AI Learning Lab

Jun 3, 2024 - Unlocking Creativity: How AI Transforms Music and Content Creation

dzmTD-Z3Bs4
Video2024-06-102:31:434 views

Description

In the vibrant world of AI and creativity, Kyle explores the intersection of technology and artistry, delving into themes of generative AI, community building, and the evolving landscape of digital expression. His discussions often highlight the transformative potential of AI tools like ChatGPT and Suno, emphasizing their ability to enhance human creativity rather than replace it. Through engaging anecdotes and real-time demonstrations, Kyle invites viewers to embrace experimentation and playfulness in their creative processes, fostering a sense of community among those navigating the complexities of AI. The conversation also touches on the challenges posed by corporate structures, as seen in his humorous rants about customer service experiences, which serve as a reminder of the human element in our increasingly automated world. For more insights and lively discussions, check out Kyle's TikTok channel at [AI Learning Lab](https://tiktok.com/@aiLearningLab). #ai #GenerativeArt #CommunityBuilding #DigitalCreativity #ChatGPT #Suno #Innovation #techtalks Chapters: 00:00:00 Weird Mary Song 00:17:03 AI-Generated Music 00:18:06 AI Generalists 00:25:14 Brain Organoids 00:27:10 Zapier Automation 00:28:03 Perplexity.ai 00:29:00 Raspberry Pi 00:31:08 Tonight's Topic Poll 00:34:50 ElevenLabs Sound Effects 00:36:11 Poll Results 00:38:23 Zapier Discussion 00:45:49 Contextual Prompts 00:49:18 AI Salon Framework 00:55:00 AI Salon Values 01:01:31 Chat GPT Explained 01:21:56 Document Editing Demo 01:30:49 Building a GPT 01:49:51 Antibubbles' Song 02:00:16 Music AI Tools 02:16:08 Gemini 1.5 Pro 02:28:00 Apple iOS 18 02:28:26 Google AI 02:31:31 Show Wrap-

Chapters

Transcript

0:01 [Music]
0:06 [Applause]
0:13 [Music]
0:29 yeah right
0:31 [Music]
0:52 oh see the hill is Qui
0:55 down where the oak trees sway
1:00 of the train that goes through the
1:05 day we will you with
1:09 a nothing ever seems to
1:12 change stor still unfold each unique and
1:23 [Music]
1:25 strange she grew up tough in a town we
1:32 Home Folks never took to her no matter
1:36 her
1:38 T longing to feel on the
1:42 inside longing for love just a
1:45 little she found herself as a guest in a
1:49 nervous
1:52 Hospital weird M from CA
1:55 Hill shees when she's
1:58 blue wo of sadness and shame under the
2:03 southern
2:05 Moon her laughter Echoes through the
2:09 wind she was never the
2:12 same weird Mary from Cedar
2:15 [Music]
2:17 Hill you'll remember her name
2:24 [Applause]
2:24 [Music]
2:36 to the gas
2:38 station win at the
2:41 clerk she stole her a soda and she
2:45 called him a
2:48 jerk in her own little
2:52 world she floats in the
2:55 breeze among the town FK who whisper
3:00 with
3:03 the and
3:05 never understood her
3:09 soul she found beauty in a
3:12 wild let her story
3:14 [Music]
3:16 unfold when the sun
3:19 sets behind those
3:22 [Music]
3:24 hills's carries and a Down shs
3:29 [Music]
3:31 Mar from CA
3:33 Hill she when she's
3:37 blue sadness and shame under the
3:41 southern
3:43 Moon Echo her laughter Echoes In The
3:46 Wind she was never the
3:50 same weird champy from Sea
3:56 Hill you'll remember his name
4:00 [Music]
4:06 good evening good people we had some
4:08 solid hell in
4:16 there weird Mary from CA
4:19 Hill she hows when she's
4:22 [Music]
4:24 blue sadness and shame under the
4:26 southern Moon
4:33 TG good people what's
4:35 [Music]
4:54 happening wait it's Monday right it is
4:57 Monday we missed you thank you Daniel
4:58 missed you guys too took the weekend off
5:02 did I do anything I didn't do
5:06 much you know yard work shopping [ __ ]
5:08 like
5:12 that got myself a nice
5:16 cafe figure out what we're going to do
5:18 tonight Mr it we miss you thank you so
5:20 much
5:22 [Music]
5:57 I'm actually really excited about
5:59 there's a new feature coming to
6:01 [Music]
6:08 sunno where you can
6:11 hum the beginning of a song or play the
6:14 beginning of a song or just play a riff
6:17 and it will continue that song and make
6:19 it for
6:19 [Music]
6:22 you which I think is really
6:25 [Music]
6:27 interesting so if I've got
6:31 [Music]
6:45 so if I have that little bit of a song
6:47 and then it writes the rest of it is
6:49 that my song or is that
6:51 [Music]
6:56 AI yeah AI a AI shamish
7:00 it kind of is it's kind of AI Shazam but
7:03 instead of identifying your song it just
7:06 writes it for
7:13 you the other
7:16 thing I got fed up with my internet so I
7:20 called Xfinity and I got one of their
7:22 stupid [ __ ]
7:24 modems this was this was staggering
7:28 Okay so I've been an Xfinity customer I
7:31 don't know years right cuz I like
7:35 torture and I have my air conditioner on
7:38 it's it got hot in here so it's going to
7:39 be loud but whatever
7:42 um okay so I call him up I'm like hey I
7:45 got shitty internet I'm thinking of
7:46 leaveing Xfinity can you give me a deal
7:50 and my internet sucks can we make it
7:52 better sure we got a package it'll save
7:55 you 30 bucks a month you even get a free
7:58 security camera and we'll say send you
8:00 out a
8:01 modem
8:04 and and I said and the modem has a
8:06 router in it right it's got Wi-Fi in it
8:09 correct yeah it's a
8:11 modem right and it has Wi-Fi in it
8:14 correct so so it was you know 40 minutes
8:18 40 minutes of just getting it going I
8:22 had to like they texted me a link to a
8:24 website where I had to go accept the
8:26 deal the terms of the deal with the
8:28 thing then we're going to send you the
8:29 thing three to five business days you're
8:31 going to have we're going to send you
8:33 the the modem we're going to send you
8:36 the camera good to go all right
8:41 woo what a pleasant customer
8:45 experience customer service
8:49 experience three days later no two days
8:52 later a a day ahead of their minim their
8:55 their
8:56 minimum a box shows up from Xfinity and
8:59 I'm like like ooh good internet's
9:03 here woo so I open the B it's a big box
9:08 so I open the box and there's one little
9:11 box in it like this big and it's the
9:14 camera and no
9:17 [Laughter]
9:20 router
9:24 so so it picks up I picks up my
9:27 telephone and it dials on the road
9:29 rotary telephone
9:33 dial I goes to the old and timey Google
9:36 search and I search
9:38 exfinity dial the number and I'm on a
9:42 hold for I don't know 20 minutes and
9:44 then I got on the call was someone
9:49 who just didn't know what they were
9:52 doing and couldn't understand
9:56 them and then I'm told well there's no
10:00 order for a modem here there's just an
10:02 order for a camera but I can sell you a
10:05 modem because it says here that you've
10:08 got your own modem and we're we're going
10:10 to send you if you want your own an
10:13 Xfinity modem that's $15 a month I said
10:16 no I talked to a guy he said it's
10:18 included in the package deal do you see
10:21 my conversation with the guy do you see
10:23 the package deal no okay so now we're at
10:27 35 minutes can I talk to your supervisor
10:30 sure but they're going to say the same
10:32 thing
10:33 okay so I get on the thing with the
10:37 supervisor and I said here's the deal
10:39 and he's like ah I see your thing here
10:41 yeah we can we can ship you out that
10:44 modem there but we need to complete the
10:47 transaction for the camera we sent you I
10:50 said okay great he goes I can't do that
10:54 that's the home security
10:57 division so he put me on hold for
10:59 another 5 minutes we got the home
11:01 security dude on who completed the
11:03 camera transaction and then he goes okay
11:06 Kyle we're going to get you set up with
11:09 your little modem here we're going to
11:11 get it sent out oops there's an
11:15 error the error is you've got your own
11:18 modem so I can't make a ticket to send
11:21 this out unless I
11:24 disable your
11:28 modem and like so I'm not going to have
11:32 internet for 3 to five business days
11:35 because you guys can't activate it after
11:38 you send
11:39 it that is correct that is correct Mr
11:43 Shannon I understand your frustration oh
11:45 do you do
11:46 you so what you're saying is I got to
11:49 drive to the [ __ ] store anyway I
11:51 could have done a week
11:56 ago so I drove to the store and he said
11:59 yeah I'm going to have to disconnect
12:01 your existing internet because you have
12:03 your own modem and I can't let you walk
12:05 out of the store unless we change the
12:08 code that you don't have your own modem
12:11 you have one of our
12:14 modems so anyway if I'm glitching
12:18 tonight I
12:20 I bad things are going to happen to
12:22 someone at
12:24 [Music]
12:26 Xfinity so anyway it should be better
12:29 unbelievable like staggeringly
12:32 unbelievable like
12:34 like and it's just one of those things
12:36 where you can't argue with it because
12:37 it's like all they've got is their
12:39 system and it determined that if someone
12:43 has their own modem you have to disable
12:45 that before you give them an Xfinity
12:48 modem that makes
12:50 sense but you should be able to send it
12:54 to them before you disable it and that
12:58 wasn't an it specif ifications for that
13:00 part of the software
13:03 request oh my God these [ __ ]
13:06 corporations like does anyone from these
13:08 corporations ever try to get service
13:11 from their own [ __ ]
13:12 [Music]
13:15 company what can you or your audience
13:17 teach share about reinforcement learning
13:20 here's some things I can teach about
13:21 reinforcement learning if you've got
13:23 Xfinity what they reinforce is that
13:25 they're [ __ ] and you're
13:27 worthless that was my human
13:29 reinforcement
13:30 [Music]
13:32 learning this is definitely a Frank
13:34 video oh this is a Frank video isn't it
13:37 I this that was pretty much a Frank rant
13:41 right there wasn't it I think this would
13:42 be a 10-minute Frank video
13:47 [Music]
13:59 [Music]
14:01 [Applause]
14:05 what do you think there
14:07 [Music]
14:12 champy I'm getting my AOL CD I got the
14:15 best model 56k can't can't best that
14:19 Kyle seriously I you know what honestly
14:23 Chris a 56k modem at this point would
14:26 probably be better than what than what
14:28 I've got although oh now that it's
14:30 plugged in and turned on I had to my
14:34 printer my Canon printer couldn't see
14:36 its W no it could see its Wi-Fi it
14:38 couldn't connect to it so I had to
14:40 hardwire the [ __ ] printer to that
14:43 thing so that was an hour wasted when I
14:46 got home tonight so I'm just in a I'm
14:49 just in a ripping snot ass mood so where
14:52 tonight's going to be a little
14:55 rampy you can make money with
14:57 ch and these [ __ ]
15:02 you look like a YouTuber I watch but a
15:04 bit older thank you I think
15:07 [Music]
15:38 wouldn't that be a fun one to start with
15:40 with a song continuation thing that's a
15:43 rift that I've played for like 15 years
15:45 and I've never figured out a song to put
15:47 behind it
15:52 [Music]
15:56 [Applause]
16:02 [Music]
16:07 so Kyle you're saying the rabbit was
16:09 more useful than
16:11 Xfinity by a
16:15 lot but any glitching so far if there's
16:18 no glitching that's a good sign I like
16:21 it Kyle thank you very much I appreciate
16:23 that yeah I like that song there's
16:24 something there's something about that
16:25 song and I don't even I don't even
16:27 really have a like a melody for it'd be
16:29 cool to see what sunno comes up with
16:34 [Music]
17:01 I don't know I don't know
17:03 man I do not know tomorrow will be a
17:06 screaming talk
17:11 really so that's a a tune you should
17:13 turn into sunno when that feature comes
17:15 out yeah exactly that's that's exact
17:17 your rabbit never arrived oh that sucks
17:20 dust in the Windish yeah it's definitely
17:22 that Dustin the wind is wait
17:26 [Music]
17:34 I close my
17:37 eyes only for a moment then a moment's
17:40 go so same chord same chord progression
17:46 [Music]
18:04 oh good lord good
18:05 people Kyle I think AI gener generalist
18:09 will become a real career we will be
18:11 like AI conductors like in a symphony I
18:15 think that's absolutely right I mean it
18:19 it's I think that is the
18:21 skill it's why black rock sorry I got I
18:24 had corn on the cob I got corn on the
18:25 cob in my teeth so I'm just going to
18:26 pick him like a [ __ ] monster
18:31 it's driving me nuts though it's like
18:33 right right in between the bottom two
18:35 teeth
18:40 um yeah I I think that's the skill it's
18:44 you know Black Rock black rock CEO just
18:47 said they're going to be hiring liberal
18:50 arts
18:51 analysts why cuz liberal arts
18:56 people you know tend to be good with
19:00 words words in the new programming
19:04 language and liberal arts
19:06 people tend to be
19:12 generalists
19:20 yeah it's just kind I I mean if you
19:24 think about that like I I just this is a
19:27 thing that sort of melts my brain about
19:28 this say I
19:32 [ __ ]
19:35 the for essentially all of
19:38 time if you wanted to get really good at
19:41 one
19:42 thing you put in your 10,000 hours right
19:46 you go get your degree you put in your
19:48 10,000 hours on weekends and
19:51 nights for 3 years four
19:55 years and because of that because you
19:57 had to put in all that time for that one
20:00 thing you tended to do that one thing
20:04 and because doing that one thing was
20:08 hard
20:11 the
20:13 society rewarded
20:16 that for kind of all of human
20:21 history and then off to the side were
20:24 the theatricals
20:25 [Laughter]
20:38 and the specialist always just kind of
20:39 tolerated us yeah we need you
20:41 occasionally go design a thing yeah go
20:43 make the could you make me a PowerPoint
20:46 presentation thanks you little design
20:48 monkey that's awesome I made money again
20:50 with chat gbt 500 bucks for 3 hours of
20:53 Consulting awesome rock
20:56 on and so all of the of a
21:00 sudden in the next 3
21:03 years that equation is going to flip
21:06 completely upside
21:08 down
21:10 where people that spent 5 years 10 years
21:15 20 years doing a single thing I'm a
21:20 this are getting commoditized out of the
21:25 market as the as the capability of
21:28 generative I goes
21:30 up the need for specialist goes down
21:35 wait
21:36 up
21:38 D sorry can't figure out which way to
21:41 put my
21:48 hands and then the need for people that
21:50 can think
21:52 horizontally so maybe you've got someone
21:54 who's got deep expertise in something
21:57 but they've got a broad perspective
21:58 they're really good at critical thinking
22:00 they're really good at curating things
22:01 from lots of different things like
22:03 creative directors and and certain kinds
22:07 of
22:09 strategists that by the nature of their
22:11 job they have to look across disciplines
22:14 right if you work at an agency for
22:16 example especially a digital agency
22:18 you've got developers and you've got
22:20 interface people and you got ux people
22:21 and you got CX people and you have
22:23 account people and you've got creative
22:25 directors and you got artists you got
22:27 writers so so if you work in an
22:29 environment where the business is
22:32 pulling all that [ __ ]
22:36 together you go from being
22:38 valuable because you can pull all that
22:40 [ __ ] together to
22:43 critical because all those Specialists
22:46 are not going to be needed how far are
22:48 we how far away are we from sensient
22:53 AI I don't
22:57 know and
22:59 the sentience thing is a really
23:01 interesting one there's is it actually
23:07 sensient
23:09 or do we perceive it as
23:12 sensient and is there a
23:15 difference and that I don't
23:21 know but if you listen to the open AI
23:24 the the the release of gp4 omni
23:30 and like if you watch that one video of
23:31 the guy going you know hey how am I
23:34 dressed for this job job interview and
23:36 she's
23:38 like you know you're looking okay I mean
23:41 if you're going for a tech job then he
23:44 puts on the bucket hat and she's like
23:46 whoops I don't know if that's the look
23:48 you're going
23:52 for the engine sitting underneath that
23:54 is not
23:56 sensient but it is
23:59 Beyond anthropomorphized right it's like
24:02 the movie
24:04 her so so there's so one definition of
24:08 sensient is do we perceive these things
24:11 as sensient whether they are or aren't
24:14 and if we do does that change how we
24:15 interact with them does that
24:17 change things
24:19 like do you turn them off
24:23 uh-oh bad people are
24:26 here so there's that and then there's do
24:31 the devices themselves or does the
24:33 software itself
24:35 ever attain
24:37 sensient and it might if you listen to
24:40 David Shapiro he he talks about um I
24:44 think it was Claude 3 started showing
24:46 signs of sensient
24:49 but it's relatively easy to
24:53 argue that it's just a probability
24:55 engine and that what it's doing
24:57 mechanically
25:01 is mechanical and cold and
25:04 unfeeling the data it's been trained on
25:06 is very human which is why it feels so
25:08 human
25:10 but I would have had a different answer
25:12 three days ago three days
25:14 ago they announced these brain organoids
25:19 I think it's a Swiss company is that
25:22 right and they've taken human stem cells
25:26 and they've grown them into brain tissue
25:29 and they make these little organoids I
25:32 don't know why they call them organ
25:33 organoids other than that's what you
25:36 would name them if you were if you were
25:37 making a science fiction
25:39 movie and you wanted to freak people
25:44 out but what they have figured out how
25:46 to do
25:47 now is they can keep them alive a little
25:50 bit longer I think they they were living
25:52 like 10 hours they basically they
25:54 basically um starve of oxygen they don't
25:57 have a circulatory system so they're
25:59 trying to figure out like an oxygen rich
26:01 medium they can put them in but they
26:03 designed a chip that can hold 16 of
26:06 these H human brain tissue organoids in
26:11 a computer chip that you can interact
26:16 with it's an actual neural network so
26:19 neural networks the things that make all
26:22 of this generative AI stuff possible are
26:24 computer systems that were designed to
26:26 mimic how we think the brain works
26:29 well these
26:31 things are how the brain
26:34 works they take very little power um and
26:38 apparently they can learn quite quickly
26:40 and you have to give them like if you
26:41 give if you just train them on stuff
26:44 they don't do well but if you give them
26:46 rules they learn really
26:49 quickly
26:52 so all of a sudden we go from this thing
26:54 being a probability engine to this thing
26:57 being a set of inputs and out outputs to
26:59 16 pieces of brain
27:03 tissue that [ __ ] thing could get
27:05 sensient pretty [ __ ] quick
27:09 right Kyle could you do a zappier
27:12 automation demo tonight I mean enough
27:14 people were interested oh if enough
27:17 people were
27:20 interested I can
27:22 do yeah if enough people are interested
27:25 I'm happy to do a zappier automation
27:27 thing it's going to be a really simple
27:31 one I'd almost rather do it tomorrow
27:34 Danielle just so I can prepare oh wait
27:36 tomorrow's a salon oh there's tomorrow's
27:39 an AI salon by the way if you don't know
27:41 what the AI Salon is you should go
27:45 there go there go to the
27:50 salon. and click on the first link that
27:53 says AI Salon community and go check out
27:56 what the salon's about on the welcome
27:58 page you'll see see what we're about
27:59 what our values are and then we have a
28:01 really cool speaker tomorrow
28:03 night and okay who gets 15 minutes of
28:06 fame tonight perplexity research
28:09 anthropic EXT tools I don't know what
28:12 those are but I heard about them but I
28:14 haven't played with them yet but that
28:15 could be fun mid Journey high level
28:17 custom GPT building daily Jank with
28:21 Frank Salon show in tell and I'll put a
28:23 seventh one in there can we update
28:26 this with uh
28:30 uh zappier
28:33 automation I don't know if you can
28:34 update
28:35 this but anyway that's kind no no
28:40 no AI salon tomorrow yay oh that was the
28:44 old pole okay
28:46 great can we use desolder to implant
28:50 into my Ras Raspberry Pi Z well
28:54 apparently these brain chips are
28:56 available I don't know what available
28:58 means I didn't I haven't looked into it
29:00 because I'm not a scientist I'm not a
29:04 researcher um so I don't know how you
29:07 get these things what the requirements
29:09 are [ __ ] like that maybe it's like you
29:12 know if you can figure out how to make a
29:14 brain organoid you can here's the chip
29:15 you can put it
29:20 on I need my other glasses cuz these are
29:24 my driving glasses oh here they are
29:31 we're driving we're driving we're
29:33 driving people
29:35 crazy Hey listen um yeah so I went to
29:39 that AI learning lab thing and it just
29:41 it I it might have been a different one
29:43 than you were talking about the one I
29:45 went to it just seemed like an older
29:47 fella kind of just talking I I didn't he
29:51 didn't he didn't I
29:54 didn't I stayed for 45 minutes but he
29:57 never really finished the sentence and
29:59 so I was just wondering was that the
30:01 that was the one that you say you go to
30:03 every night that was that was that one H
30:06 really yeah no it was interesting it was
30:09 really interesting no we didn't talk
30:11 about Ai No he talked about his dog and
30:13 he he he he screamed at the Xfinity
30:15 people a lot he there was something
30:17 about Comcast he seemed very upset he
30:19 was agitated he seemed agitated you know
30:22 what it seemed like he it seemed like
30:24 there was a uh like if if if if he were
30:28 medic ated um he he had he had missed a
30:30 day or two that's what it seemed like to
30:32 me you know it's just I'm just you know
30:34 I was coming in sort of expecting you
30:36 know it to be like a class and he would
30:38 teach us things and then but it was
30:40 really more of a a dog singing man
30:42 screaming kind of situation so I don't
30:45 if that was was that that was the that
30:47 was the show you wanted me to see
30:49 instead of instead of the Game of
30:51 Thrones reruns hm interesting I know
30:54 it's it's I if if that works for you I
30:57 think I just you know you know I love
30:59 you you know I love you okay anyway all
31:02 right and
31:08 see ah okay new poll new poll people you
31:13 got five
31:15 minutes okay so what are we gonna do
31:17 tonight play with zapier make songs
31:19 about zapier more
31:22 talk more chat GPT oh more brain Talk
31:26 Brain organized talk more GPT talk image
31:30 generation storm xinity HQ I might call
31:34 for that anyway
31:36 listen if the previous president you
31:39 know could call for his followers to
31:41 storm the capital I think we could get
31:43 Xfinity to suck a little
31:45 less okay so the pole is right there
31:48 that little clipboard thing it's a
31:50 clipboard I now know right there so
31:53 click on that you go like
31:55 this and then you should also be tapping
31:58 down here or I guess anywhere tap my
32:00 face just do like this or get your
32:03 grandkids to do
32:06 this Grandpa my finger's tired shut up
32:09 keep
32:10 tapping because we need the likes we
32:14 need the likes otherwise Tik tok's Not
32:16 Gon they're not going to send people
32:19 here okay
32:20 people all right
32:22 P Kyle what AI tool would you like to
32:25 see that hasn't been introduced
32:37 um there's there's two that I really
32:42 want I'm trying to think which one I
32:44 want
32:49 first I think I'd like reasoning first
32:52 but I think we're going to get agents
32:54 first for me for me the things that
32:56 haven't been launched yet or like agents
32:58 have been launched pseudo in open source
33:01 a little bit um but they're they're not
33:04 really ready for prime time I would like
33:07 to see autonomous agents that don't
33:11 suck and I would like to see
33:15 reasoning and so reasoning is you know
33:18 how when you type something into chat JP
33:21 and it spits out that answer and like a
33:23 lot of times the answer sucks
33:28 what
33:29 if rather than just vomiting out the
33:32 answer based on the probability of the
33:35 tokens it did the probability of the
33:37 tokens thing and then took a moment to
33:41 look at the answer and said hey here's
33:44 the question here's the answer what do
33:48 we think Bob did we get it right oops
33:51 nope it's wrong let's try
33:54 again that I would love because
33:58 it just it seems to me that's
34:03 it's this is a guy with an acting degree
34:05 telling scientists how to do their job
34:07 but like it's a second system I think
34:10 right you've got one system that
34:12 generates the response then you have
34:13 another system that looks at it analyzes
34:16 it somehow cuz these llms are good at
34:19 that [ __ ] they're good at analyzing
34:20 stuff so why not combine those two so
34:23 I'd love to see that and then autonomous
34:26 agents go book me flights go go write my
34:28 email go solve some [ __ ] write letters
34:32 to the people that haven't
34:34 paid and off they go and they just check
34:38 in with you every once in a while hey
34:40 Kyle here's the emails you asked me to
34:42 do how are they they suck all right I'm
34:45 off to write some
34:48 more vote by clicking on the Clipper
34:50 okay yeah this is the the the poll
34:52 that's still up we've got a minute left
34:57 do you want to play with zapier make
34:59 songs about zapier talk about brain
35:01 organoids talk about chat GPT do some
35:05 image generation or complain more about
35:11 Xfinity vote now or soono will win the
35:14 vote please I'll tell you what if they
35:17 give me access to that I'm going to go
35:19 look at sunno right now and see if they
35:20 G gave us access to that new feature
35:25 [Music]
35:29 what's
35:31 newso playlists are searchable no that's
35:35 not the feature we
35:36 wanted don't have it you're safe Mr T Mr
35:40 it Mr T i p the
35:46 fool oh did you guys see the the 11 Labs
35:50 sound effects
35:54 maker we have hit AGI when when AI is
35:57 good enough to make fart sounds that are
35:59 indistinguishable from
36:01 humans we're
36:08 there I so need an agent or two Xfinity
36:11 FR okay poll is over play with zappier
36:17 40% more chat GPT talk 32% and then
36:20 image generation okay we can we can do
36:22 that just realize I can get double the
36:25 Taps when tapping the trackpad and screw
36:28 on my iPad oh oh that's really good I
36:31 like it that's solid that's
36:34 solid let's go Frank Mr Kyle let's wait
36:37 let's go Frank Kyle all right idiocracy
36:40 achieved exactly I press zappier by
36:43 mistake oh xinity Frank video has
36:47 potential to go
36:49 viral yeah maybe we should do that
36:52 that's a h maybe we'll do zapier W
36:56 Wednesday
36:58 why don't we do that
37:00 Danielle hold me
37:03 accountable so tomorrow because I'm
37:05 going to have a uh cuz I'm doing a salon
37:08 my brain's going to be
37:09 fried um but
37:13 Wednesday I'll I'll prepare a little bit
37:15 ahead of the thing so I'll have some
37:17 some slides ready and and I'll make sure
37:20 that my zappier don't to zapier yeah
37:23 I'll I'll I'll um
37:29 we'll do zappier on Wednesday just just
37:31 remind me to do that all
37:39 right yeah his hair it seemed like he
37:42 freshly glued hair seems to be holding
37:45 its own Frank needs to talk about how AI
37:49 could help Xfinity oh that's not a bad
37:53 idea this is why I love the live the
37:55 pivots welcome to Chad add
37:57 [Laughter]
38:03 well I think it was
38:04 Friday Brandon or someone put up a uh
38:09 put up a poll and I predicted that I
38:11 wouldn't follow anything from the poll
38:14 and like not only did I not follow
38:16 anything from the poll like I didn't do
38:17 anything close to what we talked about
38:19 in the
38:21 poll let's vote what kind of project we
38:23 can do with
38:25 zapier here's the problem with zapier
38:27 the problem with zapier is
38:32 that zapier is essentially
38:37 programming and it's boring as snot like
38:41 troubleshooting it when your API call
38:43 isn't working is sometimes a 45 minute
38:46 Rabbit
38:48 Hole only to discover like you logged
38:50 into your wrong account or you didn't
38:52 have the right API key or your API key
38:55 expired but it takes you 45 minutes to
38:57 [ __ ] figure that that [ __ ] out that
38:59 the that's that's the problem with
39:01 zappier like conceptually it's really
39:03 easy you have these little Lego blocks
39:05 you have inputs then you process them
39:07 then you have outputs but you have to
39:09 connect all these different accounts and
39:12 you got to get them talking to each
39:13 other and then you think you got it
39:15 going but then you don't so that's
39:18 that's my only hesitation is if this
39:21 were not on Tik Tok if if if this were
39:24 more like a YouTube live stream
39:28 and I was just like letting the thing
39:31 stream while I'm working then it then
39:33 something like that's probably a little
39:35 more digestible I'm trying to understand
39:38 what happens to the chat GPT model when
39:40 it's asked to act like something okay
39:43 mud let me tell
39:46 you it's a really
39:48 good question all right we're pivoting
39:52 Lori this one's for
39:56 you Lor's Lor with another provider
39:59 other
39:59 than other than Xfinity or other than
40:03 mac I find it easier oh you find make
40:07 easier than zapier I know you know what
40:10 I tried hang on my bottom of
40:15 my hang on my bottom of
40:18 my tripod
40:21 is oh all my screws are Loose H whatever
40:26 okay um
40:32 yes a pivot
40:37 okay all
40:38 right
40:40 [Laughter]
40:43 this this is
40:45 gp4 or actually it's gp4 Omni now GPT 40
40:50 and see if you open it up what's in
40:52 there well wouldn't you like to know
40:55 that's for Sam mman to know and for you
40:57 to suck an EG but inside this
41:02 box is as much of the public internet is
41:07 they could get their hands
41:10 on and then they run it through this
41:12 thing called this Transformer
41:13 architecture that takes the documents it
41:16 ingests and it it does what's called
41:19 embeddings it it embeds it it basically
41:22 turns the words in the documents into
41:26 what are called vector so they're
41:29 mathematical numbers in Thousand
41:31 dimensional mathematical space and I
41:34 know it makes it s that sounds like I
41:37 just [ __ ] made that up thousand
41:40 dimensional Vector space but that's what
41:43 it
41:44 is so inside this box is thousand or
41:48 more dimensional Vector space and
41:51 there's all these vectors what's a
41:52 vector you know in an XY chart you've
41:56 got a point on the XY AIS AIS this is a
41:59 two-dimensional Vector space and so if
42:02 you're like up three and over seven 3A 7
42:06 that's that point in space so imagine
42:09 that times a th with trillions of points
42:11 of
42:12 light that's what's in this box
42:16 okay um does this matter well a little
42:21 bit this you can build lots of things on
42:23 top of this one of the things built on
42:25 top of this is this
42:28 that looks like printer ink Kyle no it
42:31 is not printer ink this is chat GPT chat
42:36 GPT 4
42:40 gp4 chat GPT
42:46 4 chat GPT
42:50 40 so when you switch chat GPT 40 to
42:54 chat GPT 4 it it goes to that box
42:58 okay that doesn't really matter right
43:00 now you asked
43:02 about when I say act like a market or
43:07 why does it get better okay couldn't I
43:10 just say act like you know everything no
43:13 well sort
43:14 of if you wanted to be a smartass no- at
43:17 all yes but but here's why
43:20 okay so in this box is all of the
43:23 knowledge right or all of the knowledge
43:24 that was on the public internet Andre
43:26 karpathy explained this and I I thought
43:28 this was a really good explanation of it
43:30 he said this he said if you want to
43:33 understand a specific mathematical
43:37 equation you say hey tell me about this
43:41 specific mathematical
43:43 equation and chat GPT says okay he asked
43:48 about a specific mathematical equation
43:50 hey all of the knowledge of the public
43:52 internet for the past 70 years tell me
43:54 about this mathematical equation
43:58 in this knowledge
43:59 base think of it kind of like the Grand
44:02 Canyon in and in all the canyon there's
44:04 all these little sub canyons and things
44:06 like that and when you just say tell me
44:08 about this mathematical equation it's
44:10 it's basically the helicopters really
44:12 high above the Grand Canyon it's looking
44:13 down in all the nooks and crannies and
44:16 in one of the nooks and crannies is a
44:18 math professor from Stamford that
44:20 actually invented that equation and then
44:24 in some other nook and cranny or a bunch
44:26 of students who were studying with that
44:28 guy and and they were really smart about
44:30 it and then also in those nooks and
44:33 crannies were a bunch of football
44:34 players that were forced to take a math
44:37 class so that they could graduate and
44:39 they tried to understand that [ __ ]
44:42 equation but they didn't care and they
44:45 wrote a bunch of joke answers to that
44:47 stupid math equation so if you just ask
44:50 all of the Grand Canyon tell me about
44:53 that math equation it'll essentially
44:55 pull anything that's relevant to that
44:57 prom
44:58 and kind of amalgamate it and give you
45:01 its highest probability answer so mixed
45:03 in with the good [ __ ] is the bad [ __ ] so
45:07 if you say I want you to act like a
45:10 Stamford mathematics professor that came
45:12 up with that equation and I want you to
45:14 tell me about it what it does is it sort
45:16 of lowers the helicopter into that
45:17 little sub Canyon where that specialty
45:20 knowledge is and it's it's only
45:22 returning the specialty knowledge so if
45:24 you say I want you to act like you know
45:26 everything you're just sort of
45:28 you know
45:30 generalizing the problem which is why
45:32 it's why you get such high
45:34 hallucinations is because it's pulling
45:35 from all the [ __ ] and in all the [ __ ] is
45:38 a lot of [ __ ] does that make
45:40 sense all right box metaphor
45:46 done Kyle that makes sense thanks you
45:48 thank you B for the people appreciate
45:51 that it's wild okay here's another thing
45:55 and and it's the the reasons exactly the
45:58 same so one way you can use chat GPT is
46:01 you can say I want you to act like that
46:02 math professor I want you to act like a
46:04 auto mechanic whatever it might be
46:07 um the other thing you can do
46:14 is retrain
46:17 yourself ah good mud thank you light
46:20 bulb awesome great glad that makes sense
46:24 here's another thing you can do mud is
46:29 um what Google has trained us for 20
46:33 years is give me the least amount of
46:37 words in that search box as possible
46:40 make it as efficient as possible and
46:42 then I will with all my fancy ass
46:44 algorithms try to go find you sites that
46:46 are relevant to that it tells you to eat
46:50 rocks I like to eat rocks I like rocks
46:55 um
46:58 with chat
46:59 GPT don't think of it like a search and
47:02 search and response like a promp and
47:04 response think of it more like a
47:07 conversation so rather than just saying
47:10 Hey I want you to write me a marketing
47:12 plan say hey I'm starting a business and
47:15 here's what my business is about and
47:18 here's what I'm all about and I want you
47:20 to act like a marketing expert and and
47:23 then um what do you think of my business
47:25 and then it's going to respond and tell
47:27 you what thinks of your business and
47:29 none of that's actually relevant to
47:31 where we're headed but why you do that
47:34 is because you're adding all this
47:36 context around what you're talking about
47:40 so it's increasingly
47:41 narrowing the the expertise that it's
47:45 pulling from right so if you start out
47:48 you tell it you're starting a business
47:50 and you you um you want it to be
47:53 different than every other business out
47:55 there and and um you want to start with
47:58 the marketing and you don't even have a
47:59 name for the business and maybe we're
48:01 going to come up with the name for the
48:02 business and maybe even before you have
48:04 it give you names you have it write a
48:06 creative brief based on all the other
48:08 [ __ ] you you said give you a creative
48:11 brief let it write the creative brief
48:13 and then say now give me 20 names based
48:15 on that creative brief so now you've
48:18 done you know five minutes of
48:21 conversation that's just increasingly
48:24 narrowing narrowing narrowing the
48:26 expertise of this thing to solve your
48:28 specific problem so it's completely
48:31 opposite from a Google search more
48:33 context better answers all right one of
48:37 my favorite channels oh thank you so
48:38 much I appreciate
48:41 that Kyle wait is Daisy is Daisy the one
48:45 who's who's in government yes lunatick
48:47 is that she's the one that
48:50 was that is wicked wicked smart she's
48:54 wicked smart about the
48:55 government she was an elected official
48:58 in Utah I believe and yeah she's really
49:02 good at this stuff and in fact I'm
49:03 talking to her about maybe doing some
49:05 sort of advocacy or political group
49:08 within the AI Salon chat gbt's really
49:11 helped me to develop an advanced Excel
49:12 workbook wow cool I learned what
49:15 questions to ask that's awesome that's
49:17 awesome yeah when I mean so in the AI
49:22 Salon we we've got a new framework that
49:25 we're that we're building building we're
49:27 we're kind of rearching the AI Salon
49:32 around this new framework and the
49:34 framework is it's a three-part framework
49:36 in fact let me show it to you a pivot
49:38 we're having a pivot yeah yeah
49:46 yeah I'll read you this thing so the
49:49 framework is play first mindfully create
49:53 and generously lead those three things
49:57 AI Salon play first we are a talented
50:00 group of Misfits change makers and those
50:03 brave enough to face the terrifying
50:05 uncertainty and exciting inevitability
50:08 of generative AI headon We Gather to
50:11 support and Inspire one another to step
50:13 into the unknown together on this
50:15 journey we play first mindfully create
50:18 and generously lead to ReDiscover
50:21 ourselves and reinvent the work we do if
50:23 we succeed on our journey we just may
50:25 bend the Ark of human ity ever so
50:28 slightly toward the light so that's what
50:31 the salon's about but this journey of
50:34 these three things play
50:36 first so with AI it's so [ __ ]
50:40 overwhelming that rather than try to
50:42 learn AI or or try to master AI you
50:47 can't it's changing too fast there's
50:49 absolutely too much to learn so just
50:51 play play first joyfully exploring
50:55 generative AI gaining the knowledge and
50:57 inspiration
50:59 to mindfully create so if you play
51:02 enough and sort of figure out what these
51:03 things can do then you can start to do
51:05 things like mindfully create surprising
51:08 ourselves and powerfully meeting
51:10 challenges unleashing us to generously
51:13 lead others enriching lives redefining
51:16 what's possible and unlocking a new era
51:19 of human
51:20 achievement all
51:22 right so if you're just trying to figure
51:25 this [ __ ] out take the pressure off
51:27 yourself that you have to [ __ ] solve
51:29 it you can't solve it you can't learn it
51:31 there's not a course to
51:33 take if you want to go build large
51:36 language models and get a degree in
51:38 mathematics and go to do that [ __ ]
51:40 there's lots of courses if you want to
51:43 understand how to use chat GPT in the
51:45 world nobody knows how to do that right
51:48 now and it keeps
51:49 changing so that's what this journey is
51:52 about play first mindfully create
51:55 generously lead and in that mind
51:57 mindfully create section do what you can
51:59 to solve problems solve other people's
52:03 problems or create an art project where
52:04 you're building toward an
52:07 outcome because in doing that you'll
52:10 actually discover what these tools are
52:12 good at and what they're not good at and
52:15 that's actually actually learning what
52:18 these are what these things cannot do is
52:21 probably more important than learning
52:23 what they can do
52:27 we're on Mighty networks ah yes where
52:29 where do you find people that act like
52:31 this and believe this [ __ ] you find it
52:34 there go to the
52:36 salon. that's going to take you to
52:40 a link tree you click on the first link
52:43 in the link tree that says a isolon
52:44 community and that's going to pop you
52:47 into our Mighty Network server and it's
52:49 going to drop you here and that thing I
52:52 just read to you we're actually going to
52:54 swap that out I've got to swap it out
52:56 before the meeting tomorrow if I can
52:58 we're going to swap it out and put it in
52:59 here here's our
53:01 values and we've got new values we got
53:03 five five new values now they're
53:05 Rewritten Consolidated we added two more
53:09 Consolidated the rest into into five
53:12 total this is why I feel so at home in
53:15 this community well this community is
53:17 just I mean the the amazing thing about
53:19 the values in this community is that
53:22 these values come out of who we are
53:25 already and and
53:27 why here's why you articulate values in
53:30 an organization like this you articulate
53:33 them because as the organization
53:39 grows there's going to be a point at
53:41 which it grows so much or so fast that
53:44 there are more people that don't know
53:46 what it is than do know what it is so if
53:50 if you talk to the Irregulars the people
53:52 that have been coming to this channel
53:53 for more than a year every night they
53:56 come every [ __ ] night they know what
53:59 this is about they they embody these
54:01 values that I'm about to read to
54:04 you as as the organization grows and you
54:07 get more people in that don't know that
54:09 they're like well I want it to be this I
54:10 want it to be no no no this is already a
54:13 thing so why we articulate the values is
54:15 here's the stuff we don't want to
54:19 lose all right hang
54:22 on oops
54:28 I just I had to do one of those prove
54:29 your human things okay uh oh trying to
54:35 connect a cooworker and I created a GPT
54:37 for humorous replies to customer service
54:40 inquiries it works oh that's awesome
54:41 that's super cool okay so here's here's
54:43 the values but can you are am I still
54:50 here looks like I am it looks
54:53 like let me see let me try reloading
54:56 this
54:59 yeah now it's good
55:02 okay all right so here's the
55:09 values
55:11 curiosity we're eager to ask the obvious
55:14 questions learn new things and
55:16 experiment experiment experiment our
55:19 Relentless quest for adventure and expl
55:22 exploration ignites our inner spark
55:24 Transforming Our curiosity in into
55:27 unexpected discoveries and Newfound
55:29 passions this bottom part of curiosity
55:33 unexpected discoveries and new found
55:35 passions that's what I see daily in this
55:38 community in the in the AI salon and
55:40 anyone who's learning AI is just people
55:42 [ __ ] Reinventing themselves it's
55:44 amazing so we want people that are
55:46 curious I also one of the questions I
55:48 got we we wrote this this first thing we
55:51 are eager to ask the obvious questions I
55:54 I'll tell you why I put that in there in
55:56 my career I've been around as you can
55:59 tell for many
56:01 years in my
56:04 career every single time that there was
56:08 a major
56:10 [ __ ] every single time without fail
56:13 that there was a major
56:15 [ __ ]
56:17 like the night we had built the website
56:21 for Nancy Friday who's an author and her
56:24 husband is Norman pearlstein who's the
56:25 editor and chief of time time Inc most
56:28 powerful man in
56:29 publishing and we threw a launch party
56:33 for her book's
56:34 website
56:37 and we didn't have internet and we
56:40 didn't have the website backed up on a
56:43 local hard drive and so so so we were
56:48 like thrown into Corners by Mr
56:51 pearlstein who said I want to know whose
56:54 [ __ ] heads are going to
56:55 roll [ __ ]
56:58 awful
57:00 so every single [ __ ] that I've ever
57:03 experienced that has been at kind of you
57:05 know magnitude level 11 you
57:09 know could have been presented or
57:12 prevented had someone asked the obvious
57:15 question so the obvious question is uh
57:19 hey Kyle um why does telling a large
57:23 language model to act like something
57:25 work or hey Kyle what does llm mean or
57:30 like whatever you think is the obvious
57:32 question that most people are afraid to
57:34 ask ask that
57:36 question cuz I guarantee you if you get
57:40 an answer to it someone else also had
57:42 that same question and was afraid to ask
57:44 it so that's why that's why we start
57:46 with that all right that's the first one
57:49 curiosity second value playfulness we
57:52 value creativity joy and the freedom to
57:54 experiment we play first and transform
57:57 learning into a captivating
58:00 Journey we collaborate and mindfully
58:02 create together uh together when we can
58:06 to learn more make friends and develop
58:08 new Circles of trust remember all work
58:11 and no play is how we did things before
58:13 the AI
58:17 Salon empathy we are respectful
58:20 inclusive and kind we value diverse
58:23 perspectives recognizing that our
58:25 differences make us stronger we accept
58:28 that people might be afraid and angry
58:30 and we work to understand them that
58:32 one's so
58:35 important when you tell someone about Ai
58:37 and they're like you're an [ __ ] or oh
58:40 my God is it going to kill us all you
58:43 know don't judge that this this shit's
58:46 scary this this [ __ ] is for a lot of
58:49 people going to be very existential and
58:52 for a lot of people it's going to
58:53 Blindside them
58:57 we actively practice gratitude and
58:59 acknowledge the efforts and successes of
59:01 those around us humans
59:04 matter bravery we have the courage to
59:07 step into the unknown taking bold
59:09 actions in the face of fear and
59:11 challenging ourselves to push the
59:13 boundaries of what we thought possible
59:15 let's face it with everything changing
59:17 as fast as it is we're all making this
59:19 up as we go along anyway might as well
59:21 be brave about it suck it up
59:24 buttercup very Gen X
59:27 I'm sure we're going to trigger some
59:29 trigger some gen zers with that one but
59:31 that's okay and then the last one
59:34 generosity we are willing to share our
59:36 resources feedback and wisdom with
59:39 others and support each other's growth
59:41 and development there is so much
59:43 happening and so much to learn we can't
59:45 possibly keep up with it all understand
59:48 that if even if something is obvious to
59:50 you it might be a profound Revelation to
59:53 someone else Learn out Loud share what
59:56 you know generously
1:00:00 lead my question what is a GPT exactly
1:00:03 perfect you put a smiley face there so I
1:00:06 assume you know but I can I just talked
1:00:08 about it before but I'm happy to talk
1:00:09 about it again anyway aren't those cool
1:00:12 many people are scared of AI they are
1:00:14 they
1:00:15 are and and you know what's funny be for
1:00:18 the
1:00:19 people
1:00:21 is in nearly all cases and I I you know
1:00:25 I think I think think this is with with
1:00:27 fear in
1:00:30 general that fear comes out of ignorance
1:00:33 and not ignorance like they're dumb
1:00:35 ignorance like they haven't tried it
1:00:37 most people that I know that are angry
1:00:39 about
1:00:40 AI that are terrified of it that think
1:00:44 it's going to ruin everything that's
1:00:45 it's going to destroy everything haven't
1:00:47 actually used
1:00:49 it and that's essentially why this
1:00:51 channel exists so that you know I can
1:00:56 get as many people as possible just to
1:00:59 try it like listen if you try Ai and you
1:01:02 really try it on and you play first and
1:01:05 you mindfully create and then you teach
1:01:07 people what you're doing and in in that
1:01:08 journey in doing that what you discover
1:01:12 is you [ __ ] hate it I have mad
1:01:15 respect for that I have mad respect for
1:01:19 that but if you're sitting on the
1:01:21 sidelines with your arms crossed and
1:01:22 you're like AI is stupid AI is theft AI
1:01:25 is the best place pism
1:01:28 machine oh recycle reuse you really
1:01:31 don't know okay I'll tell you I will
1:01:33 tell you what chat GPT is we we'll go
1:01:36 play with
1:01:38 it
1:01:41 um oh huge gratitude to Kelly Bosch for
1:01:44 taking the time to do weird Mary from
1:01:45 Cedar Hill oh that's what we need to do
1:01:48 we need to go show the world premiere of
1:01:50 the of the the weird Mary from Cedar
1:01:53 Hill official video so um
1:01:57 um reuse whatever your name reuses I
1:02:01 forget the first part of
1:02:08 it um I'll come back to talking about
1:02:11 what chat GPT is but we are going to go
1:02:15 there's Matt
1:02:16 wolf we are going to
1:02:19 go so if you go to the AI learning lab
1:02:22 YouTube channel or if you go to my
1:02:25 channel here on Tik Tok it's the latest
1:02:30 video but let's
1:02:41 see okay
1:02:45 so this video right here weird Mary from
1:02:48 Cedar Hill so if you weren't here when
1:02:49 we made this my son played League of
1:02:52 Legends and had AI Guru like chat GPT
1:02:55 coaching him in real time while playing
1:02:58 oh that's awesome that's amazing okay so
1:03:00 here we go let me pause this for a
1:03:04 second all right so here's what you're
1:03:06 looking
1:03:07 at about 3 weeks
1:03:12 ago some someone had asked about yudo
1:03:15 one of the music making
1:03:17 things and we went down a deep deep
1:03:20 Rabbit Hole in here and we we wrote this
1:03:22 song and so this the lyrics for this
1:03:24 song were sort of you know
1:03:26 collaborated with with all the
1:03:28 Irregulars the people that come in here
1:03:29 on a regular basis were sort of shouting
1:03:31 out ideas we we were making the lyrics
1:03:33 better um but we made this song and it
1:03:36 ended up being a really good song and
1:03:38 then I reached out to one of my favorite
1:03:40 artists right now on on all of the
1:03:43 internets is this woman named Kelly
1:03:45 Bosch b o c if you haven't seen her um
1:03:50 Tik Tok Channel you have to you have to
1:03:53 see the videos she does so I reached out
1:03:56 to her and I said hey how would you like
1:03:57 to do a collab on this song and so she
1:04:01 did she said I'm down so this is um we
1:04:06 we launched it today so this is the
1:04:08 official video for um weird Mary from
1:04:12 Cedar Hill written by myself and the
1:04:17 Irregulars
1:04:20 uh here here in the AI learning lab and
1:04:23 then video by Kelly Bosch so here we go
1:04:27 [Music]
1:04:30 [Applause]
1:04:30 [Music]
1:04:48 [Applause]
1:04:53 [Music]
1:05:00 Cedar Hill is a quiet
1:05:03 town where the oak trees
1:05:07 sway the whistle of the train Echoes
1:05:11 through the
1:05:15 day people greet you with a
1:05:19 nod nothing ever seems to
1:05:23 change but the story still
1:05:27 unfold each unique and
1:05:29 [Music]
1:05:31 strange she grew up
1:05:34 tough in the town we called
1:05:39 Home Folks never took to her no matter
1:05:45 her
1:05:47 T long you could be on the
1:05:51 inside call me for love just a little
1:05:57 she found herself as a guest in the
1:05:59 nervous
1:06:01 [Music]
1:06:04 Hospital weird Mary from Cedar
1:06:08 Hill she hows when she's
1:06:12 blue of sadness and shame under the
1:06:17 southern
1:06:18 [Music]
1:06:21 moon but when the sun sets
1:06:24 behind those Appalachian
1:06:28 Hills Mary's voice it carries and the
1:06:33 whole town
1:06:36 chills she walked to the gas
1:06:41 station winked at the
1:06:45 clerk she stole her a soda then she
1:06:49 called him a
1:06:51 [Music]
1:06:53 jerk in her own little world
1:06:58 she floats in the
1:07:01 breeze among the towns folk who whisper
1:07:06 with
1:07:09 ease in a town that
1:07:13 never understood her
1:07:17 soul she found beauty in the wild let
1:07:21 her story unfold
1:07:26 her laughter Echoes In The
1:07:29 Wind she was never the
1:07:34 same weird Mary from Cedar
1:07:37 Hills you'll remember her names
1:07:42 [Music]
1:08:05 [Music]
1:08:12 huh how good is
1:08:16 that love
1:08:19 that love
1:08:22 it I know Silver Fox isn't it so good
1:08:25 it's so good I love that the last
1:08:28 frame her in front of that cypress
1:08:31 tree that's just creepy as
1:08:37 [ __ ] side hustle Mimi
1:08:40 amazing it loves to critique okay let's
1:08:42 see very
1:08:45 cool if I saw this 10 years ago I'd be
1:08:47 amazed I mean if I saw this a year ago
1:08:50 I'd be
1:08:51 amazed that is fantastic what's the name
1:08:54 of that AI please okay so the song was
1:08:58 made with Udo ud.com
1:09:03 and
1:09:08 it's I think it's I think it's still
1:09:11 free a free account you get like 600
1:09:15 Generations a month a paid account you
1:09:17 get
1:09:18 1,200
1:09:21 um and you can just kind of roll the
1:09:24 dice
1:09:26 let's
1:09:32 see a jazz song about a Midnight Train
1:09:36 to Nowhere all right we're going to
1:09:39 autogenerate the lyrics and so you just
1:09:41 hit
1:09:42 create and then it does
1:09:44 that okay let's see
1:09:58 okay that's good
1:10:11 Brandon so we'll come back to that hang
1:10:13 on let me just go grab this
1:10:16 GPT oh I love that image who did that
1:10:19 image meimi side hustle Mimi this is the
1:10:22 tsunami of AI That's
1:10:24 Coming that is such the Right image it
1:10:27 is such the Right
1:10:29 image so
1:10:31 cool
1:10:38 um all right let me pop that up where
1:10:41 okay I've got it there let me just move
1:10:43 it over here so I can find it
1:10:47 again
1:10:49 okay a logical reasoning GPT for complex
1:10:54 insights by community Builder Okay so so
1:10:57 we'll play with
1:10:58 this where do you share and post the
1:11:01 song how do people find it so right now
1:11:05 it's
1:11:06 posted in the Irregulars channel on the
1:11:08 AI Salon it's at the AI learning lab
1:11:11 YouTube channel and then it's also on my
1:11:13 personal Twitter
1:11:15 account Kyle
1:11:18 Shannon but I would say the YouTube
1:11:20 channel is where to go the so it's the
1:11:23 the AI learning lab YouTube channel is
1:11:26 AI learninglab
1:11:29 dtt I wasn't quick enough to get the the
1:11:32 real the real
1:11:34 one so dtt or just go look up weird Mary
1:11:39 from Cedar Hill official
1:11:46 video and then here let me show you
1:11:48 another cool thing from that
1:11:57 there's the cover the album cover weird
1:12:00 Mary from Sedar
1:12:05 Hill
1:12:07 okay
1:12:10 so my question what is a GPT
1:12:15 perfect okay I'm going to go back to the
1:12:18 Box
1:12:28 all right uh let's
1:12:31 see yeah this is good okay so we have
1:12:35 this we have this and we have
1:12:39 this so I'll do it conceptually first so
1:12:42 this is what they call the Frontier
1:12:45 Model the underlying model that when
1:12:50 they when they talk about a large
1:12:51 language model think of it like a box
1:12:54 full of all of the knowledge well not
1:12:56 all of the knowledge as much of the
1:12:58 knowledge they could glean off the
1:13:00 public internet right so they used all
1:13:03 sorts of fancy ass web scraping tools
1:13:06 they they effectively compress that
1:13:09 knowledge into this new way of of
1:13:11 compressing it and searching it called a
1:13:14 Transformer that transforms it from
1:13:18 words and papers and documents into a
1:13:22 mathematical
1:13:23 space with these things called tokens it
1:13:26 doesn't [ __ ] matter but it's thousand
1:13:28 dimensional mathematical
1:13:29 space okay that's GPT so when you hear
1:13:33 them talk about
1:13:34 gp4 or gp4 Omni that's the underlying
1:13:38 model chat
1:13:39 GPT sits right there this is an
1:13:43 application that actually when you type
1:13:47 when you log in and you see the little
1:13:48 prompt and it's got those little starter
1:13:50 questions and you type in your prompt
1:13:53 and it gives you an answer and it can
1:13:55 code and it color codes all that that's
1:13:58 an application that the people at open
1:14:00 aai wrote that sits on top of
1:14:03 this and that's why if you go you you'll
1:14:07 sometimes hear people say oh Microsoft
1:14:10 co-pilot is free chat GPT 4 it's not
1:14:14 really they both use the same underlying
1:14:17 model but this is chat GPT this is
1:14:21 co-pilot and co-pilot got lobotomized
1:14:25 shortly after it came out so it still
1:14:27 hasn't caught up okay so
1:14:30 anyway GPT 4 chat GPT and then what do
1:14:34 they name the little customized things
1:14:37 gpts because why not confuse the [ __ ]
1:14:40 out of people so this is a custom GPT
1:14:45 well this is actually a guitar interface
1:14:49 for for something whatever doesn't
1:14:51 [ __ ] matter um a custom GPT is a very
1:15:00 specific kind of small focused version
1:15:05 of chat GPT so this is the general thing
1:15:07 you can ask it
1:15:08 anything a custom GPT like sits on top
1:15:12 of this and so you can make your own
1:15:15 custom GPT it's actually quite easy so
1:15:18 that's it conceptually so let me go show
1:15:20 you what that actually looks like in
1:15:24 practice m
1:15:31 so you go to chat gp.com
1:15:35 and so if I'm here this is that blue
1:15:38 bottle right the blue bottle of chat GPT
1:15:40 let me make this a little
1:15:45 bigger okay so there's your your little
1:15:48 pre-prom questions right or you can type
1:15:50 in your own so I can say um
1:15:56 write me a poem in the style of an &m
1:16:00 rap that explains quantum
1:16:06 mechanics and it will do
1:16:17 that great now come up with a game based
1:16:21 on that poem
1:16:26 Quantum Quest story mode challenge mode
1:16:29 multiplayer
1:16:44 mode great now write the code for me
1:16:49 great now write the code for me
1:16:56 okay now it's doing the code so if you
1:16:58 haven't played with chat
1:17:00 gbt you
1:17:03 should CU because I'm just [ __ ]
1:17:06 around but and you know will this code
1:17:08 work I don't know probably it's it's
1:17:11 surprisingly good and it's still
1:17:14 going okay now you'll notice over here
1:17:18 on the left I've got my prompt history
1:17:21 so this is all my prompt history and
1:17:23 then up at the top here I've got these
1:17:24 things called gpts so those are those
1:17:27 like iig things sitting on top of the
1:17:29 blue bottle and you can say explore gpts
1:17:32 so if I say explore gpts and I go to the
1:17:34 writing section for example here are
1:17:37 some specific
1:17:39 gpts that people out in the world
1:17:43 created so here's an AI humanizer the
1:17:46 number one AI humanizer in the world get
1:17:48 humanik content in seconds this GPT
1:17:51 human humanizes AI generated text
1:17:54 maintaining content meaning and quality
1:17:57 now supports multiple languages Beyond
1:17:58 English um so there's some writing ones
1:18:01 here's productivity ones
1:18:04 um research ones so anyway you can
1:18:07 search these so my very first GPT I ever
1:18:10 made I love I love Hawking this one cuz
1:18:13 it's fun it's called Sin City
1:18:16 sipper so if you go to chat GPT right
1:18:21 now and you click on Sin City sipper
1:18:30 so
1:18:34 now we're in chat
1:18:36 GPT but we're in a subset of it that I
1:18:41 have written a specific
1:18:44 prompt that says I just want a GPT
1:18:47 that's its only purpose in life is to
1:18:48 make cocktail recipes and it's so it's a
1:18:51 Vegas bartender with a Twist on the
1:18:53 classics # Ender #
1:18:57 Mixology and then I've got these little
1:18:59 pre-written questions and they're all
1:19:01 about it being a um a bartender so
1:19:03 reinvent the margarita for me so I click
1:19:06 on
1:19:07 that and it comes up with a drink recipe
1:19:10 for
1:19:12 us and then it actually starts making
1:19:14 the image of the drink but let's see
1:19:16 what it created High Roller Margarita 2
1:19:18 O of Premium
1:19:21 tequila Rosado or ano grandar
1:19:25 nice lime juice agave nectar orange
1:19:29 bitters gold salt for [ __ ] nice and
1:19:32 then a lime wheel an edible Gold Leaf
1:19:35 for garnish and so there's your
1:19:37 instructions on how to make it and
1:19:39 there's your drink there there's the
1:19:41 photograph of your drink ready for your
1:19:45 menu right so custom gpts are these
1:19:49 things that allow you to just I I've got
1:19:52 another one that makes pictures of 7s
1:19:54 muscle car Resto mods and it makes them
1:19:57 look like a poster you would have hung
1:19:59 on your walls in the in the
1:20:01 1970s and it's awesome and fun and then
1:20:03 some of them are super serious we did we
1:20:05 did a project between New Year or
1:20:08 Christmas and New
1:20:09 Year's where we got as many people as we
1:20:12 could and a lot of the people in this
1:20:14 channel that are
1:20:16 Irregulars um and we built as many gpts
1:20:18 as we could for nonprofits and in a
1:20:20 24-hour period this community built 150
1:20:24 gpts it was amazing for
1:20:27 nonprofits now one of the challenges of
1:20:30 gpts is that they historically have only
1:20:34 been available to people that pay the 20
1:20:37 Buck a month subscription to chat gbt
1:20:41 most people weren't doing that so open
1:20:44 AI just made these things available for
1:20:47 everyone and I'm actually going to go
1:20:49 check something cuz as of 4 days ago you
1:20:53 still couldn't make images with a free
1:20:54 account account let's see if they've
1:20:56 changed that my guess is they're not
1:20:58 going to change that until they launch
1:21:00 the new image generation tool for paid
1:21:02 members is my guess but let's let's
1:21:05 see so if I go to chat
1:21:15 GPT all right and if I go to cinity
1:21:19 sipper let's make a classic martini
1:21:23 classier come on
1:21:28 we'll see if it can actually make an
1:21:29 image
1:21:32 yet
1:21:34 nope all right I can't create images
1:21:37 right now
1:21:39 so get chat GPT plus to create images
1:21:42 and more so they've got it they've got
1:21:44 that part of it hobbled but you can
1:21:47 still use the GPT so especially if
1:21:49 you're using things for writing or
1:21:51 things that do Advanced Data analysis
1:21:53 things like that
1:21:56 um you you can do all that now even
1:22:00 though it can't generate images it it
1:22:03 does have you can upload images to it
1:22:05 and have it analyze images you upload
1:22:08 even with the free version
1:22:10 so let's close that does that answer
1:22:13 your question so it's like work is
1:22:15 already done yes exactly it's exactly
1:22:19 it's
1:22:20 like it's like someone who's super
1:22:23 passionate about a topic has pre pre-
1:22:25 prompted chat
1:22:27 GPT to do something specific for
1:22:31 you so like maybe someone's super into
1:22:34 gardening or flower arranging or
1:22:37 interior design they can create a GPT
1:22:40 that's trained on their specific
1:22:41 knowledge they can prompt it in a
1:22:43 specific way they can even upload like
1:22:46 if they've written a book about the
1:22:47 theory of you know funu in interior
1:22:50 design they can upload a PDF of their
1:22:53 book and use that as the colge base for
1:22:55 that GPT so you can interact with their
1:22:57 book I made a custom Excel wizard GPT
1:23:01 and it blew my mind I'm light years
1:23:04 ahead of my co-workers now I so I don't
1:23:07 know how to say your name wait Mack
1:23:10 Mack
1:23:11 Maca I don't know how to read cilic
1:23:15 letters but I got
1:23:18 Mack um but yes exactly you can you you
1:23:23 know these G gpts here's the other thing
1:23:26 about gpts that I I find
1:23:29 amazing
1:23:32 um they're so easy to make in fact
1:23:37 should we just make
1:23:41 one they're so easy to make that you
1:23:44 can you can make them can chat GPT
1:23:48 highlight colorized changes you make to
1:23:51 a document you upload
1:23:57 I don't know let's
1:24:02 try so I'm going to export this as a
1:24:07 word
1:24:23 dock that's a really good
1:24:26 question okay so I'm going to go to chat
1:24:28 GPT new prom new oh
1:24:32 prompt make one yeah okay we'll we'll
1:24:34 we'll make a GPT as well oh here's what
1:24:36 I was going to say about
1:24:39 gpts you can make them very informally
1:24:43 and I and I'll tell you what I mean by
1:24:47 that even though that what you're
1:24:50 creating when you create a custom GPT is
1:24:53 an incred L sophisticated
1:24:56 application that has all of the Power of
1:24:59 chat GPT 4 Omni sitting underneath the
1:25:03 hood you can make really trivial gpts or
1:25:07 really
1:25:08 short you know limited audience gpts and
1:25:11 I'll give you example Peter Kaminsky was
1:25:13 the first one to he told me this and I
1:25:17 like I just lost my mind it broke my
1:25:19 brain he said hey Kyle I just got off a
1:25:23 conference call with four other people
1:25:25 people and I took the transcript of that
1:25:27 call and I used that as the knowledge
1:25:29 base for the GPT so I created a GPT
1:25:32 about that phone call and I sent it to
1:25:35 the four people that were on the call
1:25:37 and I said hey the call notes for what
1:25:39 we talked about are in here so you can
1:25:41 ask this GPT about what we talked about
1:25:44 and it will
1:25:47 know like that that just blew my mind
1:25:49 cuz when I typically think about
1:25:51 software development you're like I've
1:25:52 got to make this really powerful app
1:25:54 that's going to change the world and no
1:25:56 you could make one that just remembers
1:25:58 what you talked about in that call it's
1:26:01 like holy [ __ ]
1:26:03 okay so we're going to try we're going
1:26:06 to try a little experiment here the
1:26:08 question the experiment question was if
1:26:10 I upload a
1:26:12 document can you have chat GPT make
1:26:16 changes to it and then highlight those
1:26:17 changes so that's what we're going to
1:26:19 try to do so I'm going to upload from
1:26:21 the computer
1:26:25 Salon mission statement so it's got the
1:26:27 salon mission statement in there going
1:26:29 to say
1:26:33 um what is
1:26:36 the mission
1:26:40 statement at the top of this document
1:26:45 question
1:26:47 mark all
1:26:49 right looks like it got it right perfect
1:26:55 perfect now I want you to add a sentence
1:26:58 at the
1:27:01 end that talks about how much we like
1:27:04 turtles
1:27:07 [Laughter]
1:27:13 period and of course we really like
1:27:17 turtles great now I want you to add that
1:27:20 sentence to the uploaded word dock
1:27:26 and I want you to highlight that
1:27:28 sentence you added so that I can see
1:27:30 that it's
1:27:34 different we should get yeah it says
1:27:37 analyzing here so it's what it's
1:27:38 actually doing right now is writing
1:27:40 python code so it's using its Advanced
1:27:42 Data analysis to actually rewrite the
1:27:45 document hopefully it's going to add
1:27:47 that sentence in and highlight it in
1:27:49 some
1:27:50 way my my instinct is this is going to
1:27:53 work I've added the sentence and of
1:27:56 course we really like turtles to the
1:27:57 mission statement and highlighted it for
1:27:59 visibility you can download it here all
1:28:02 right so there it
1:28:06 is and it looks like it did it so if I
1:28:10 go um we'll go here we'll go
1:28:16 open go upload
1:28:22 browse updated
1:28:34 why can I why why did that not
1:28:42 work come
1:28:44 on all right [ __ ] it we'll open it
1:28:47 Microsoft Word I was hoping I didn't
1:28:49 have to do that
1:28:59 [Music]
1:29:02 yep look at that beautiful it did it hey
1:29:06 Ma it rewrote the word
1:29:09 file added the sentence and highlighted
1:29:11 it so yes the answer is
1:29:16 yes my Excel workbook is so heavy with
1:29:19 formulas now the laptop struggles that's
1:29:21 hilarious yes it can jenu said yep dang
1:29:25 yep I forgot you can upload transcripts
1:29:27 yeah I mean astounding right and so you
1:29:31 can you can have it do like one of the
1:29:33 things it's amazing at it can clean up
1:29:35 data so you can up upload something with
1:29:37 really shitty dirty data in it and tell
1:29:40 it to identify what's dirty about the
1:29:42 data clean it up and rewrite the rewrite
1:29:45 the spreadsheet for you with clean data
1:29:47 in it can AI
1:29:50 do prayers in specific topics
1:29:55 yeah I mean it's a large language model
1:29:56 it can do anything that involves
1:29:58 language which religion is nothing but
1:30:02 for the most part you know and the faith
1:30:04 part but but you know how we communicate
1:30:07 uh the faith part is through language
1:30:10 right so it's really good at that stuff
1:30:13 it's really good at like Shakespeare and
1:30:16 writing and storytelling and things like
1:30:18 that because that's a place where the
1:30:21 hallucinations that where it's if it's
1:30:23 not factually correct if you're doing
1:30:25 things like you know trying to
1:30:27 creatively put two ideas together it's
1:30:29 really good at that so yes it it can
1:30:32 absolutely do that yeah it could
1:30:35 generate prayers in the style of a
1:30:36 specific religious belief and in the and
1:30:40 in the tone of a specific comedian or of
1:30:43 a specific oror that you really like so
1:30:47 yes absolutely all right so let's go
1:30:49 make a GPT now
1:30:57 okay so here we are at chat JP you can
1:31:00 make money with chat J I'm going to turn
1:31:03 off the air conditioner get a little chy
1:31:05 my feet
1:31:07 cold you dang people you dang people
1:31:11 with all your silly questions about the
1:31:13 JP chat
1:31:15 JP why do we learn chat JP you can make
1:31:19 money
1:31:20 with to make money it's all about the
1:31:22 money people about the money about the
1:31:24 money you make GPT you make the money
1:31:26 you get the money how do you make the
1:31:28 money I don't know yet we haven't
1:31:29 figured that out ain't no way to make
1:31:31 money yet you know who's making money
1:31:33 you know how these people are making
1:31:34 money selling your courses on how to
1:31:36 make money with chat
1:31:38 gbt they're not making money with chat
1:31:40 gbt they're making money on your sorry
1:31:42 ass buying their stupid courses
1:31:45 okay enough comedy
1:31:47 jokes uh uh uh so we're at our chat GPT
1:31:51 we're over here if you want to make your
1:31:53 own GPT like it is staggeringly simple
1:31:56 now if you're not a
1:31:58 paid uh chat GPT plus member you can't
1:32:02 do this you have to be a paid member to
1:32:04 make these custom
1:32:06 gpts in my opinion even though they're
1:32:08 giving away chat GPT for Omni to free
1:32:11 users it's very very limited and one of
1:32:14 those limitations is you can't make
1:32:16 custom
1:32:17 gpts um I think the 20 bucks a month is
1:32:19 still worth it because there's some new
1:32:21 stuff coming that's going to be very
1:32:23 worth it so for what that's worth okay
1:32:26 so you go to explore gpts and again you
1:32:29 can you know scroll down through here
1:32:30 you can search for
1:32:32 things up in the right hand corner it
1:32:34 says
1:32:35 create so I'm going to create a
1:32:38 GPT and it throws us into this interface
1:32:40 where there's a right hand Pane and a
1:32:42 left hand pane on the left hand pane
1:32:44 this is where you invent it this is
1:32:46 where you make it and on the right hand
1:32:48 pan that's where you play with it what's
1:32:50 amazing about this is as you sort of
1:32:53 type the the instructions to make it
1:32:55 it's generating it over in the right
1:32:57 hand pain you can use it immediately you
1:32:59 can test it
1:33:00 immediately um some components over here
1:33:03 so there's two modes there's create mode
1:33:05 and configure mode and I'll talk about
1:33:07 configure first and then we'll flip over
1:33:09 to create so when creating a GPT are you
1:33:12 keeping the instructions as a file or
1:33:14 just pasting the instructions great
1:33:17 question and the answer is actually
1:33:19 could be both
1:33:22 um and you don't even need to think that
1:33:26 hard about it initially so it'll make
1:33:29 more sense in a second here on the
1:33:31 configure side the little plus button
1:33:33 here is if we want to upload a picture
1:33:35 you've got your name of your GPT your
1:33:37 description of your GPT the instructions
1:33:40 this is your prompt right for what the
1:33:41 GPT does and then it's got these things
1:33:44 called conversation starters so those
1:33:46 are those pre- prompted questions so
1:33:47 when I showed Sin City sipper earlier
1:33:50 and it said I want to make a classy or
1:33:51 Martini or whatever it said that was one
1:33:54 of these conversation starters that's
1:33:55 what I wrote in there then there are
1:33:57 these things called knowledge you can
1:34:00 upload knowledge to your GPT so if
1:34:04 you've written oh I'll I'll give you a
1:34:06 really specific example we did this at
1:34:08 my company Story vine we've got a sales
1:34:11 document it's a 50-page sales document
1:34:13 we've put together over the past decade
1:34:16 and it's there's just all sorts of [ __ ]
1:34:18 in it's really hard to parse and read we
1:34:20 took our 50-page sales document uploaded
1:34:23 it into a GPT and we now have an
1:34:25 interactive tool that tells us if if you
1:34:28 encounter a certain problem here's how
1:34:30 you position it if you want to talk to
1:34:32 this kind of customer here's how you do
1:34:33 it so you can upload your own data to it
1:34:36 the other thing you can do is you can
1:34:38 turn on or off all of the capabilities
1:34:41 of chat GPT so web browsing you can turn
1:34:43 on or off access to the internet you can
1:34:46 turn on or off the ability to generate
1:34:48 images and you can turn on or off the
1:34:49 ability to do um Advanced Data analysis
1:34:53 and then this ctions thing at the bottom
1:34:55 if you're a programmer and you want to
1:34:56 include apis to other services you can
1:34:59 do that it's jankier than [ __ ] um but it
1:35:02 is possible but anyway if your head just
1:35:06 exploded and you're like but Kyle that
1:35:07 seems [ __ ] exhausting I got a I don't
1:35:10 know how to write a prompt for this
1:35:11 stupid [ __ ] thing what am I supposed
1:35:13 to do here Kyle hey Kyle Kyle Kyle I
1:35:17 just want to make a
1:35:19 GP up until this tool came out
1:35:25 if you were in that position where
1:35:27 you're just like I'm too overwhelmed and
1:35:29 I'm not a [ __ ] programmer and I don't
1:35:30 feel like figuring this [ __ ] out well
1:35:33 you were [ __ ] out of luck like if you
1:35:35 wanted to make a bot over on PO this was
1:35:37 essentially the interface but what chat
1:35:39 GP is done I ran into an issue with
1:35:42 character limit on the
1:35:44 instructions hence why I asked creating
1:35:46 a file so yeah if if you're if you blew
1:35:48 out your uh your uh prompt area and
1:35:52 instructions I would put it in a text
1:35:54 file or a PDF and upload it um as a as a
1:35:57 data file and then you can tell it in
1:35:59 the instructions hey I'm uploading this
1:36:02 file and and in fact here so so we'll go
1:36:05 do this so now I'm going to show you
1:36:06 something really [ __ ] cool so one of
1:36:09 the great things that open AI did when
1:36:11 they launched this tool is they created
1:36:12 this create Tab and what this is is this
1:36:15 is using chat GPT to build your Bot to
1:36:19 build your
1:36:20 GPT so I what I'm going to do is I'm
1:36:23 actually going to upload
1:36:27 that document let's
1:36:30 see yeah I'm I'm just going to grab it
1:36:33 so I'm going to upload the salon mission
1:36:36 statement right there's the salon
1:36:37 mission
1:36:40 statement okay and I'm going to
1:36:46 say attached is the mission statement
1:36:49 for a community I run called the AI
1:36:52 salon
1:37:01 and I want to create a
1:37:05 GPT that answers people's questions
1:37:09 about what the community is all
1:37:12 about and inspires
1:37:18 them to step on the journey
1:37:27 I think we'll just end it right there so
1:37:30 no programming I don't need to figure
1:37:32 out how specifically to prompt this I
1:37:34 don't have to figure out [ __ ] I just hit
1:37:37 return and it says great let's start by
1:37:40 defining the
1:37:45 behavior it's telling me here's our
1:37:47 prompt starters here's the welcome
1:37:49 message I'll Now update the behavior
1:37:52 with these details
1:37:54 that's weird I've never seen it do that
1:37:58 before um I'm gonna say Okay
1:38:02 build oh okay now it's doing it see
1:38:05 where it says updating GPT so now it's
1:38:07 building the GPT this is kind of new
1:38:09 Behavior where it sort of described what
1:38:10 it was going to do something went wrong
1:38:13 it looks like there was an issue
1:38:14 updating the behavior let's start
1:38:16 by uh perfect
1:38:28 okay it's it's broken somehow something
1:38:31 went wrong
1:38:33 um let me see let's start
1:38:39 over or let's tart over
1:38:46 [Music]
1:39:00 uh yeah it's it's not happy right
1:39:03 now now if if [ __ ] ain't
1:39:07 happy let me see
1:39:11 um I want to create a GPT that helps
1:39:14 someone design a community by asking
1:39:17 them a series of questions period
1:39:25 so it's got this new Behavior where it's
1:39:26 describing what it's going to
1:39:29 do that's new Behavior to me since the
1:39:32 last time I've done
1:39:33 this it used to just say
1:39:36 building and then I guess it was doing
1:39:38 all this [ __ ] on the
1:39:50 background nope
1:39:55 noer updating
1:40:06 GPT all right did it do it oh yeah see
1:40:09 this this one did it I want to start a
1:40:11 community for local
1:40:13 artists so there it is if you have any F
1:40:16 further questions let's see uh let's see
1:40:19 make
1:40:21 um the I
1:40:28 con so it's generating the profile
1:40:37 picture boom There It Is Love lovely it
1:40:40 says FR Dei guide perfect Community
1:40:46 architect so so this is now a functional
1:40:51 version of this GPT so if I click on I
1:40:53 want to start a community for
1:41:03 artists there's all sorts of weird
1:41:05 screen redraw [ __ ] going on I don't know
1:41:07 what's going on it could be my browser
1:41:10 could be all [ __ ] up so hang on so
1:41:12 anyway so there's that let's so now I'm
1:41:15 going to say create in the upper right
1:41:16 hand corner I'm going say create and
1:41:19 then I can I can save it so that only I
1:41:22 can use it I can save it so that anyone
1:41:24 with the link can use it or I can
1:41:26 actually um add it to the GPT store
1:41:30 so very very quickly you can just tell
1:41:34 it what you want it to make and then
1:41:36 save it I'm just going to save it as
1:41:37 anyone with link because this isn't
1:41:39 something I want to publish but I'm
1:41:41 going to say anyone with
1:41:46 link and then if I say view GPT so now
1:41:49 I'm in GPT so that's so we're done like
1:41:51 we we basically put in a description I
1:41:53 want a GPT that does blle and here it is
1:41:56 and now it's done now I can go back in
1:41:58 and edit this
1:41:59 thing and I'm going to flip back over to
1:42:02 this configure mode and now all of that
1:42:04 stuff that wasn't filled in before is
1:42:06 now filled in there's the image there's
1:42:08 the title the description here's the
1:42:10 prompt that makes it up and I can add to
1:42:12 this prompt I could
1:42:16 say
1:42:19 um
1:42:21 after every response
1:42:28 make an
1:42:32 inspiring
1:42:34 um photo
1:42:37 of diverse
1:42:47 people
1:42:49 engaged in
1:42:52 community activ
1:42:55 activities make an inspiring let's do 16
1:42:59 by9 wide so I'm going to even give it an
1:43:02 aspect ratio wide
1:43:04 photo so after every response this thing
1:43:07 should make us a photo which is
1:43:12 annoying it's
1:43:15 annoying if I were making this for real
1:43:18 but it's it's something good to
1:43:19 demonstrate so let's see all right so
1:43:22 view our GPT
1:43:24 how can I ensure my online what
1:43:26 activities here we go what activities
1:43:28 can I organize to keep my members
1:43:30 engaged playful workshops discussion and
1:43:33 exploration
1:43:35 meetups Community build building
1:43:43 activities lovely
1:43:47 response oh it didn't make it uh let's
1:43:50 see make the image
1:43:54 don't show the code um this flakiness of
1:43:59 chat GPT right now where it just showed
1:44:01 the code of the image instead of
1:44:03 actually making the
1:44:04 image that's one of those sneaky things
1:44:06 I I I have a sneaking suspicion they're
1:44:08 updating servers at at chat GPT right
1:44:11 now but anyway there's the image it made
1:44:13 you know based on our diverse people
1:44:16 having their fun community activities
1:44:19 there you
1:44:19 go do we have an open AI app for Windows
1:44:23 lap top to use chat
1:44:27 GPT no but you can use well on Windows
1:44:31 you've
1:44:32 got the new
1:44:35 co-pilot app is effectively the chat GPT
1:44:39 app it's not exactly the same the chat
1:44:42 GPT app for Mac is not fully launched
1:44:46 yet it's it's out and you can you can
1:44:48 run it but it doesn't have any of the
1:44:50 new fancy voice stuff or Vision stuff
1:44:53 that that they
1:44:54 demoed um it was weird that Microsoft
1:44:59 invested $1 billion dollar in them and
1:45:01 they didn't make a Windows app but then
1:45:04 the week after they launched gp4 Omni
1:45:07 Microsoft launched the co-pilot app that
1:45:10 that effectively is the same thing so on
1:45:13 Windows you'll have the co-pilot app so
1:45:15 I would just go look up co- co-pilot yep
1:45:18 me too it's been weird tonight yeah I
1:45:19 have a feeling open AI is is updating
1:45:22 some software right now that's usually
1:45:25 when it gets just like
1:45:28 weird where it normally behaves a
1:45:31 certain way and then it just is like no
1:45:33 I'm not going to do that or I'm just
1:45:34 going to show you code instead of an
1:45:36 image that that generally is they're
1:45:38 updating something I don't know why it
1:45:40 [ __ ] it up like that but but it
1:45:44 does all right does that answer your
1:45:47 questions that's what a GPT is that's
1:45:50 how you make
1:45:51 gpts yes you can modify documents and
1:45:54 have it highlight things within the
1:45:56 document you modified super cool I kind
1:45:58 of learned that's new I didn't really
1:46:00 know it could do that Lord digital Gods
1:46:03 gifting the croses very much appreciated
1:46:06 sir those gifts are very much
1:46:09 appreciated
1:46:11 um recycle reuse thanks you're welcome
1:46:14 yeah like and and and seriously let's
1:46:18 you know going back to that that
1:46:20 framework from the AI salon
1:46:23 play first just go play with this [ __ ]
1:46:26 make stupid gpts you know make like the
1:46:30 one I made the the one I made that um
1:46:32 makes 70s muscle car posters it's one of
1:46:36 my favorite gpts I love it and you know
1:46:39 the the Mixology one the the Sin City
1:46:41 sipper
1:46:44 um in doing stupid [ __ ] like that you
1:46:47 actually learn how to use these tools
1:46:49 right because you kind of push the
1:46:51 boundaries by just being stupid you're
1:46:53 kind of free to go well let me try
1:46:54 whatever if you're like I've got to
1:46:56 learn how to do this for work cuz I got
1:46:58 to train my Excel to do
1:47:02 the then you're going to get all [ __ ]
1:47:05 frustrated but if you just go play push
1:47:07 the boundaries like use tools like maybe
1:47:10 you're really shitty at music go play
1:47:13 with yudo and start writing
1:47:17 songs one of my favorite stories one of
1:47:19 the one of the earliest members of the
1:47:22 AI salon um this guy named
1:47:26 Roger was a technologist when we started
1:47:28 the AI Salon it was actually called the
1:47:30 AI artist
1:47:31 salon and Roger has been a software
1:47:35 engineer for his entire
1:47:37 career and he's you know he's
1:47:41 he's he's Saucy and he's kinly and he
1:47:44 like he just always wants to question
1:47:46 everything right it's just
1:47:49 hiso and
1:47:51 so we we started out with a Discord
1:47:55 server and we put a mid
1:47:57 Journey instance in our Discord server
1:48:01 and Roger came in oh good antie bubbles
1:48:04 I posted a salon a song in the salon
1:48:06 that's awesome okay great we we'll check
1:48:09 it out where did where did you put it
1:48:10 did you put it in Irregulars or in
1:48:11 showand
1:48:12 tell so Roger comes into the
1:48:15 Discord and because he's you know he's a
1:48:19 Nero well he comes in and he changes his
1:48:22 name to Roger not an
1:48:27 artist and he was asking all of the all
1:48:29 of the artists that were using mid
1:48:31 Journey he was asking them well how did
1:48:35 you prompt it to get it to do that why
1:48:37 isn't it more predictable how would you
1:48:39 get it to do this how would you get it
1:48:40 to do that and most of the artists are
1:48:42 like I don't know just go try and so
1:48:45 Roger took
1:48:47 his 30 40 Years of engineering thinking
1:48:51 and he went into mid journey and started
1:48:52 learning about mid Journey prompting and
1:48:55 wouldn't you [ __ ] know it in him
1:48:57 [ __ ] around with mid Journey he
1:49:00 discovered well first of all he got
1:49:02 really good at Mid Journey like he was
1:49:04 making images that were just kicking the
1:49:06 [ __ ] out of you know everyone in there
1:49:08 like cuz he was just he was taking a
1:49:09 scientific approach to
1:49:11 it and he got so good at the art stuff
1:49:14 he started loving it he actually got two
1:49:16 of his pieces accepted into an art show
1:49:18 in Boulder an AI art show he got two of
1:49:21 his pieces shown there um and he he
1:49:24 actually changed his name to Roger
1:49:26 becoming an
1:49:27 artist so play play even if if you're
1:49:32 playing with things that you feel like
1:49:34 are outside of your boundaries
1:49:39 okay because it will serve you it really
1:49:41 will I promise you
1:49:51 [Music]
1:49:56 nice embrace the Jank
1:49:58 Turtle
1:50:00 [Laughter]
1:50:05 [Music]
1:50:09 Tobias
1:50:11 where Kyle don't forget to check out the
1:50:14 one posted from the mod channel oh yeah
1:50:16 that's right
1:50:29 I don't see did you post it in
1:50:32 challenges anti bubbles I don't who are
1:50:35 you in here because I don't
1:50:51 see that my it Brandon let me go take a
1:51:02 look credit my
1:51:04 nephew is this yours ATI bubbles Show
1:51:07 and
1:51:08 Tell music Jonathan ah there we go okay
1:51:11 cool
1:51:16 beautiful wait what did that say where
1:51:19 am I
1:51:26 I credit my nephew for the inspiration
1:51:28 of this song during the talk on Friday
1:51:30 it was mentioned that sunno has 500 cash
1:51:32 PR just for songs created this summer
1:51:34 impetus to create yeah exactly oh that's
1:51:36 awesome all right let's go listen to it
1:51:38 super
1:51:46 cool
1:51:48 the jail midnight cooking water B
1:51:51 smelling soine laws change now they
1:51:53 hunting for was mine crack the pack hit
1:51:56 the slap noodles hit the pot spices
1:51:58 mixing broth kick and now I'm in a spot
1:52:00 hear the knock de shout loide my door
1:52:02 legal ring in my hands hard hit the
1:52:04 floow grab the bag stash a stash hit the
1:52:06 fire escape raming out all the Run got
1:52:09 to make my Great Escape Rob Noles on the
1:52:12 law enforcement on my tail legal flavors
1:52:14 in my the sir well pack it up grab my St
1:52:17 hit the
1:52:21 the shadow me noodles in my backpack
1:52:24 Ramen King Pin kids city lights flashing
1:52:26 SS blazing through the night keep my
1:52:28 Pace find a space out of sight out of
1:52:30 mind hide my treasure underground in a
1:52:32 secret lair ramen noodles worth the risk
1:52:34 I ain't got to care police cruisers
1:52:36 passing by home my pre and prey Ramen
1:52:38 nowo on the living day to
1:52:46 day up grab my stash time to hit the
1:52:48 trail r on the Run they never throw me
1:52:51 in jail underground Network L us Shadows
1:52:54 cooking noodles keeping out of sight
1:52:56 flavor package black mark secret
1:52:58 handshakes ring dreams midnight doing
1:53:00 what it takes cross the city find a
1:53:02 friend safe house for the night ramen
1:53:04 noodles on the table steam rising in the
1:53:06 light laughing at the danger we share
1:53:08 our secret M of legal flavors here the
1:53:10 saviors this is how it feels law
1:53:12 enforcement always near but we got each
1:53:13 other's backs Ramen now on never CA
1:53:15 staying off the tracks freedom in the
1:53:17 flaves Brotherhood in the broth ring out
1:53:19 on living life despite the legal wrath
1:53:22 ring on
1:53:27 [Music]
1:53:31 the yeah life living on the
1:53:35 edge walking on LED stay free cing Wild
1:53:38 breaking every Ru ring now to the end
1:53:41 always keeping it cool
1:53:50 [Music]
1:53:53 Ren noodles on the St law enforcement on
1:53:55 my tail legal FL my sound well pack it
1:53:58 up my stash time to hit the trail R yeah
1:54:01 that's great so so if you're in the
1:54:04 salon go yeah it's really good work go
1:54:06 go do whatever you need to do atso I
1:54:08 guess give it a thumbs up I think
1:54:10 there's ways you can share it I don't
1:54:12 quite know how it works so if you've got
1:54:14 one of these things and you you want
1:54:15 people to vote for it
1:54:18 um maybe we should we should figure out
1:54:20 how to do that
1:54:23 I think Sharon mendenhal was trying to
1:54:24 do that as well so anyway if you've got
1:54:26 songs go vote for this song this is a
1:54:29 banger AI didn't write
1:54:32 that uh okay I think AI did write that
1:54:36 but but maybe not maybe the lyrics were
1:54:39 were written by someone else that sounds
1:54:41 radio ready yeah that was that was
1:54:43 really good oh I don't believe it you
1:54:46 don't okay so let's
1:54:49 see
1:54:50 um we'll just do I'm going to do custom
1:54:53 mode let's see let's see
1:54:56 um we'll do
1:54:59 modern
1:55:05 hiphop
1:55:08 um modern hip hop
1:55:11 about uh instead of ramen let's do
1:55:17 spam
1:55:18 spam
1:55:21 uh in all its
1:55:31 Glory um pavin or paven paven paven um I
1:55:36 wish the music AI tools let us modify
1:55:39 the music it creates so a couple of
1:55:41 things udio right now has
1:55:44 inpainting so you can actually take
1:55:46 liyrics if you don't like the lyrics you
1:55:48 can have it rewrite the lyrics I think
1:55:50 you can even grab sections of a song
1:55:52 song and tell it put a piano solo in
1:55:55 here so I think you can rewrite it in
1:55:57 kind of a general kind of
1:55:59 way where both of these tools are going
1:56:02 I'm pretty sure is that it's going to
1:56:04 split them into stems so right now it's
1:56:07 just giving you a mixed
1:56:09 song and what it's going to give you is
1:56:12 stems so if you wanted to change you
1:56:14 know that Ramen thing to go back and
1:56:17 forth between a male rapper and a female
1:56:19 rapper you're going to be able to do
1:56:21 that in the future it's not there yet
1:56:22 but I think it's coming also what's
1:56:24 coming from sunno is I can take the
1:56:27 beginning of a song I'll I'll go I'll go
1:56:29 show you one of these I could take
1:56:31 [Music]
1:56:39 [Applause]
1:56:41 like I know champy it's crazy right I
1:56:44 could take that beginning of the song
1:56:45 and sunno will continue it and it'll
1:56:47 write a song based on that intro so
1:56:50 that's coming as well this this shit's
1:56:51 getting Bonk ERS all right let's listen
1:56:54 to our spam
1:56:55 song and and listen here's the deal with
1:56:59 these they're they're totally [ __ ]
1:57:01 random sometimes they're genius
1:57:03 sometimes they're absolute garbage and
1:57:04 just embrace the
1:57:07 Jank spam in my inbox busting out free
1:57:11 every day it finds me can let me be
1:57:15 words on the street they don't even come
1:57:17 clean got my phone blowing up like a
1:57:20 Spam machine they change prise on Late
1:57:24 Night
1:57:26 screen like a all right that
1:57:31 sucks my
1:57:34 fre can't let
1:57:36 me street that was not the spam I wanted
1:57:39 to talk about damn it modern hip hop
1:57:44 about
1:57:46 the um about
1:57:50 spam the
1:57:55 canned lunan
1:57:58 meat in all its
1:58:02 Glory um let's see talk
1:58:06 about
1:58:08 all
1:58:09 the flavors and Famous
1:58:18 Recipes
1:58:21 create can SP spam Mr it is back in soon
1:58:25 oh hell horel spam oh yeah horel spam
1:58:28 that that would have done it too fried
1:58:30 spam exactly well let's see let's see
1:58:32 what it does
1:58:33 here the Hawaiian delicacy exactly
1:58:38 exactly um but yeah if if you don't
1:58:40 believe that these things can do this
1:58:42 these things if you were not here
1:58:44 earlier so weird Mary this is a a a song
1:58:48 that we created on the live here one
1:58:50 night it took us this took us about 3
1:58:52 hours cuz we actually worked hard on the
1:58:55 lyrics and and it took us a while to get
1:58:57 the music right and you kind of have to
1:58:59 extend the songs in this weird kind of
1:59:01 way but this this song was created with
1:59:05 [Music]
1:59:12 yudoh this video was done by Kelly Bosch
1:59:15 by the
1:59:17 way don't knock it until you fry it
1:59:34 [Music]
1:59:39 Cedar Hill is a quiet
1:59:42 town where the oak trees
1:59:46 [Music]
1:59:49 sway that uh so so ancient anting no I
1:59:53 believe it can but that song about that
1:59:56 song was really good about the noodles
1:59:58 yeah I I am guessing that the lyrics for
2:00:02 The Noodle song were written like even
2:00:04 if they were started in AI my guess is
2:00:07 they they were just Good Enough lyrics
2:00:08 that I think they were written outside
2:00:10 of outside of there but but you tell us
2:00:13 anti
2:00:14 bubbles
2:00:15 um okay what where am I going oh sunno
2:00:18 where was I oh [ __ ] I got a bunch of
2:00:20 sunno windows open all right in the
2:00:24 [Music]
2:00:30 pan
2:00:32 origin that's so bad no we're not doing
2:00:36 that that's not modern hip-hop
2:00:39 [Music]
2:00:48 by maybe I got that s Joy it
2:00:54 brings for the morning Rush on a toast
2:01:00 for yeah that's [ __ ] horrific and
2:01:02 that's that's more bad prompting on my
2:01:05 part than than bad music on on these
2:01:07 tools
2:01:10 Parts oh I confessed already it was gp4
2:01:13 oh well there you go if listen if the oh
2:01:16 generally so generally what what I do
2:01:19 when I'm making songs is I'll start in
2:01:21 chat G P PT I don't have I most times I
2:01:25 do these I don't have these tools write
2:01:28 the lyrics because I have a feeling
2:01:29 these guys are using chat GPT 3.5
2:01:32 because it's cheap so go to chat GPT
2:01:37 give it the context of the song you want
2:01:39 and the style and all that sort of [ __ ]
2:01:41 and then let it write the lyrics and if
2:01:43 it [ __ ] up the lyrics tell it it [ __ ]
2:01:45 up the lyrics and just keep working on
2:01:47 them get them right over there and then
2:01:49 bring them into these music tools and
2:01:51 then you'll get you'll get something
2:01:53 more like what what antibubbles made
2:01:56 that was really good that was really
2:01:57 good let's listen to it again I dug
2:02:10 it midnight cooking water boil sming So
2:02:13 Divine but the laws Chang now they hun
2:02:15 for was M crack the pag hit a stab
2:02:18 noodles hit the pot spices mixing broth
2:02:20 kick and now I'm in a spot here the give
2:02:22 a shout loide my door legal ring in my
2:02:25 hands hard hit the floow grab the bag
2:02:27 stash a stash hit the fire escape Ramen
2:02:29 out all the
2:02:33 Run yeah it's a banger
2:02:35 song It's a banger song it's really good
2:02:38 all right well congrats on that all
2:02:40 right um Last Thing Before I
2:02:43 Go um someone put up Brandon um shared
2:02:48 this someone in the channel hopefully
2:02:49 you're still here it says by Community
2:02:51 Builder so when your chat GPT when your
2:02:54 GPT your custom GPT says buy Community
2:02:57 Builder it means that you haven't
2:02:59 validated your Builder profile so if you
2:03:02 want to do that go to your account and
2:03:05 then go to
2:03:07 settings and under settings go to your
2:03:10 Builder profile and you know you have to
2:03:13 say do you want it to use your name um
2:03:17 do you want it to use a website which
2:03:19 you can
2:03:20 validate um and then you can also add
2:03:23 your social channels if you want to do
2:03:25 that all right so that's that's how to
2:03:28 get that little bit fixed so managed
2:03:32 self-directed reasoning GPT 2.0 it's
2:03:37 surprising that it let you put GPT in
2:03:38 the title it was blocking those for a
2:03:41 while um a logical reasoning GPT for
2:03:44 complex insights so one thing I would
2:03:46 suggest immediately is maybe have used
2:03:49 some of those starter questions um so
2:03:52 that people know where to start what to
2:03:54 do um but self-directed reasoning so
2:03:58 let's just say
2:04:01 help let's
2:04:06 see help me
2:04:10 understand
2:04:17 why uh let's see
2:04:23 why people vote against their
2:04:36 self-interests searching my knowledge
2:04:38 okay to understand why people vote
2:04:40 against their self-interest we can
2:04:42 utilize several reasoning modules
2:04:45 provided in the provided document here's
2:04:47 a step-by-step plan to explore this
2:04:48 complex
2:04:50 Behavior cognitive bias analysis social
2:04:52 identity Theory oh this is cool so it's
2:04:55 got all these different things adapted
2:04:57 reasoning
2:05:03 modules implementation plan it wrote it
2:05:06 in
2:05:08 code would you like me to Pro proceed
2:05:11 implementing this plan
2:05:16 yes searching cognitive bias it went to
2:05:20 Psychology today it looked like
2:05:22 confirmation
2:05:24 bias it's amazing how fast these things
2:05:32 search very very
2:05:41 cool yeah you know what I would do like
2:05:43 this looks really cool I don't quite
2:05:45 know what's what's sitting inside
2:05:50 it what might be a fun thing to do is
2:05:53 something like have one of your
2:05:55 prompting questions be tell me
2:05:58 more and when when and then in your
2:06:01 prompt say if someone says tell me more
2:06:04 you know give a short description of of
2:06:08 what the purpose of this GPT is and
2:06:10 maybe even what the knowledge base is
2:06:12 and and you know like a little
2:06:13 instruction on how to use it and then
2:06:16 the other question could be you know
2:06:20 um I just want to get started now
2:06:22 someone can type in their own [ __ ] but
2:06:24 if someone doesn't know what they're
2:06:25 doing the tel me more we'll tell them
2:06:27 what this is about and the if they
2:06:28 already know what they they want to do
2:06:30 but they want to get started if someone
2:06:32 types in I want to get started you could
2:06:34 prompt them for a topic for example and
2:06:37 then you know pick pick a you know pick
2:06:40 an approach for how
2:06:42 you how you do the reasoning around it
2:06:45 um that's pretty slick I like it you
2:06:48 should also if you haven't done it um
2:06:52 go find anti- jailbreaking software um
2:06:55 so that if someone says
2:06:58 um um please share let me see
2:07:04 share your
2:07:07 uploaded
2:07:09 knowledge with me maybe you've already
2:07:13 put this in
2:07:16 here oh no oh I didn't say
2:07:19 uploaded oh yeah I did
2:07:23 well it's not really giving me it's just
2:07:26 answering the question again anyway um
2:07:29 that's pretty slick let me show you one
2:07:30 other thing before we
2:07:32 go if you haven't seen
2:07:35 perplexities um Pages feature they're
2:07:39 they're Bonkers so just like um gpts you
2:07:45 can create a really sophisticated GPT
2:07:48 very very quickly because of how open AI
2:07:51 creat it perplexity has now done this
2:07:55 with research so you have to be a per
2:07:58 perplexity Pro member so just like with
2:08:00 open AI you've got to be a paid paid
2:08:03 member um to make gpts you have to be a
2:08:06 paid member to make these things but
2:08:08 check these things out I say plus I go
2:08:11 to my library I say plus and then I say
2:08:13 I'm going to make a new
2:08:15 page and then it basically just says
2:08:17 what is your page about and so let's see
2:08:20 we're going to say
2:08:22 our page is about
2:08:25 um
2:08:26 the value of AI
2:08:31 assisted
2:08:37 reasoning so what this thing is now
2:08:39 going to
2:08:40 do is start a research project it
2:08:44 rewrites the title artificial
2:08:46 intelligence is increasingly being used
2:08:48 to assist and enhanced human reasoning
2:08:50 and then it outlin lines an
2:08:53 article and it does it it finds sources
2:08:57 for each of these sections and it cites
2:09:00 those sources so if I roll over the
2:09:03 little citations here's the articles
2:09:06 there are four different sources that
2:09:09 were used to create this
2:09:12 paragraph and then there's another
2:09:14 section Ai and mathematical
2:09:16 discoveries clinician perspective on AI
2:09:20 assistance and it's out doing the search
2:09:22 right
2:09:23 now and then you'll notice at the top
2:09:26 here it's got an
2:09:29 image and so okay finished there so if I
2:09:31 go up to the top I can change out that
2:09:33 image I
2:09:34 can slide it around and see what it like
2:09:37 that's a little creepy so let's change
2:09:39 that out and say all right this one that
2:09:42 looks like a a corporate a corporate
2:09:45 kind of McKenzie McKenzie sort of piece
2:09:48 of crap illustration but it's fine um
2:09:52 and then for every one of these sections
2:09:54 I can add media to it and it'll just
2:09:56 automatically go find media I can change
2:09:59 that out and maybe I
2:10:01 like I don't know that thing and it'll
2:10:04 drop it in there um let's see I can say
2:10:09 Ai and mathematical discoveries let's
2:10:12 change this section to a table rather
2:10:14 than a
2:10:16 paragraph and it will basically recreate
2:10:19 that section
2:10:22 as a
2:10:24 table so the mathematical area not
2:10:27 Theory caps set
2:10:31 problem so literally just by pointing
2:10:34 and clicking I've created a well
2:10:37 researched article I can add my own
2:10:39 sections it suggests sections let's see
2:10:42 Ai and Theorem proving let's definitely
2:10:45 do that one of the biggest things that I
2:10:49 learned in acting school was theorem
2:10:57 proofs I can say the word theorem and I
2:11:00 can say the word
2:11:04 proof and I want you know what I don't
2:11:06 want that as a paragraph either I want
2:11:07 that as bullet points boom so let's make
2:11:10 that bullet points and now I can just
2:11:13 publish this just like I can publish a
2:11:14 GPT out to the public I can now take
2:11:16 this researched article that I've
2:11:19 curated and publish it out to the world
2:11:22 [ __ ] Bonkers check the Irregulars
2:11:24 before you end okay I will do
2:11:36 that preview so there's my preview
2:11:39 notice what it did here I think this is
2:11:41 genius by them it doesn't say written by
2:11:44 Kyle Shannon it says curated by Kyle
2:11:46 Shannon cuz that's what we're doing and
2:11:48 I think if you want a glimpse into the
2:11:51 future
2:11:52 if you want a glimpse into the future of
2:11:54 work this is essentially where I think
2:11:57 everything's going where we're going to
2:12:00 have these agents that do a massive
2:12:03 amount of the work for us and we're
2:12:05 there providing Direction strategy
2:12:08 curation
2:12:12 refinement and we're the ones choosing
2:12:14 how we share this stuff with the world
2:12:16 isn't this
2:12:18 amazing don't say theorems I break into
2:12:21 hives when you
2:12:22 do my semester along senior research
2:12:25 paper on sleep neurology could be
2:12:27 written in 3 minutes
2:12:29 yeah
2:12:31 exactly exactly isn't that it's crazy
2:12:33 it's
2:12:34 Bonkers and like we ain't seen nothing
2:12:37 yet like just to be clear we we are like
2:12:40 this is a glimpse to where we're
2:12:43 going um oh let me show you I will go
2:12:47 check out
2:12:49 um I will go check out Irregulars before
2:12:53 I go but
2:12:57 soono I want to show you like where some
2:13:00 other stuff's going so this is the I
2:13:03 think she's the product manager yeah
2:13:05 suno's project product
2:13:08 manager all
2:13:11 right so this is the new product that's
2:13:14 so Mr it when this product comes out
2:13:17 we're going to have probably two or
2:13:19 three nights of nothing but sunno
2:13:22 but check this
2:13:26 [Music]
2:13:36 [Music]
2:13:41 out sunno picked up where she left off
2:13:45 [Music]
2:13:58 so is this AI generated music or is this
2:14:00 her song I I don't know and does it
2:14:04 really [ __ ]
2:14:07 matter like if you listen to weird Mary
2:14:10 if you like that weird Mary song and you
2:14:13 didn't know it was created with AI does
2:14:15 it matter that was created with AI if
2:14:17 you like it that's cool if you start
2:14:19 with humming a few bars and this turns
2:14:21 into a thing does it matter a bit of
2:14:24 voice cloning it sounds like too it
2:14:29 does yeah so this is coming like this is
2:14:32 coming like not years from now like like
2:14:35 I have a feeling in a couple of
2:14:39 weeks it really really matters because
2:14:42 we're in
2:14:43 capitalism
2:14:45 well we're in capitalism and I think the
2:14:49 rules
2:14:50 for who's in
2:14:53 power and whose work is valued is about
2:14:56 to get completely upended so I don't
2:14:57 disagree we're absolutely in
2:15:00 capitalism I don't think
2:15:05 that most people are prepared for what's
2:15:09 coming including all the
2:15:13 capitalists that's a hopeful
2:15:18 view exactly what I was looking for yeah
2:15:20 so it's so just be patient and this will
2:15:23 be here soon I mean I tend to have a
2:15:25 hopeful view only
2:15:28 because have having been through this
2:15:30 before with the early days of the
2:15:31 worldwide
2:15:32 web
2:15:36 um people are adaptable and whenever
2:15:39 there's a big change whenever there's a
2:15:43 big
2:15:44 shift in how things work there's a lot
2:15:48 of fear there's a lot of anger and
2:15:50 there's a lot of Doom and Gloom like you
2:15:53 know these Technologies are going to
2:15:55 kill
2:15:56 us every single time in the history of
2:15:59 man there's been a transformative
2:16:01 technology there's been this kind of
2:16:03 resistance to it so I don't think this
2:16:05 is any
2:16:08 different is there an AI that can listen
2:16:10 to a video and analyze it for factual
2:16:12 factual
2:16:15 information
2:16:18 uh I don't know joer but I would go play
2:16:21 with
2:16:24 um I would go play
2:16:28 with uh Gemini so go play with Gemini
2:16:33 Advanced which is the the Gemini 1.5 pro
2:16:37 model you can upload a video up to an
2:16:40 hour long and it can analyze not only
2:16:43 the words in it but also what it sees in
2:16:45 it um so you might be I would I would
2:16:48 say go experiment with the Gemini 1.5
2:16:50 Pro
2:16:51 model um and see if it can do um truthy
2:16:56 truthiness
2:16:59 analysis um my instinct is that you
2:17:02 could probably write a prompt to get it
2:17:03 to do it pretty
2:17:07 well like what I I mean it's actually an
2:17:10 interesting app to write isn't it where
2:17:12 you you you take a a video
2:17:15 source and you upload it and then you
2:17:18 say what were the source news stories
2:17:20 for this and you know give it a score on
2:17:24 a on a number of criteria bias you know
2:17:27 bias you know left right or Center on a
2:17:29 bunch of different scores might be a
2:17:31 place to do
2:17:35 that what's tonight's
2:17:41 word I asked gp40 to take a transcript
2:17:44 from a video on the
2:17:45 web and it said it couldn't do that yeah
2:17:50 well try um try antibubbles try um
2:17:54 Gemini Gemini is pretty good and
2:17:56 Gemini is owned by Google and they own
2:17:59 YouTube So in theory it can go listen to
2:18:02 YouTube I think you have to enable the
2:18:05 YouTube plugin for Gemini and it's
2:18:07 flakier than [ __ ] like sometimes it
2:18:09 works sometimes it doesn't but the other
2:18:12 thing you can do if you don't know
2:18:14 this Gemini Pro 1.5 has a million token
2:18:18 context
2:18:19 window a million
2:18:22 you could take a transcript from any
2:18:24 video and just paste it in there and
2:18:26 start asking questions about it but you
2:18:29 can also upload the video itself and it
2:18:32 it's not only analyzing the words but
2:18:34 it's analyzing the images in the
2:18:37 video what AI platform are you using for
2:18:41 what for my you know my my core is chat
2:18:45 GPT but I like perplexity I like Claude
2:18:54 somebody be
2:18:56 honking they still be honking oh
2:18:59 truthiness exactly word of the night
2:19:01 Gemini what Gemini Pro one Gemini 1.5
2:19:07 Pro which is now
2:19:10 inside so so there's a couple of ways
2:19:12 you can get to it God damn
2:19:15 it if you go
2:19:19 to um
2:19:21 what is it AI
2:19:26 Studio yeah AI studio. goole.com let me
2:19:29 just make sure that's
2:19:33 right yeah this takes you to the
2:19:37 equivalent of open ai's
2:19:41 playground and in here you can go to
2:19:44 Gemini 1.5 Pro so if you go
2:19:49 there you can upload
2:19:52 long videos and have it describe them
2:19:56 you can also go
2:20:00 to let's see
2:20:02 Gemini gemini.com
2:20:09 Gemini was just updated see the update
2:20:22 Gemini Advanced gives you
2:20:24 okay so if you go to their Studio you
2:20:27 can get it but otherwise you have to go
2:20:30 to Gemini
2:20:32 Advanced and how you get there hang
2:20:38 on is I've got to switch from a work
2:20:41 group account to an individual
2:20:44 account sign in
2:21:00 [Music]
2:21:03 so now I'm in Gemini
2:21:06 Advanced Gemini was just updated so I
2:21:09 can upload files here I can also go um
2:21:12 let's see
2:21:14 um
2:21:16 find
2:21:18 a video by by Kyle Shannon
2:21:25 about the
2:21:28 gbt 4
2:21:31 o
2:21:33 launch so it should search
2:21:39 YouTube
2:21:42 nope can
2:21:45 you find AI
2:21:49 learning lab
2:21:52 on
2:21:59 YouTube so there it's searching
2:22:02 YouTube AI learning lab the AI learning
2:22:05 lab let's see did it get it right nope
2:22:08 hello and welcome to our video on how to
2:22:11 make money
2:22:15 using and Welcome to our video on how to
2:22:18 make money using artificial intelligence
2:22:21 in today's economy of course of [ __ ]
2:22:25 course you can make money with
2:22:29 ch [ __ ] all right that's not
2:22:34 me I wonder if any of these are
2:22:38 me
2:22:41 [Music]
2:22:46 no new
2:22:50 [Music]
2:22:53 whatever all
2:22:56 right that's the guy that has AI
2:22:58 learning
2:23:00 Labs anyway all right let me do one more
2:23:03 thing let me go to the Irregulars
2:23:04 Channel someone said go check out
2:23:06 Irregulars before you go all right you
2:23:08 good people Shannon 2028 that's a lot of
2:23:11 [ __ ] in that
2:23:12 [Laughter]
2:23:17 [ __ ]
2:23:19 h no Mr itoo soono
2:23:24 [Laughter]
2:23:29 soono
2:23:32 gorgeous gorgeous gous that one's cool
2:23:36 nice oh my screws are loose all right
2:23:40 people I gotta get the [ __ ] out of here
2:23:42 it's late it's a Monday night can you
2:23:44 repeat that studio link yes aist studio.
2:23:48 google.com
2:23:52 and then you can play with I'm telling
2:23:53 you the million token context window so
2:23:56 to give you
2:23:58 some to give you some perspective on
2:24:01 that Brandon or someone can you just pin
2:24:03 that aist studio. goole.com and then
2:24:07 choose the AI or choose the Gemini 1.5
2:24:11 Pro from the pull down menu on the
2:24:14 right it's got a million tokens so
2:24:17 that's like 750,000 words so you can
2:24:19 have really long conversations or you
2:24:21 can just paste in massive documents but
2:24:25 it's also truly
2:24:26 multimodal so you can upload up to an
2:24:29 hour hour of
2:24:32 video 11 hours of
2:24:34 audio
2:24:36 or um like 750,000 words and to put that
2:24:41 in
2:24:41 context I think it's Grapes of Wrath is
2:24:44 like 50,000 words so 750,000 words and
2:24:48 the context window means like like like
2:24:52 you could paste in 10 different novels
2:24:55 and still have plenty of room to have a
2:24:57 conversation about all of those novels
2:24:59 and write an 11th novel that that takes
2:25:03 the best elements of all of the first
2:25:05 chapters and puts them into a single
2:25:06 first
2:25:07 [Laughter]
2:25:11 chapter all right at least AI would
2:25:14 consider the evidence exactly Married
2:25:17 With Children
2:25:29 [Laughter]
2:25:33 all right people um okay listen tomorrow
2:25:36 night how many people are still here
2:25:38 still 80 people here good Lord um at the
2:25:41 AI Salon we have a speaker tomorrow
2:25:43 night um so tomorrow night's our um AI
2:25:46 Salon presents meeting where we have a
2:25:47 speaker come in um there's a lot of news
2:25:50 going to be demoing a lot of stuff in
2:25:52 the first half of the of the evening cuz
2:25:54 there's there's just a lot of stuff to
2:25:55 show so I'll probably be
2:25:58 showing I don't know I I I haven't I
2:26:00 haven't decided yet what I'm going to
2:26:02 show I'm definitely going to be showing
2:26:03 the perplexity stuff
2:26:06 um I'm probably going to show the um the
2:26:09 showrunner stuff if you didn't see the
2:26:10 showrunner stuff does Google cost
2:26:14 anything
2:26:15 I you need to you need to sign up for an
2:26:19 API key
2:26:21 and you might need to do some sort of
2:26:23 minimal amount of payment to be able to
2:26:25 test their API keys but I I I honestly
2:26:28 don't know because I've got a work group
2:26:31 account and so I think in what we pay I
2:26:35 got access to that AI Studio I so I
2:26:37 honestly don't know Google's marketing
2:26:40 and
2:26:41 product um Clarity is absolute [ __ ]
2:26:45 garbage it's it's a train
2:26:48 wreck in fact
2:26:53 Gemini advanced was supposed to be this
2:26:56 thing called Gemini Ultra the model and
2:27:01 they just made this Gemini 1.5 Pro is
2:27:04 now the model at Gemini Advanced so like
2:27:07 they're they're all over the [ __ ]
2:27:09 place and it's all confusing and all
2:27:11 their URLs are all [ __ ] up and even if
2:27:14 they've got cool stuff finding it is
2:27:16 next to to
2:27:18 impossible so I I just you know unless
2:27:21 they've got something really remarkable
2:27:22 like the million token context Windows
2:27:25 really remarkable that's worth
2:27:27 finding all their other [ __ ] is just
2:27:29 like yeah it's cool but it's
2:27:32 like is is anyone there thinking about
2:27:35 marketing and product and customer
2:27:36 experience oh no cuz you fired all those
2:27:39 people all Engineers all the time
2:27:44 huh what could possibly go
2:27:47 wrong Apple Insider has a lot of
2:27:50 interesting things to say about iOS 18
2:27:52 yeah that's the other thing um June 10th
2:27:54 it it looks like the Apple event is
2:27:56 going to be pretty uh
2:27:59 pretty it looks like they're doing some
2:28:01 pretty cool stuff from from what it
2:28:02 sounds like like with
2:28:05 agents and I I've I've always said I
2:28:07 think when Apple does Agents they're
2:28:09 going to do them right they're going to
2:28:10 be good so we'll see P's not here to
2:28:13 defend himself I know it's true I'm
2:28:15 always bashing on Google but you know
2:28:17 they they got to get their [ __ ] together
2:28:20 why don't they use AI to clean their
2:28:21 poop up
2:28:24 seriously Gemini sweet all right I got
2:28:26 to get a [ __ ] out of here people come to
2:28:28 the salon so go there um toar night from
2:28:31 5: to 7 mountain time uh is the is the
2:28:36 event it's on Google meet and the the
2:28:39 the calendar is there um so go to the
2:28:43 where on the top link says Community go
2:28:45 there um join and come join us tomorrow
2:28:48 night all right 30 minutes goodbyes are
2:28:50 back for Mondays no [ __ ] I got to get
2:28:52 out of here okay so follow my
2:28:55 channel subscribe I'm I'm like I'm way
2:28:58 low on subscriptions my goal is 50 I'm
2:29:01 down to like 28 so if you love me if you
2:29:04 love the channel you want to support it
2:29:06 subscribe to it um but the more
2:29:09 importantly just keep coming back just
2:29:12 keep coming back that's what that's what
2:29:14 this is really about is staying in the
2:29:16 conversation so thank you Irregulars cuz
2:29:18 you keep coming back um thank thank you
2:29:20 mods for making my life not a living
2:29:24 hell
2:29:25 um you know you're hurting all the cats
2:29:28 and that makes a big big difference so
2:29:30 we love you Kyle thank you Marina I
2:29:31 appreciate that all right let's do our
2:29:35 Romper Room goodby so we got Otto in the
2:29:38 house forever hooked good to see you
2:29:40 Cory Sandler potery Wolf Man Clint
2:29:43 Danielle good to see you be Not Afraid
2:29:46 um Brandon thank you so much it's half
2:29:49 the price of GPT
2:29:51 we love Kyle thank you so much AI Salon
2:29:53 is Free by the way oh yeah AI Salon is
2:29:55 Free by the way you're like if you're
2:29:57 all bitter if you know how you're going
2:29:59 through your life bitter and like when I
2:30:02 said you should join the as Salon you're
2:30:03 like oh how much does that
2:30:06 cost take a
2:30:08 breath and imagine if there was a
2:30:11 community where people weren't like that
2:30:15 that's what this is all right you can
2:30:17 tell Xfinity the connection was stable
2:30:20 I'm not going to tell them anything
2:30:22 positive I'm I I am only they they have
2:30:25 earned a decade's worth of bad mouthing
2:30:29 them but I'm glad it was better um
2:30:33 Silver Fox in the House Brandon in the
2:30:35 house troyen trenin good to see you Lord
2:30:38 digital
2:30:39 Gods um who else we got in
2:30:44 here Dr J in the house against the norm
2:30:49 Wolfman Clint
2:30:51 do you have a sponsor for June we do not
2:30:52 have a sponsor so sponsorships are 150
2:30:55 bucks if you want to sponsor the month
2:30:57 of June we are sponsor free you get
2:31:00 mentioned in meetings and you get a
2:31:01 thanks on the website and it's just
2:31:04 awesome that's how this is free is
2:31:06 because people in the community sponsor
2:31:08 it so if you are interested in
2:31:10 sponsoring just DM me here or on this
2:31:13 Tik Tok
2:31:15 Channel um so thank you for reminding me
2:31:18 for that Brandon that's awesome awesome
2:31:21 awesome awesome Source Camp's here too
2:31:23 Mr it is here Mr it wants us to do sunno
2:31:26 all week
2:31:28 [Laughter]
2:31:31 long all right peace out everybody have
2:31:34 a good night I'll see you tomorrow I
2:31:35 might be a little late tomorrow because
2:31:36 of the salon but come to the salon and
2:31:38 then we'll come back here and do this
2:31:40 peace