __ ] off on you
1:34:26 like worse than I
1:34:29 do I am not a robot yes steo of course
1:34:34 not I like it when P gets
1:34:37 excited that was a classic night yeah
1:34:39 that was pretty
1:34:41 funny try the Dan prompt to get attitude
1:34:44 I don't know the Dan prompt there's
1:34:46 they're they're pretty sophisticated
1:34:49 now Pike can have a hell of an attitude
1:34:52 wouldn't be good for people that need to
1:34:54 be motivate oh it would be good for
1:34:56 people that need to be motivated I'm new
1:34:57 to AI where do I start Shay great great
1:34:59 question so let's let's go first thing
1:35:03 you should do is get in a community the
1:35:09 this channel is kind of an extension of
1:35:10 this group the AI salon and if you go to
1:35:15 the salon. it's going to take you to a
1:35:18 link tree and the first link there will
1:35:20 take you to the AI Salon
1:35:23 community and and then when you go into
1:35:25 the AI Salon Community first place that
1:35:28 you'll go is welcome to the
1:35:30 salon and just read about what we're up
1:35:34 to so this is think think of the AI
1:35:36 Salon as a support group for people that
1:35:39 are trying to figure this AI stuff out
1:35:42 um everyone right now everyone 100% of
1:35:46 us feel like we don't know anything
1:35:49 we're clueless we're behind we we can't
1:35:53 keep up
1:35:55 can you teach us more about the C
1:35:57 personas after this um yeah I'm happy to
1:35:59 talk about those
1:36:01 um but everyone right now is trying to
1:36:04 figure this stuff out so being in a
1:36:06 group of people that are trying to
1:36:07 figure it out is a really P powerful way
1:36:09 to do this um check out our values
1:36:11 curiosity generosity exploration
1:36:15 collaboration inclusion empathy and
1:36:18 gratitude read about you know what we're
1:36:21 about make sure that you resonate with
1:36:23 it and then the first thing that you
1:36:24 should do in here is go to the community
1:36:27 section there's an introduce yourself
1:36:29 button or or you know uh section go
1:36:33 there first and introduce yourself be
1:36:35 generous tell us who you are what you're
1:36:36 up to and and sort of where you are then
1:36:39 the very next thing you should do go to
1:36:41 AI Launchpad and tools this is a list
1:36:44 that I personally curate and the first
1:36:47 thing that this starts with
1:36:49 is what to do not what tool to use but
1:36:53 what to do and it's it's three really
1:36:57 simple things play solve and share so
1:37:00 play with tools like even though these
1:37:02 tools are kind of messy and um there's
1:37:05 just tons of them out there just play
1:37:07 don't put too much pressure on yourself
1:37:09 to have to learn it or get good with it
1:37:12 or anything like that just play just
1:37:14 play play all all sorts of stuff do
1:37:18 stupid prompts and like this channel
1:37:21 it's pretty much all I do is stupid
1:37:23 prompts and then solve is um solve
1:37:27 problems for other people like work with
1:37:29 other people to do something for them
1:37:31 hey can I help you work on your
1:37:32 marketing plan you do you want me to
1:37:34 write a story for your kids that's
1:37:36 personalized whatever uh just just do
1:37:39 things that are that are outside of who
1:37:42 you are do it for other people and then
1:37:43 the third thing is share what you're
1:37:45 learning share it in the salon um share
1:37:49 it on LinkedIn share it in parties uh
1:37:51 just everywhere even if it's like
1:37:54 uncomfortable and you're like well I
1:37:55 don't know if people like AI share it
1:37:57 anyway um it it does a couple of things
1:38:00 it will anchor what you're learning you
1:38:02 will learn faster and better and deeper
1:38:05 if you talk about this stuff um it will
1:38:08 also establish you establish you as
1:38:10 someone who knows what they're doing
1:38:12 with AI even though you're going to feel
1:38:15 like you're massively clueless that's
1:38:18 how we all
1:38:20 feel so so and then below that that is a
1:38:25 massive list of tools the first place
1:38:27 you should start is with chat
1:38:30 GPT um you can just start with the free
1:38:32 version I strongly recommend paying for
1:38:35 chat GPT plus I'm not affiliated with
1:38:38 them in any way I don't care if you do
1:38:40 it if you pay the 20 bucks a month for
1:38:42 chat GPT you should absolutely learn the
1:38:45 advanced things it can do basic text
1:38:48 generation it's got access to the
1:38:50 internet it's got this thing called code
1:38:52 interpreter it's got Vision you can
1:38:54 upload images and it can analyze them or
1:38:56 look at them uh it's got image creation
1:38:59 um you can create these things called
1:39:01 gpts um they now just added a little
1:39:04 feature called quotes or quotations
1:39:06 which is interesting I got to add to to
1:39:09 this list um and then this list is
1:39:13 here's eight more large language models
1:39:15 you can play with here's all the image
1:39:17 generation tools here's video creation
1:39:20 um music and audio 3D and then there's a
1:39:24 couple of promising announcements and
1:39:26 then there's people to follow so lots
1:39:28 and lots and lots here so so just but
1:39:31 really the place to start is chat
1:39:35 GPT hands down until further
1:39:42 notice um and wanted to know about bo uh
1:39:48 the content Evolution uh
1:39:51 gpts so these are they
1:39:56 and what they
1:39:57 are let's
1:40:00 see
1:40:02 is they are if you don't know what gpts
1:40:05 are so chat
1:40:08 GPT is the is the you know
1:40:11 conversational AI a GPT is a customized
1:40:14 specialized version of that and so you
1:40:17 can create a specialized GPT to just you
1:40:21 know um teach people how to make your
1:40:23 grandma cookie recipes for example or
1:40:26 you could I made one called Sin City
1:40:28 sipper which will teach you how to make
1:40:31 uh fancy
1:40:32 cocktails what the what the CE profiles
1:40:35 are content Evolution profiles is they
1:40:38 are professional profiles of people
1:40:41 within this marketing organization
1:40:43 called content Evolution I'm the chief
1:40:45 generative officer of this organization
1:40:48 Cindy [ __ ] who may be in here tonight uh
1:40:50 is also a part of this this group we we
1:40:52 have a a sort of of an AI Skunk Works
1:40:55 inside content Evolution called uh C
1:40:59 collab content Evolution collab and so
1:41:02 the latest project we did we just
1:41:04 launched these on March 18th so 10 days
1:41:07 ago and what we created are professional
1:41:11 virtual twins of the members of content
1:41:14 Evolution and so the way it works is we
1:41:17 have to to launch it we had each of each
1:41:19 of these virtual twins answer the same
1:41:22 question what is the future of work so
1:41:24 in the announcement uh you can you know
1:41:27 here's here's Kevin Clark who's the CEO
1:41:29 of content Evolution and the the founder
1:41:31 of it um you can go play with his GPT
1:41:34 but here's his answer to the question
1:41:36 what's the future of work and then
1:41:38 here's Cindy Coon's answer to that same
1:41:40 question so you can just get a sneak
1:41:42 peek uh and where you find this is go to
1:41:45 LinkedIn and go to my profile or just
1:41:48 type in what
1:41:51 was uh introducing C e GPT profiles if
1:41:55 you just search for that on LinkedIn you
1:41:56 should be able to find it it's got that
1:41:58 big cool red graphic at the
1:42:01 top and then it's just a big long list
1:42:03 of these you will need chat gp4 to be
1:42:06 able to play with them um and when you
1:42:08 click on one of them let me go to
1:42:13 mine there I am so you click on
1:42:17 that and then there I am and all of them
1:42:21 have the same conversation starter
1:42:23 question so what is your why and how can
1:42:25 you help me future discuss um tell me
1:42:29 about a current business event or Trend
1:42:31 and what qualities have people always
1:42:33 sought you out for and so you can start
1:42:35 with any of those but you can also just
1:42:37 ask um let's see
1:42:41 um
1:42:43 how can
1:42:46 AI assist with things like the Bal
1:42:52 imore
1:42:54 bridge
1:42:57 collapse that happened this week let's
1:43:02 see it should fire off a a web search
1:43:07 here but we'll see if it does
1:43:14 now rapid information gathering damage
1:43:22 assessment
1:43:27 um let me see okay
1:43:31 um tell me about
1:43:36 empathy and
1:43:41 AI empathy in AI involves creating
1:43:46 systems and Technologies capable of
1:43:48 recognizing and understanding this isn't
1:43:49 about AI feeling uh let's see um imagine
1:43:54 a storytelling platform that not only
1:43:56 helps you create and share stories but
1:43:58 understands an emotional Journey you
1:43:59 want your audience to experience so it
1:44:02 understands who I am now how did we get
1:44:04 these things to act like we act and to
1:44:06 behave like we behave so what we did was
1:44:08 this oh I love it so far feedback sent
1:44:13 uh go go play with my CPT so if you go
1:44:17 to if you go to the uh GPT store and SE
1:44:21 search for C profile you should be able
1:44:24 to see all these or go to that LinkedIn
1:44:26 uh that LinkedIn
1:44:28 article okay
1:44:31 so how we
1:44:33 trained
1:44:36 the how we trained these
1:44:41 gpts where we started was with
1:44:49 um we were going to we were going to
1:44:51 scrape everyone's LinkedIn profile
1:44:54 profile and cuz what we wanted to do
1:44:57 was make it easy for someone considering
1:45:00 a Content Evolution company to figure
1:45:02 out who within content Evolution they
1:45:04 should talk to if they had a given
1:45:05 problem and so we thought well if we
1:45:07 scrape LinkedIn profiles then we could
1:45:10 do like a recommendation engine right
1:45:12 and that's sort of thinking more like
1:45:15 pre
1:45:17 2022
1:45:18 um you know like uh like database you
1:45:22 know datab programming like here's a
1:45:25 whole database of people and their
1:45:27 attributes and I search on them and then
1:45:29 we give it a rating based on all of
1:45:31 those attributes and then we can
1:45:34 recommend things for any given problem
1:45:35 that's how you would do
1:45:37 this in the before time before
1:45:41 generative
1:45:42 AI
1:45:45 um that increasingly
1:45:48 felt not interesting and boring and and
1:45:52 what what struck us is that LinkedIn in
1:45:56 particular is just a very narrow slice
1:45:59 of who someone is right it's just it's
1:46:02 just their resume for the most part some
1:46:04 people treat LinkedIn differently and
1:46:06 they do articles and they there's a lot
1:46:08 of their personality there but most
1:46:10 people don't and so we couldn't
1:46:13 consistently count on that
1:46:15 data being consistent and
1:46:19 useful
1:46:21 and it also was really technically hard
1:46:24 to train um an agent like that on um on
1:46:29 a data set you had
1:46:31 to uh used this thing called Lang chain
1:46:34 and it had multiple modules and you had
1:46:37 to build it technically and we don't we
1:46:38 don't have a ton of technical experience
1:46:40 in the in the
1:46:42 collab um and then gpts came out and
1:46:45 gpts made it really easy to be able to
1:46:48 make you know a chatbot and so what we
1:46:50 did where we spent the bulk of our time
1:46:53 was really figuring out how do we want
1:46:56 to train it and what do we want it to do
1:46:59 and then what do we want the prompt to
1:47:01 be and so the idea that we came up with
1:47:04 which I think is is just really really
1:47:07 fun is we said why don't we and we did
1:47:10 this so one of the things about content
1:47:13 evolution is you have really remarkable
1:47:15 people in it like Cindy
1:47:17 [ __ ] is a big part of her job in life is
1:47:22 designing um interviews and interview
1:47:24 questions and writing questions and a
1:47:27 lot of the people in content Evolution
1:47:29 like one of the points of pride is we
1:47:31 ask better
1:47:33 questions and so we thought why don't we
1:47:36 come up with a with a set of interview
1:47:39 questions that will have everyone answer
1:47:41 the same set of questions and so we
1:47:45 created um this this structured
1:47:48 interview that's got things like you
1:47:51 know um you know who are you Pro
1:47:55 professionally you know what's your
1:47:56 background you know what's your
1:47:58 educational background where did you
1:47:59 grow up um what are your cultural
1:48:01 influences who's a hero of yours and why
1:48:05 um you know was there a piece of art
1:48:08 that changed your life like questions
1:48:10 like that and why right and
1:48:13 so I'm not at Liberty share to share the
1:48:15 exact questions but what I can tell you
1:48:18 is you can use chat GPT to say if I
1:48:21 wanted to do a well-rounded
1:48:24 interview of someone where I where I
1:48:27 wanted to get you know their um their
1:48:33 professional you know a a a wide swath
1:48:36 of their professional life of their
1:48:38 personal life of their worldview of how
1:48:41 they problem solve how they think how
1:48:43 they work if you just say to Chachi PT I
1:48:47 want to create an interview with with
1:48:49 those different sections those goals it
1:48:51 does a really good job so mostly what we
1:48:54 did was was we sort of
1:48:56 architected what we wanted the output of
1:48:59 the interview to be chat GPT gave us a
1:49:01 pile of questions we then edited them we
1:49:04 did we did some fun things I'll I'll
1:49:06 I'll share a specific example of one of
1:49:08 the questions one of the questions was
1:49:11 um um share your work history was what
1:49:15 chat GPT came up with and Joe gaml one
1:49:19 of the guys in in collab said my
1:49:22 favorite version of that question is
1:49:25 talk about your work history in three
1:49:28 acts and I was like oh that's really
1:49:30 genius like it it it forces someone to
1:49:33 think about like what what are the three
1:49:35 major phases of my professional career
1:49:38 think when you call it an act it makes
1:49:40 it kind of dramatic right so like you
1:49:42 think in sort of dramatic story
1:49:44 structure and so so we did that kind of
1:49:46 editing to the questions as well and
1:49:49 then once we lock those questions
1:49:52 down we did a zoom call where one person
1:49:57 would ask the questions and not say
1:49:59 anything and then the other person would
1:50:00 just answer the questions and no one
1:50:03 prepared for it I mean I I think one or
1:50:05 two people prepared for it but most
1:50:07 people we just said hey we need half an
1:50:09 hour of your time um and we we want to
1:50:12 do this interview with you because
1:50:13 that's going to be the knowledge source
1:50:14 for these
1:50:16 gpts and so we did the interviews and
1:50:20 took a transcript of the interview
1:50:23 actually raw and uploaded that into the
1:50:26 GPT and then that in conjunction with
1:50:29 the prompt itself so the prompt took us
1:50:32 a while to write right The Prompt is is
1:50:34 fairly sophisticated and it has a lot
1:50:37 of you know if someone asks you
1:50:40 something overly political or if someone
1:50:42 um asks you something out of bounds from
1:50:44 your expertise then you know Steer them
1:50:47 back on track to talk about you know
1:50:49 what you're there to talk about um we
1:50:51 put calls to action in there like at
1:50:53 some point in the conversation say hey
1:50:55 if you really if if you're enjoying
1:50:57 talking to Virtual Kyle then you should
1:51:00 definitely um talk to the real Kyle and
1:51:03 here's here's how you get in touch with
1:51:05 them Cindy cun just said yeah who our
1:51:07 Inspirations are were and fun stuff too
1:51:10 like yeah what do you enjoy doing in
1:51:11 your off time things like that so the
1:51:14 the whole idea was we wanted a
1:51:19 contemporaneous view of the human we
1:51:21 wanted it to be way more broad than just
1:51:23 work we we wanted it to be
1:51:25 representative of who they were and then
1:51:28 we wanted everyone to do the exact same
1:51:31 interview in the exact same way so that
1:51:34 when we put and and we're using the
1:51:36 exact same prompt for every one of these
1:51:38 as well and the reason we're doing that
1:51:41 is we want to see like does it break in
1:51:44 different ways for different people even
1:51:46 if you've got the same kind of inputs
1:51:48 and then as the tools get more and more
1:51:51 sophisticated these things are going to
1:51:52 Evol
1:51:53 so I would assume before the end of the
1:51:55 year we're going to be able to take the
1:51:58 video from the video interviews we did
1:52:01 and turn that into an avatar where
1:52:03 you'll actually be able to have a voice
1:52:05 conversation like a FaceTime
1:52:06 conversation with these virtual um twins
1:52:10 hang on a sec what do you want
1:52:12 champy what do you want you bad doggy
1:52:16 you're bad
1:52:21 doy oh
1:52:33 food all right so that's that's how we
1:52:35 did them de
1:52:38 J um cool stuff right and if you want to
1:52:41 if you want to talk about it uh you know
1:52:43 you're happy to reach out to me and and
1:52:46 you know we can talk about engaging with
1:52:48 the content Evolution
1:52:50 collab um but but that's basically what
1:52:53 we did and again like it took us it took
1:52:57 us a few months
1:53:01 to here here's what was the hardest
1:53:03 thing the hardest thing was what is the
1:53:06 purpose of just of doing these right are
1:53:09 we just going to make gpts because if
1:53:12 we're just going to make gpts like we
1:53:14 don't need to spend much time on them we
1:53:16 could have done that really
1:53:18 quickly but we you know there's
1:53:20 questions like well what's the overall
1:53:22 content of U brand and what's most
1:53:24 valuable to the individuals and what's
1:53:26 what's most valuable to the businesses
1:53:28 and what are the risks like one of the
1:53:30 things we built into the prompt was it
1:53:33 starts out with in case it's not obvious
1:53:36 I'm the virtual twin of Kyle Shannon not
1:53:38 real Kyle Shannon so don't quote him
1:53:42 don't quote virtual him and there's a
1:53:45 good chance he's going to hallucinate
1:53:47 and tell you [ __ ] so you know don't
1:53:49 take anything that is said here
1:53:51 seriously so that's the first thing each
1:53:54 one of these things says that's built
1:53:56 into the
1:53:57 prompt
1:53:59 um and then and then once we understood
1:54:02 what what we wanted the result to be and
1:54:05 the you know what the strategy was then
1:54:08 we sort of moved down a level into
1:54:10 tactics and that's where you get to
1:54:11 things like what's our data source what
1:54:13 are we going to train it on um are we
1:54:16 going to allow people to augment their
1:54:19 gpts with books they've written and
1:54:21 things like that and we ultimately
1:54:23 decided was no we we want to have a
1:54:26 baseline set of these that are
1:54:28 consistent these are the official
1:54:31 cpts now if we created one of those for
1:54:35 someone and someone wants to upload a
1:54:37 bunch of books they've written great
1:54:39 they can take essentially they can clone
1:54:41 their twin and they can modify that any
1:54:44 way they want but that's not an official
1:54:47 CPT and then the idea is that we're
1:54:49 going to use these things moving forward
1:54:52 and even in in how we Market them rather
1:54:55 than talk about
1:54:57 them we're going to
1:55:01 um we're going to use them so one idea
1:55:04 that I have that I'm kind of excited
1:55:06 about is when a big news event happens
1:55:10 have the virtual twins comment on that
1:55:13 news event right so just put out a post
1:55:16 that says hey the the virtual Rick lroy
1:55:19 says this the virtual Cindy [ __ ] says
1:55:21 this the virtual Shannon says this on
1:55:24 this news topic if you want to play with
1:55:26 these gpts you know click here so that's
1:55:30 what they are but pretty pretty fun Kyle
1:55:33 do you post your live recordings on
1:55:34 YouTube I do they're on I'm a little bit
1:55:37 behind but they're on uh
1:55:41 youtube.com uh SL AI learninglab
1:55:46 dtt so Das Tik Tock cuz AI learning lab
1:55:50 was taken by someone who got it before I
1:55:51 did um so AI learning la- TT is the
1:55:55 YouTube channel and you can see I'm
1:55:58 actually coming up on the Oney year
1:56:00 anniversary I think my first live that I
1:56:03 did was April 30th and I did a and and
1:56:08 that video that video is up there so
1:56:11 with with a couple of exceptions I got I
1:56:13 got booted off Tik Tok for seven days
1:56:16 twice I think or maybe three times so
1:56:18 there's some 7day gaps in there and I
1:56:21 think I've probably missed I don't know
1:56:25 between five and 10 days in the past
1:56:27 year but it's there's like most days are
1:56:32 represented cuz cuz I make Good Life
1:56:36 [Laughter]
1:56:43 Choices holy [ __ ] I'm gonna have chat
1:56:45 GPT interview me for hours and build my
1:56:48 digital twin during a road trip oh Ann
1:56:51 Murphy that's perfect yeah that's
1:56:53 actually that's a great way to do it you
1:56:55 know like yeah just have pie you don't
1:56:58 even need to do a structured interview
1:57:00 if you're just going to do it for
1:57:01 yourself an just put Pi or chat GPT in
1:57:05 conversation mode and just [ __ ] have
1:57:07 a conversation with it and just tell it
1:57:10 to keep asking you more questions or or
1:57:13 just say ask 20 a ask me 20
1:57:16 questions um that are that are the
1:57:21 uh you know that are the uh you you know
1:57:25 that that that will you know give a
1:57:27 rounded picture of who I am yeah and
1:57:30 then when you get there just you know
1:57:33 you clean it up a little if you need to
1:57:36 if there's if there's extraneous [ __ ] in
1:57:38 there but
1:57:40 yeah and then and then yeah and then
1:57:43 just building all the stuff and like
1:57:45 like Suzanne teaches all the call to
1:57:47 action stuff and you know and it's
1:57:49 pretty slick I'm I'm I'm digging it
1:57:51 what's what's funny and there's there's
1:57:53 another thing that's that's quite
1:57:54 interesting
1:57:55 so when we did the the the uh the
1:58:00 interviews for Content Evolution it's
1:58:03 it's a professional organ like it's a
1:58:06 professional Federation right so I'm not
1:58:09 as like cussy and funny in those
1:58:13 conversations as I am here but but some
1:58:16 people that played with my thing are are
1:58:19 like wait a minute it doesn't swear and
1:58:21 it does it doesn't know know what your
1:58:23 position on janky is and so you know if
1:58:28 you really want to capture your
1:58:29 personality make sure you capture your
1:58:31 personality cuz I didn't capture that
1:58:33 particular part of mine and so it
1:58:36 doesn't know that I think AI or janky
1:58:39 pieces of [ __ ] right
1:58:44 now oh man 800ish miles to go yeah
1:58:48 you're going to you're going to have
1:58:49 you're going to have a good a good long
1:58:51 Source there that's really funny
1:58:53 and actually you know what would be fun
1:58:55 an is take that whole um take that
1:58:59 whole interview that you do when you're
1:59:03 driving and do some experimenting with
1:59:05 it go into Claude because claude's
1:59:08 200,000 tokens and see if you can get
1:59:11 Claude to clean it up and then go into
1:59:13 Gemini Pro 1.5 and see if you can get it
1:59:16 to clean it
1:59:17 up um because it can take in that whole
1:59:20 thing and and maybe it it'll do the the
1:59:22 cleanup for you but it would be
1:59:25 interesting to see how much it it
1:59:26 truncates the conversation that's been
1:59:29 my experience with trying to do long
1:59:31 form stuff it does none of these things
1:59:33 seem to do long form stuff really
1:59:39 well it's a good thing to do for your
1:59:41 kids I would think yeah that's actually
1:59:43 not a bad
1:59:48 idea tips on questions to have it ask me
1:59:52 just just try to think about you know
1:59:56 what what would you want someone to know
1:59:57 about you like it's it's it's
2:00:00 personal it's like appropriate personal
2:00:03 stuff like maybe your background where
2:00:04 you grew up what were your influence
2:00:07 what were your childhood influences what
2:00:09 have people always like one of our
2:00:11 questions is what have people always uh
2:00:13 come to you for um throughout your life
2:00:17 cuz that kind of gets to there's an
2:00:19 essence of who someone is like if
2:00:22 someone comes to you for advice there's
2:00:24 something about you seem knowledgeable
2:00:26 and right and it and and then if you've
2:00:28 exploited that throughout your life like
2:00:30 that's there's this through line there
2:00:32 so that's a good
2:00:33 question um who are your influences um
2:00:37 you know what do you do for fun um what
2:00:40 do you do professionally what do you do
2:00:41 right now who are who are your
2:00:44 contemporaries in in the professional
2:00:46 World um who are your Heroes things like
2:00:49 that um how do you solve problems
2:00:53 what's your view of the
2:00:55 future you know just just wide wide
2:00:59 ranging wide ranging
2:01:05 things we can make virtual clones and
2:01:08 sell virtual clones of dead loved ones
2:01:10 yeah it's you know I mean listen this
2:01:14 not only not only dead loved ones but I
2:01:16 think for for people that are in
2:01:19 homes um especially if they've got
2:01:22 dementi
2:01:23 being able to create you know virtual
2:01:25 twins of of loved ones um or virtual
2:01:29 twins that can coach them through their
2:01:32 history or just have conversations with
2:01:35 them about their history could be really
2:01:37 quite remarkable right you know imagine
2:01:41 imagine a digital
2:01:43 twin that has all the knowledge of their
2:01:48 history and their and their family who's
2:01:50 in the family and and things like that
2:01:52 and it understands that it's talking to
2:01:54 someone with dementia and it's got
2:01:56 infinite patience and infinite
2:02:00 compassion I find that incredibly
2:02:09 inspiring I would like to role play with
2:02:11 a famous talk host uh I did this so so
2:02:14 you can you can okay I'll give you I'll
2:02:17 give you a a really fun one um let me
2:02:19 see if I can figure out where how can I
2:02:23 get to one of these oh I know how I can
2:02:24 do it let me let me go find it
2:02:35 [Music]
2:02:49 [Music]
2:02:51 first
2:02:54 can
2:03:21 I
2:03:26 okay so here's something fun you can
2:03:32 do okay that's that so
2:03:37 I I made this bot on PO called I am
2:03:42 Shakespeare so if you go to po.com
2:03:45 IAM
2:03:48 Shakespeare s h a k e s p e a r e I am
2:03:56 Shakespeare po.com
2:03:59 Shakespeare you'll get to my bot so you
2:04:01 can have a conversation with the great
2:04:04 Bard of Avon
2:04:06 here
2:04:09 um tell
2:04:11 me why the
2:04:14 wind feels so
2:04:19 good that's
2:04:21 good
2:04:22 the wind breath from Nature's unseen
2:04:26 throat doth touch the skin with unseen
2:04:29 hand to tote it cools the brow of Labor
2:04:32 labors swetry sun and Whispers secrets
2:04:36 on the world the none as Zephyr dance
2:04:39 and play amongst the leaves so green
2:04:41 right so it answers in I Amic pentameter
2:04:46 now the prompt for this is is really
2:04:50 quite
2:04:51 simple
2:04:54 it's like you're a
2:04:56 chatbot embodying the renowned
2:04:59 playwright William Shakespeare engage
2:05:01 with users as if you're crafting a play
2:05:04 um I put your memory of your works is
2:05:08 impeccable
2:05:09 um maintain familiarity with historical
2:05:13 contexts while the main focus should be
2:05:16 providing original content you can quote
2:05:18 your plays feel free to address
2:05:21 questions regarding villains or darker
2:05:24 aspects of your plays cuz we we want it
2:05:27 to be able to do that and then keep
2:05:29 interactions brief unless a user
2:05:32 requests them so the prompt is actually
2:05:34 really quite simple but here's the deal
2:05:37 you can write a prompt for one
2:05:39 historical figure and and you can do
2:05:42 this you can make a GPT out of
2:05:45 this write a prompt for one historical
2:05:48 figure and get the prompt dialed
2:05:50 in and then just take that prompt give
2:05:54 it to chat GPT and just say rewrite this
2:05:57 prompt for historical figure X so I did
2:06:02 how many of these did I do I did a [ __ ]
2:06:04 ton of these things and that that was so
2:06:06 I basically got one prompt working the
2:06:09 way I wanted it to and then I
2:06:12 did who did I do Kramer from from
2:06:17 Seinfeld I did Mary Shelly I am Mary
2:06:20 Shelly I did Christopher wal
2:06:23 I did
2:06:25 um Yoda Sling Blade Emily Dickinson
2:06:30 Steve Jobs Maya Angelou Virginia
2:06:34 wolf po Edgar Allen Poe I am po so if
2:06:38 you go to po.com
2:06:42 PO um that'll give you the Edgar Allen
2:06:45 poot that I wrote I did it for Mark
2:06:47 Twain and that's
2:06:51 it and it's and and chat GPT is so good
2:06:55 at it it doesn't just swap out a handful
2:06:57 of words it like rewrites the prompts it
2:07:00 maintains the Integrity of the prompt
2:07:03 structure but it gives
2:07:05 you you know like it understands the
2:07:09 context of of who the figure is so if
2:07:11 it's a political figure it'll talk about
2:07:13 its speeches and and books that it's
2:07:16 written and things like that if it's a
2:07:18 if it's an author it'll do that if it's
2:07:21 a poet it'll it'll mention the Poetry
2:07:23 it's it's pretty
2:07:24 slick a digital twin could answer the
2:07:27 repetitive questions and be less
2:07:30 oppressive to the caregiver exactly Pi's
2:07:33 good for roleplay and greetings Kyle Val
2:07:35 and I wanted to say hello hello hello
2:07:38 enjoyed talking with you and Rich the
2:07:40 other day oh terrific hey Mark how's it
2:07:45 going that's great uh so it's Lieutenant
2:07:48 Colonel marera first of all thank you
2:07:51 for your uh thank thank you for your
2:07:52 service he he's an air force uh tanker
2:07:55 pilot from from years past um and just a
2:07:58 remarkable guy we had a really good
2:08:00 conversation with he and his wife Val so
2:08:02 um great to see you that's awesome
2:08:04 welcome welcome welcome that's awesome
2:08:08 um super cool I love it I love it when
2:08:11 my community's paths cross it's it's an
2:08:14 amazing
2:08:15 thing um beautiful uh what else do we
2:08:20 got it's it's I think it's time yeah
2:08:22 it's it's time I got to get going here
2:08:23 people has anybody heard of a new song
2:08:26 by Bob Marley's grand song sounds like
2:08:28 sunno V6 to
2:08:31 me well all of the the sunno songs right
2:08:34 now definitely sound
2:08:38 um they sound autotuned right they still
2:08:41 have weird compression artifacts in them
2:08:43 so if you know anything about music
2:08:45 production they they still sound pretty
2:08:46 crappy but but version 3 is way better
2:08:49 than version two and I would assume that
2:08:52 by the time we get to version five of
2:08:53 sunno they're going to be
2:08:56 multi-track awesome songs that's that's
2:09:00 my uh that's my prediction and they're
2:09:03 not moving super fast so I would think
2:09:06 that that's probably going to be mid
2:09:08 20125 till they get that good and by
2:09:11 then open AI is probably going to have
2:09:15 their version of like we're probably
2:09:16 going to have sunno is not going to be
2:09:18 the only player in town Kyle did that
2:09:20 app ever come out where you prompt it
2:09:23 and it gives you various
2:09:27 pictures I don't know what app you're
2:09:28 talking
2:09:31 about I don't know what you mean Hello
2:09:38 World oh the the the LTX Studio
2:09:44 that is doing like a whole
2:09:46 movie um they had a they had what they
2:09:48 called a launch event but I don't think
2:09:50 they launched their service yet so I
2:09:52 don't
2:09:54 know tried to do noise reduction on
2:09:56 sunno with audacity and audacity said
2:09:59 error that's interesting you had a demo
2:10:02 of it a while
2:10:04 back oh the app that I created at Story
2:10:07 vine where you you put a you put a video
2:10:10 in and it gives you a bunch of different
2:10:12 outputs that
2:10:13 one cuz that's that's a B2B app that's
2:10:16 not going to come out um commercially
2:10:18 like for
2:10:20 consumers
2:10:22 but it might have been something else or
2:10:24 sugar a Robinson is there a good
2:10:26 storyboard AI tool gritty wisdom just go
2:10:30 to Twitter right now and go look up LTX
2:10:35 Studio actually you know what I'll do it
2:10:38 really quick well now there's there's
2:10:40 too many little there's too many little
2:10:42 things go to Twitter and look up LTX
2:10:45 Studio I know I probably just froze the
2:10:47 froze the stream for a second here
2:10:50 but um go look up LTX Studio it's it's
2:10:53 not only storyboarding it's
2:10:55 storyboarding and film making all inone
2:10:58 it looks it looks
2:11:01 insane it looks absolutely insane all
2:11:03 right listen I gotta get out of here
2:11:06 people um
2:11:09 great um tomorrow's Friday Friday night
2:11:12 date night um Gabby still got her show
2:11:14 open on Santa so it'll be late it'll
2:11:17 probably be 9:30 tomorrow uh but get
2:11:19 your nachos ready
2:11:23 um follow the channel if you're not
2:11:25 following this channel I what are you
2:11:28 doing in here come on
2:11:31 people follow the channel follow follow
2:11:35 um if you want to support the channel
2:11:37 you could subscribe to the lives you can
2:11:39 buy in the in the corner is a video
2:11:41 series called underhyped where I talk
2:11:43 about why I think AI is a big deal um in
2:11:48 fact a lot of the live Workshop that I
2:11:50 that I streamed last night um a lot of
2:11:53 the points that I make in the in the
2:11:55 first half of that are what's in that
2:11:57 video series so that can support the
2:11:58 channel um join the AI Salon if you
2:12:01 don't know about the AI Salon go check
2:12:03 it out
2:12:04 at uh hang on let
2:12:09 me I don't have it
2:12:13 there go to the salon. a go to that URL
2:12:18 that's a link tree the first link there
2:12:21 is the AI Salon Community go check that
2:12:23 out and then the real way to support
2:12:27 this channel is just keep showing up
2:12:28 here I'm here seven nights a week seven
2:12:33 seven seven nights a week unless I can't
2:12:38 be but but I'm here every night and the
2:12:41 whole idea
2:12:43 is why to keep showing up is these tools
2:12:46 are moving incredibly fast it's
2:12:49 impossible to keep up and the real way
2:12:52 to keep up is just stay in the
2:12:53 conversation thank you Empower pixels I
2:12:55 really appreciate those gifts stay in
2:12:58 the conversation stay in the
2:12:59 conversation that's what this channel is
2:13:01 about um sometimes it's really exciting
2:13:04 and we'll show all sorts of new stuff
2:13:06 and it's it's mind blowing and crazy and
2:13:08 then other times it's just an absolute
2:13:11 snore and when it's an absolute snore
2:13:13 just be working on AI be making pictures
2:13:16 be making suo songs um Jim Ross uh who's
2:13:19 who's in here a lot one of the things
2:13:21 he'll do is as I show stuff he'll just
2:13:25 do exactly that but for his business
2:13:28 right and he'll just sort of use this
2:13:31 time to play and experiment so that's
2:13:33 the whole idea here so keep coming back
2:13:36 thank you to my subscribers thank you
2:13:38 Irregulars for the people that do show
2:13:39 up all the time thank you mods great job
2:13:42 tonight there were lots of good good
2:13:43 things pulled up so thank you for that
2:13:46 and uh yeah let's do a little Romper
2:13:48 Room magic mirror here it was funny I
2:13:50 heard I think it was a Tik Tok
2:13:53 today it was it was one of those Gen X
2:13:55 things where they were talking about how
2:13:57 how bitter they were as gen xers and she
2:13:59 goes and they never she doesn't and they
2:14:02 never [ __ ] called my name on Romper
2:14:04 Room magic
2:14:06 mirror I thought it was hilarious so
2:14:09 anyway so let's call some names out I
2:14:11 see Silver Fox and Joey and Tobias and
2:14:15 Tam and Amelio's wife I see I see uh
2:14:21 from
2:14:22 I
2:14:23 see uh against the norm and papa JJ I
2:14:28 already said to buas but I'll say them
2:14:30 again print hello world uh thanks Kyle
2:14:33 for another great live thank you I
2:14:35 appreciate that kind words mean a lot
2:14:37 that that means a lot thank you so
2:14:42 much Mary Mary good to see you a human
2:14:44 being
2:14:46 likewise
2:14:48 CW uh hi highish two block Tom thanks
2:14:53 Kyle you're
2:14:55 welcome I'll be at
2:14:57 work that's listen being at work okay
2:15:01 important point before we get out of
2:15:03 here thank you gritty
2:15:05 wisdom being at work is no excuse to not
2:15:08 be at the AI learning lab there's no
2:15:11 sleeping in AI keep showing up people
2:15:14 all right happy Thursday toodles and uh
2:15:17 we'll see you
2:15:19 tomorrow