aiLL 9/4/23 - Generative AI: Our Chance to Reinvent Ourselves

Description

Chapters

Transcript

0:03 foreign
0:08 [Music]
0:19 [Music]
0:20 alrighty then I like taters
0:25 um all right then
0:26 um
0:29 they call it a sling blade um
0:32 welcome everybody welcome welcome
0:34 welcome welcome welcome
0:36 happy Labor Day
0:39 Robert Rossi welcome in
0:43 right out of the gate with a donut I
0:45 like it thank you thank you thank you
0:48 thank you shaking
0:51 welcome to the AI learning lab we'll get
0:54 this show on the road
0:58 I got myself some uh
1:02 Stumptown cold brew coffee my wife
1:05 bought it and she goes you want some of
1:07 this I'm like all right yeah
1:11 it's all right it's pretty good
1:13 pretty good A little little peppy it's a
1:17 little peppy put a little pep in his
1:19 step
1:19 welcome everybody my name is Kyle
1:21 Shannon this AI learning lab will get
1:23 rolling here in a bit let's let some
1:25 folks get in
1:26 and uh happy Sunday night on the open
1:29 window here doing a fresh air going it's
1:32 nice and cool out
1:34 this evening in general Denver evenings
1:37 are cool
1:39 although this summer that's not been the
1:41 case but tonight it is it's nice and
1:43 cool
1:44 got some Us open on in the background
1:47 and we'll get going here we'll get going
1:50 we'll get rocking and rolling
1:54 we got we got all sorts of exciting
1:55 stuff the scammer is back which is a
1:58 little demoralizing my Tick Tock shop
2:01 that I set up last night was functional
2:05 but
2:06 apparently I have to talk about and show
2:08 products the whole time I'm live so
2:10 that's shut down now and then there were
2:13 two orders on it and Tick Tock screwed
2:15 them both UPS so I think I gotta I gotta
2:19 return I gotta cancel those orders and
2:22 then if those people still want them
2:23 I'll do it on printful or something like
2:24 that but you know what are you gonna do
2:27 and then what else did I do I spent
2:30 two hours on the phone with a t and
2:32 treat a t trying to get some stupid
2:35 account thing worked out hey Ted what's
2:37 happening uh so so it was a day of of uh
2:43 technology and systems disappointments
2:47 complications and disappointments which
2:49 one of the things I think AI is going to
2:51 help us with is all these overly
2:52 complicated systems that are designed to
2:54 confuse the [ __ ] out of us are gonna get
2:57 simpler when we can just say hey solve
2:59 this problem for me and it goes off and
3:01 does it I am looking very forward to
3:04 that day
3:06 at work in life with insurance companies
3:10 with loan with all of it all of it none
3:13 of it should be this complicated right
3:15 none of it none of it should be this
3:17 complicated
3:21 error messages like something went wrong
3:24 you know no we have a I know you can
3:28 figure out what went wrong have it read
3:29 the code have it read whatever
3:31 exceptioned
3:34 and say what went wrong is you were
3:36 logged into this account but not that
3:38 account and therefore our system didn't
3:40 know how to handle that so if you go log
3:42 out of that one and into this one it'll
3:44 probably work you know right am I wrong
3:46 no I'm not wrong
3:48 [ __ ] shouldn't be this complicated
3:50 anyway lunish sticks here okay Farmers
3:53 here welcome Nancy's here
3:55 welcome welcome welcome
3:58 we'll get rolling here shortly we'll get
4:01 doing it we'll get to it
4:03 not good on that not done that not gonna
4:06 die not gonna die
4:10 gotta
4:12 uh cold brew coffee
4:14 Rocky Mountain filtered tap water
4:17 you can't buy that in a bottle well you
4:20 could I could sell it to you if you want
4:21 it I'll ship it on out
4:25 right out of the fridge door
4:30 oh and by the way if you didn't know
4:35 uh we'll get started here in a sec let's
4:37 let some some of the folks get in here
4:39 if you do what if you would do me a
4:41 favor it looks like uh Tick Tock might
4:42 be slow rolling the live tonight
4:45 um I guess I'll invite some people too
4:48 um
4:49 if you would share the uh share the live
4:52 if you're willing to make yourself look
4:54 weird in front of your friends
4:56 and let me invite some folks here
5:00 okay Farmers here
5:03 and you ran do that do that do that
5:08 do that do that
5:10 Luna sticks here
5:17 um
5:22 do two more oh Tobias is covered
5:27 sand all right there we go let's send
5:30 out a few more all right welcome
5:32 everybody this is the AI learning lab as
5:34 evidenced by the
5:36 lovely designed little poster here
5:39 Kyle did you did you get a degree in
5:41 typography why no
5:44 if you didn't know it there is now a
5:46 image generation tool that does text did
5:50 you know that did you know it did you
5:51 know it
5:52 I'll I'll talk about that in a second
5:56 um let's get rolling here so
6:00 um welcome to the AI learning lab my
6:01 name is Kyle Shannon I'm an entrepreneur
6:02 I'm a uh co-founder and CEO of a company
6:06 called storybine which is an 11 year old
6:08 company that does automated video
6:10 storytelling
6:11 and we do a lot of work in healthcare
6:13 and Pharma and rocking and rolling oddly
6:16 enough we don't use a ton of AI in Auto
6:19 automatically Auto magically editing our
6:21 videos
6:22 um
6:23 but we're rolling a lot of AI into it
6:26 um I was an entrepreneur entrepreneur in
6:28 the mid 90s in the early days of the
6:29 world wide web and the parallels between
6:31 the early days of the web and the early
6:34 days of what we're going through right
6:35 now like in the last nine months with
6:38 this generative AI [ __ ] are crazy crazy
6:42 it's cool it's really cool so anyways I
6:45 like talking about that stuff if you
6:47 have questions about AI or comments or
6:49 thoughts or just whatever pop them in
6:51 the comments below I'll do my bus best
6:54 to answer them I try to be honest I try
6:58 to be fun I make funny little voices
7:00 I'll occasionally cuss when I get all
7:02 fired up I just lose my mind and I cuss
7:04 so if you don't like to cussing get out
7:06 just because I'll do it some some
7:09 uptight dude was in here the other day
7:10 you know you're you're cut you're uh
7:13 your swearing makes you look tacky all
7:15 right I'm tacky
7:17 I'm not the most sophisticated of the uh
7:20 of the tick talkers
7:22 foreign
7:24 ly just not very sophisticated
7:28 um okay so here's what we're gonna do if
7:30 you're new to this AI stuff
7:36 I'm gonna give you some tips so this is
7:39 like this is like my top 10 list except
7:41 just like top 11 list
7:43 and these are the big three the top
7:45 three right here
7:47 if you're brand new if you've never used
7:49 chat GPT before go there right now that
7:52 one that one it's uh chat GPT that's the
7:55 official website chat.openai.com just
7:57 start playing if you don't know what to
7:59 say to it if it's just like a Google
8:00 search box but it's not a search engine
8:03 it is a just start having a conversation
8:06 with it just start asking it to do
8:07 things for you answering questions for
8:09 you or solving problems for you if you
8:11 don't know what to do with it go to
8:13 prompts.chat that thing below there and
8:16 it will teach you how to talk to these
8:17 things
8:18 and and you'll start to get a real good
8:20 sense that document gives you a very
8:22 good sense of how deep these things are
8:25 and
8:26 what they can possibly do for you it's
8:28 it's pretty mind-blowing so so that's a
8:30 good one Claude and Bard bard.google.com
8:33 and cloud.ai are those are the ones that
8:36 I'm considering right now kind of the
8:37 top three of my go-to's
8:40 they all kind of do slightly different
8:42 things like bards connected to the
8:43 internet Claude lets you put in really
8:45 big prompts and chat GPT is just best
8:48 and it has this thing called Advanced
8:50 data analytics which is incredibly
8:51 powerful so that's that the next four on
8:55 the list
8:56 are Bing
8:58 perplexity.ai pi and po
9:02 um piinpo sounds like some bad
9:05 children's
9:07 nursery rhyme that was pie and that was
9:10 pole and Poe said to pie and pie said to
9:13 Poe
9:15 um were it not for a single letter but
9:17 but anyway Bing is Microsoft's version
9:20 of chat GPT it's it's built on the GPT
9:23 for large language model it's connected
9:26 to the internet it can make images
9:28 um the reason it's on in the in the on
9:31 the kids table on the second list is
9:34 because it's the safety guard rails on
9:37 it are so overblown right now it makes
9:39 it kind of unusable sometimes
9:41 um lab stop perplexity uh is uh
9:45 perplexity AI is hosting llamas or metas
9:49 llama 2 models and their code llama
9:51 model so if you want to go play with
9:52 those open source models that meta put
9:54 out there without installing [ __ ] go do
9:56 it
9:58 um Pi dot AI is what we call chatty GPT
10:00 if you want to have a good conversation
10:02 with a large language model and you're
10:04 like but Kyle why would I want to do
10:05 that you should just try it Go download
10:07 the PI app Pi personal assistant search
10:10 for that you'll find it
10:12 it's from a company called inflection
10:14 and and put it in voice mode and just
10:15 start talking to it and it is not Alexa
10:18 it is not Siri it's it's it's a
10:22 conversation of a different nature and
10:24 it's connected to the internet it's
10:25 actually pretty interesting and then po
10:27 will let you make Bots and things like
10:28 that so that's that so the next pile of
10:31 [ __ ] on the list are the image tools so
10:35 there's kind of there's
10:38 there's a handful of major players mid
10:40 Journey's the one that's got the the
10:42 most sophisticated image generation
10:45 stable diffusion which is kind of
10:47 represented with these bottom two things
10:48 here is the most
10:51 um sort of technically adventurous right
10:53 it's open source the the developer
10:56 communities making all sorts of stuff
10:57 for it if you really want to geek out
10:59 with something that's stable diffusion
11:01 Leonardo is is a simple web front end on
11:04 top of stable diffusion and clip drop is
11:06 from stability AI the company behind it
11:08 and then ideogram this little sneaky
11:11 bastard they snuck out of nowhere it's
11:13 like it's like a 10 day old well the the
11:17 thing's been launched for about 10 days
11:19 ideogram and ideogram is what's doing
11:22 all this
11:24 typography
11:26 in image generation that was not done in
11:28 Photoshop or illustrator this is all
11:30 done in ideogram
11:32 so now these are obviously the selects
11:35 the ones that I got that actually it
11:37 made the words correctly it makes the
11:39 words incorrectly a lot but
11:43 still pretty cool and then prompt book
11:45 openart.ai prompt book will teach you
11:48 how to talk to these things
11:49 so you got prompts.chat for the words
11:51 and prompt book for the pictures and
11:53 then you got futurepedia.io which is
11:55 like 4 000 AI tools
11:57 and you can go you can go down many
11:59 rabbit holes so go do that
12:02 um couple other things
12:04 for those of you that have been here for
12:05 a while in the ongoing soap opera that
12:08 is the scammer he's back
12:11 um he's only copied about 20 of my
12:14 videos at this point instead of like 100
12:15 like he had before but he's back so AI
12:18 learning lab is me is this is is our
12:21 happy little house here
12:23 this AI learning lab with 2ls in in
12:26 front of learning that's the [ __ ] and
12:29 he comes into my lives he hangs out here
12:31 so if after the live tonight someone
12:34 purporting to me me reaches out to you
12:36 and says hey how's it going you're swell
12:41 it's not me
12:44 um we we can we can uh you know chat in
12:48 other ways like like on the AI Salon
12:50 Discord or whatever but anyway
12:53 um he's a jerk he's trying to get people
12:55 to buy his crypto multi-level marketing
12:57 scheme so just report him report him for
13:00 intellectual property theft report him
13:02 for uh impersonating a celebrity which
13:05 you're like but Kyle you're not a
13:06 celebrity um really how do we Define
13:08 celebrity I think I'm a celebrity
13:10 therefore yeah I'm a celebrity so
13:14 Tick Tock doesn't give us an answer for
13:16 uh he's trying to scam us so
13:19 um so anyway that's that all right the
13:22 other thing is last night I announced
13:24 that we had a tick tock shop with
13:25 t-shirts which we did but apparently if
13:29 you're gonna have products on Tick Tock
13:31 you have to talk about them and show
13:33 them and and the channel has to become
13:36 about selling [ __ ] well [ __ ] that I'm
13:39 not gonna do that I did that as kind of
13:40 a courtesy so I'll figure out a way what
13:43 what I'm going to likely do is I'll just
13:45 put up a printful shop and then I'll
13:48 stick it in my link tree so we'll we'll
13:51 put a link tree together and the shop
13:53 will be in there so anyway if if you're
13:56 one of the Irregulars there were two
13:58 people that ordered last night and Tick
13:59 Tock screwed up those orders I'm
14:01 probably going to cancel those orders so
14:03 if you were one of them
14:04 uh promptbook.com no prompt book is
14:08 openart dot AI slash promptbook
14:13 um if you're one of the two people that
14:16 ordered last night I'm probably gonna
14:18 cancel the orders because Tick Tock
14:20 screwed them up
14:21 uh and then just let me know and and
14:24 what are the shirts look like the shirts
14:26 have
14:28 oh this
14:32 image on them I'm not weird I'm
14:34 irregular so the the um
14:36 the the folks in here
14:38 the uh
14:40 the folks in here that that are in here
14:42 on a nightly basis with me they're
14:44 they're as crazy as I am many of them
14:46 are neuro spicy just like me uh and so
14:49 they've they've named themselves the
14:51 Irregulars and and we also have a joke
14:54 about turtles so so this is just a it
14:58 was just a quick Design This was done in
14:59 ideogram uh and and when I did it and
15:03 shared it in Discord someone said that's
15:05 gotta be a t-shirt so that's our first
15:07 t-shirt I'm also going to do a t-shirt
15:09 for
15:12 me and then I'll figure out if there's
15:14 other ones there might be things uh
15:16 you're not late uh Emilio's wife 8 13. I
15:19 just started and uh I've been I've been
15:21 doing just the housekeeping stuff you
15:23 know you know the you know the drill you
15:26 know here's stuff you should go check
15:27 out here's a scammer who's an [ __ ]
15:29 right all The Usual Suspects the
15:32 scammers Back The Tick Tock store is
15:34 gone because tick tocks have Butthead
15:38 and yeah I'm a little demoralized I've
15:41 had a rough day with with technological
15:44 realities
15:47 um anyway
15:48 welcome
15:49 so so what I'm gonna do if you have any
15:52 questions thoughts comments about AI you
15:55 want me to demo something I'm happy to
15:56 do that pop it in the comments below
15:59 um I will answer everything I can I try
16:02 to be as
16:05 um
16:07 as honest as I can as straight as I can
16:09 like if I don't know something I'll say
16:11 that if I do know something
16:13 I will answer you and yeah like last
16:16 night last night I felt bad it was like
16:18 it's like the first seven questions were
16:20 like oh I don't really know
16:23 it would be good if I had an answer to
16:25 that one you know that's a fantastic
16:26 question Bob uh yeah
16:29 so so hopefully tonight's a little bit
16:32 better
16:33 well we almost have 10 000 likes already
16:36 we're just getting started oh that's the
16:38 other thing if you are new here and
16:40 you're like my goodness these people
16:42 love him no no no they're just playing a
16:44 game see see they've got add too so they
16:48 get bored and and they're all just
16:50 tapping the screen I think some of them
16:51 have bought auto screen tappers from
16:53 Amazon or something like that so they're
16:55 just trying to beat their high score
16:57 from last night which I forget what we
16:59 got to it was like 75 000 or maybe it
17:01 was 70 000 something like that it was
17:03 pretty good last night the all-time
17:05 record was like a hundred and one
17:06 thousand
17:10 oh good lord good Lord good Lord all
17:13 right let's go answer some questions
17:15 shall we
17:17 bye
17:20 [Music]
17:26 life wasn't meant to be hard I know Sky
17:30 Timber what life wasn't meant to be hard
17:32 we're we're we are we have over
17:34 complicated everything
17:37 and uh
17:39 you know all of the stuff that was
17:41 supposed to make it easier and simpler
17:43 has just gotten more and more and more
17:45 complicated and I do believe that this
17:48 generative AI stuff is going to give us
17:52 the gift of not that
17:56 I think it's going to give us the gift
17:57 of Simplicity I think I the thing that
18:00 I'm the the thing that I'm most excited
18:03 about is I learn about these tools
18:06 is
18:07 that they can understand really
18:10 complicated things they can ingest
18:13 complicated documents and read them for
18:16 you and interpret them for you and
18:18 analyze them for you
18:20 they can take complicated processes and
18:23 understand them they can write code they
18:26 can read code they can document code
18:30 so yeah so what if they
18:34 make the ugly [ __ ] less ugly and what if
18:37 we got like so today
18:40 I woke up and it
18:42 it was clear that Tick Tock didn't give
18:45 a [ __ ] how I wanted to have my store
18:47 that they were just gonna ding me and
18:50 shut down the account or whatever I
18:51 don't know but I I don't care what it
18:53 however I was doing the story was not
18:55 acceptable to them then they also
18:57 screwed up the two orders that I got
18:59 then I went to printful and they
19:02 couldn't answer anything and I I just my
19:04 help thing went into some vacuum which
19:07 I'll probably hear back tomorrow
19:09 so then there was that then I got a note
19:11 from a t that said the thing that I
19:13 resolved on the phone two weeks ago is
19:15 now unresolved and they canceled my
19:17 discount that I had spent
19:20 two hours last week or two weeks ago
19:22 trying to fix so so just all day today
19:25 it's just been complicated system
19:27 meltdown
19:29 and uh so I'm I'm I'm particularly I'm a
19:34 little sensitive I do declare I'm just
19:37 I'm a little I'm a little rubbed raw if
19:40 you know what I mean I'm a little
19:42 sensitive
19:44 and I'm I would love for these things to
19:47 make life a little less shitty
19:51 dang it oh and the hair is starting to
19:54 go it's this could be a rough night
19:56 from a quaff perspective also so you
20:00 just listen it's Labor Day hopefully
20:04 you've all had I do you have barbecue on
20:07 on the labor day well I don't know what
20:08 you have we had what did we have
20:10 we had
20:12 tacos chicken and steak
20:15 tacos
20:16 so I don't know they were done on the
20:19 grill
20:20 so there you go
20:23 I'm not drunk I probably should be
20:26 Nancy hey everyone this is great thank
20:28 you
20:30 um
20:32 life wasn't meant to be hard no indeed
20:36 it wasn't
20:39 I used ideogram today in a client
20:41 newsletter very cool
20:43 all about it I meant all about it wait
20:47 all about what you meant Thelonious what
20:49 were you talking about
20:50 oh tell me
20:53 oh all about it Kyle oh you had a good
20:56 day today too Thelonious awesome
20:59 oh
21:00 just just every just just
21:03 like are companies purposefully trying
21:06 to make life harder like they are right
21:09 like
21:10 it's not a conspiracy theory they're
21:12 they're trying to make it harder like I
21:14 know insurance companies are trying to
21:16 make it harder they don't want to pay
21:18 you so they're like well if we make it
21:20 17 pages of regulations then they won't
21:23 most people won't figure it out we won't
21:25 have to pay them let's just do that
21:27 that I get
21:28 completely laudable like as a strategy
21:31 chef's kiss brilliant
21:34 [ __ ] people in their health just let's
21:36 make some money at least it's clear
21:40 but like the phone bill really
21:45 we we just that's got to be hard too
21:50 passwords
21:52 first it was captions then it's
21:55 two-factor authentication now it's
21:56 two-factor authentication with a a super
21:59 secret passcode and a click and a thing
22:01 and really
22:03 really I just steal my [ __ ] data like
22:07 here's my bank passcode here's my Social
22:09 Security the scammer already wants to do
22:11 that anyway we should just give it
22:12 wouldn't it be easier just give your
22:13 [ __ ] up
22:16 foreign
22:19 bitter you betcha
22:23 uh
22:28 uh we don't need no stinking
22:31 sophistication
22:32 [Laughter]
22:34 no this is not a sophisticated Channel
22:38 we can talk like the sophisticates but
22:40 we are not sophistic kids if you
22:41 understand what I mean welcome welcome
22:44 to the AI learning lab yes good day to
22:47 you sir good day
22:50 how do you make money with chat EBT
22:56 all right
22:57 uh I've been playing around with some of
22:59 the plugins cool let me know if what you
23:02 have oh no I'm late you're not late Dr J
23:04 we're just we're just getting started
23:05 I'm just I'm just babbling at this point
23:08 um yeah Luna stick let me know what
23:10 plugins you're playing with if you found
23:11 anything interesting failed Pixar
23:14 franchise
23:15 uh but the AI learning lab it is
23:18 definitely a failed Pixar franchise
23:20 we can go we can go to look I've got
23:23 I've got um
23:26 where's my Kyle Shannon dreams this is
23:28 all of my these are all of my uh
23:31 stable diffusion
23:35 failed Pixar versions of myself that was
23:38 a pretty fun one
23:39 if you go to Instagram Kyle Shannon
23:42 dreams or Tick Tock Kyle Shannon dreams
23:45 I I wrote little stories behind all
23:48 these
23:49 all these crazy characters that this was
23:51 done in dreambooth and stable diffusion
23:53 like a year ago
23:55 yeah it was you know what's funny it was
23:57 it was about this time last year
23:59 I was playing with a project called
24:01 pixel mind and I just heard about dream
24:04 Booth or like right around this time
24:05 dream Booth came out
24:07 and I'm like I gotta get stable
24:09 diffusion up and running I wanna I wanna
24:11 see what this dream Booth thing's all
24:12 about
24:13 I failed nine times
24:16 nine different sessions where I sat down
24:20 went to Reddit
24:22 found a Google collab notebook tried to
24:26 run it didn't understand what the [ __ ] I
24:28 was doing didn't quite understand how to
24:30 connect it to Google Drive
24:32 tried to run it it errored out on me
24:35 nine [ __ ] times it took me like a
24:38 month and a half of just you know trying
24:41 it
24:42 slamming my face against the desk and
24:46 then going [ __ ] it and then like
24:49 I don't know a week later I'd go yeah
24:51 maybe I should try that again I'd try it
24:53 [ __ ] nine times
24:56 and then I finally got it and when I got
24:58 it then I just spent many many
25:02 weeks making these images where where
25:05 what was cool about sort of stumbling on
25:08 this thing where I could upload images
25:10 of myself that one's [ __ ] hilarious
25:13 um
25:14 and
25:17 like seeing these images that were kind
25:19 of me but not me it was like
25:22 it was like living vicariously through
25:25 through these fictional
25:27 versions of myself
25:30 and I thought huh like that's kind of
25:33 cool like I want to know the stories of
25:36 these people like what choices did they
25:38 make in their life or like what version
25:40 of our reality are they living in that
25:43 they ended up like this and not like me
25:47 that was kind of cool anyway I don't
25:49 know what got us on to that someone said
25:51 something oh failed Pixar franchise see
25:53 that's this is the other thing that that
25:55 uh the Irregulars uh have have nicknamed
25:59 the AI learning lab is chat add so
26:01 welcome to chat add where we just
26:05 [ __ ] go you you pop prompts in down
26:08 below I'm kind of like chat gbt except
26:12 with much less knowledge and I act like
26:16 an idiot and I have ADD so so uh so
26:19 we'll just twirl around for a while
26:22 let me go find one of these that I like
26:24 and they all kind of feel the same to me
26:27 right now I got to make some more of
26:28 these
26:29 well that could be another thing if you
26:30 want to go to go to ideogram
26:34 ideogram dot a i
26:37 and just start making some uh
26:40 making some images and if you make any
26:42 cool AI learning lab ones uh you can
26:45 share them on the Discord the AI Salon
26:47 Discord which I'll talk about
26:49 in a bit
26:51 but let's put that there for now
26:54 and then we look all official look at
26:56 that
26:57 you know it almost looks like we try
27:00 [Laughter]
27:04 look at the Loco
27:07 hey your eggs from Dr J put all of Heart
27:11 of Darkness into Claude 40 years too
27:14 late for high school exactly Ted I know
27:18 but I'm doing that now like whenever
27:20 someone sends me
27:21 an article from a reputable institution
27:24 that's you know way too long for my
27:27 attention span I go straight to Claude I
27:30 copy that [ __ ] paste it in there give me
27:34 the 10 bullet points of this
27:36 it's brilliant love it
27:44 what is the money of AI
27:48 Abigail wait
27:50 AB AG nail
27:54 a bagnail Abigail 24.
27:58 I don't understand your name but it's
28:00 kind of cool sounded
28:02 um
28:03 what is the money of AI I don't know
28:05 what that means what is the money of AI
28:08 how do you make money with Chachi BT how
28:10 do you how do you
28:14 um
28:17 well right now we're so early so a lot
28:20 of these tools a lot of the tools like
28:22 chat GPT and Claude and Bard and just
28:25 all of them
28:27 um are what I like to call janky pieces
28:29 of [ __ ] they are
28:31 profoundly powerful
28:34 but they're very very early like we're
28:36 we're Within
28:38 kind of a within a nine to ten month
28:40 window of chat GPT being the first thing
28:43 out of the gate that was kind of machine
28:46 learning for the rest of us right so
28:48 chat GPT in my opinion is the equivalent
28:51 of what the World Wide Web was to the
28:53 Internet so the internet had been around
28:55 for decades Tim berners-le invents this
28:57 thing called the world wide web and you
28:59 can just click on the internet and it
29:00 takes you
29:01 to documents all over the world
29:05 and it changed everything right because
29:06 it went from you needed to be kind of a
29:08 geek to understand the internet and FTP
29:11 servers and gopher servers that you had
29:15 to dial up to all that sort of [ __ ] to
29:17 like I want that I want that I want that
29:20 Chachi PT is the equivalent of that for
29:23 these large language models
29:25 and so all of these tools right now are
29:29 um super raw and the the core technology
29:33 is like just good enough that it it
29:36 blows your mind and you can kind of see
29:37 what the future is going to look like
29:41 um but it's not really ready for prime
29:42 time so so one I think part of the short
29:45 term money for for things like chat gbt
29:47 and for these AI tools are education so
29:50 anyone who learns anything about it can
29:53 start talking about it and saying here's
29:54 what I've learned here's what I've
29:55 learned everyone's Hungry For answers
29:57 right now there are not great answers
30:00 from anyone there are no experts this
30:02 stuff's too raw there are experts in
30:05 machine learning and there are experts
30:07 in
30:08 in you know neural networks and
30:11 embeddings and Vector databases and Lang
30:14 chain and all the sort of technical
30:15 stuff lots of experts in that
30:18 but there are no experts in how do we
30:20 use tools like chat GPT and like
30:23 mid-journey and ideogram
30:25 how do we use them as part of our daily
30:27 workflow I don't know
30:30 I'm doing everything I can and then the
30:32 minute I get something figured out some
30:34 new tool comes along and reinvents it
30:36 again and then reinvents it again and
30:38 then reinvents it again so
30:41 um so so we're in a in a state of
30:43 um you know primordial ooze right now so
30:47 so a lot of the money right now is going
30:48 to be in just figuring [ __ ] out getting
30:50 your name out there as someone who knows
30:52 something about this
30:54 um as things start start to stabilize
30:56 then you know you'll either be able to
30:58 create a company really quickly or
31:00 you'll be able to create a service
31:01 business really quickly I know there was
31:03 kind of a whole rash of
31:05 um
31:06 automation agencies that are taking
31:08 tools like zapier and combining that
31:11 with calls to open AI to these large
31:13 language models and doing business
31:14 automations that are like these magical
31:17 [ __ ] applications that they can whip
31:19 up in a week or a day
31:23 um and and selling those to companies so
31:25 there might be ways to do with that my
31:27 philosophy on this stuff is that you
31:29 don't make money with chat GPT you make
31:31 money by solving problems
31:34 that other people have
31:36 that if you solve it good enough
31:38 they're willing to give you money for it
31:40 and you just happen to be using these as
31:42 the tool to get there
31:45 so I would say don't don't think about
31:47 how do you make money with chat GPT
31:49 think about what's a problem I can solve
31:51 and if I know these things well enough
31:53 maybe I can do it better faster or
31:56 cheaper than the competition so that's
31:59 the that's the secret unlock here is
32:02 what these things amplify which is your
32:05 intention and your your skills
32:08 they're not doing [ __ ] on their own
32:10 so
32:12 I think that was your question I hope
32:14 that was question that was your question
32:15 I love how we all greet each other then
32:18 see everyone start their tapping I know
32:20 Silver Fox it's hilarious there is
32:22 totally a whole underground thing that
32:24 happens that in in the comments and it's
32:28 very very cool I gotta say the community
32:30 here is really neat it warms my heart
32:32 it's it's fun to hang out with you guys
32:35 um
32:37 all right
32:39 promptbook.com nope but I am swell Kyle
32:42 or dot AI prompt book again just in case
32:46 you missed that before
32:47 openart.ai promptbook
32:50 that's a remote document if you haven't
32:53 gone and played with promptbook it's
32:54 like a 96 page PowerPoint presentation
32:57 that was designed for stable diffusion
32:59 but it's it's it's decently Universal
33:02 across all these these image things but
33:04 what it tells you is just how remarkably
33:08 um
33:09 comprehensive these image tools are in
33:11 understanding all sorts of art Concepts
33:14 like you know camera lenses and film
33:17 types and illustration types and art
33:20 types and lighting conditions and just
33:23 you know Styles like cyberpunk and retro
33:27 Punk and future Punk and elbow Punk oh
33:30 you just like all of that [ __ ]
33:32 think of any art history class you took
33:36 these tools kind of know all of that
33:40 [ __ ] so so the prompt book is a is a
33:42 really neat
33:44 um primer that that kind of reconnects
33:48 you with oh yeah there's like the art
33:50 world is deep and and using something
33:52 like prompt book and those Concepts you
33:54 can make
33:55 art that is Will it'll blow your mind I
33:59 mean it blows my mind anyway say it
34:00 slower
34:02 please what do the shirts look like we
34:04 did that but I'll show you again that's
34:07 what the shirt that's the
34:09 that's the art on the front of the shirt
34:10 I'm not weird I'm irregular and then it
34:12 says AI learning lab below it with 1l
34:16 not our not our ever loving scammer with
34:21 this 2L his 2L magic AI Lilla learning
34:25 lab
34:26 our scammer stutters
34:32 [Music]
34:41 image prompt book I didn't get it down
34:44 in time all right Thelonious we're going
34:46 one more time I think I I think I've
34:48 done this enough now but
34:52 open art dot a I slash prompt book
34:57 cool
35:04 [Music]
35:08 all right
35:09 looking loud
35:12 Thelonious is probably sitting there
35:14 watching some TV show snacking on nachos
35:19 he's like ah [ __ ] what was prompt book
35:21 again damn it I can't dag Jabet
35:27 [Laughter]
35:35 [Music]
35:41 wait let Tobias
35:44 take care of your t-shirts and swag okay
35:47 if Tobias is good at that
35:49 it wants to manage that [ __ ]
35:53 I
35:54 I mean
35:56 after my last two days I [ __ ] made
35:59 this shirt I made all the things I made
36:00 all the patterns I figured out how to
36:02 hook it up to [ __ ] Tick Tock I do it
36:05 I'm 10 minutes into my thing yesterday
36:07 and they're like Tick Tock violation
36:10 you're doing the store wrong you [ __ ]
36:12 jerk no
36:14 it's like
36:16 oh
36:19 yeah I need to I need to connect with
36:22 some of the Irregulars and do get some
36:25 mods in here and if if Tobias wants to
36:27 help with that that would actually
36:28 probably be a good thing
36:31 it would probably be good yeah
36:35 file a complaint on the scammer and
36:37 attach a screenshot yes please Nancy
36:39 thank you for that
36:42 I reported him Kyle thank you Robert
36:44 Rossi chat gpt4 keeps airing out when I
36:47 try information from having it look
36:49 through plugins
36:50 not sure what I'm doing wrong it might
36:52 not be you let's go see sometimes chat
36:55 gbt that sometimes the plugins just wonk
36:58 out
36:59 again back to that idea that these tools
37:01 are janky pieces of [ __ ] it's just
37:03 they're new and sometimes they just
37:04 don't work they don't work good
37:08 um with plugins right now
37:11 um if you're new to chat GPT this is
37:14 chat gbt so you've got down the left
37:16 hand side you've got your prompt history
37:20 um you've got GPT 3.5 which is the free
37:24 model it's also the fastest model it's
37:26 super fast and then for gpt4 you've got
37:30 default GPT for Advanced Data analysis
37:34 which used to be called code interpreter
37:36 and you've got plugins
37:39 um if you don't see those oh the other
37:41 thing is to get gpt4 with all that extra
37:44 [ __ ] you have to have uh chat GPT plus
37:46 which is 20 bucks a month
37:50 um
37:52 I'm not affiliated with them in any way
37:55 so whether or not you pay that or not I
37:57 think it's worth it but it's up to you I
37:58 don't really give like I don't make
37:59 anything from it just full disclosure
38:01 I'm not making money from it
38:04 um if you pay your 20 bucks and you
38:06 don't see those things you have to go to
38:07 Beta features in settings and turn them
38:09 on
38:10 so you have to explicitly turn on
38:12 plugins and Advanced Data analysis this
38:15 also used to have browse with Bing it
38:17 was connected to the internet but they
38:19 killed that I I don't it it seems to me
38:22 they're not bringing connection to the
38:24 internet back which I find absolutely
38:26 [ __ ] weird but whatever okay so let's
38:29 go check in check on plugins and see if
38:31 they work so let's go grab
38:35 see
38:37 Let's see we can grab an article we'll
38:39 go to we'll go to
38:42 we'll go to the Twitter
38:45 I'll go to General generative AI
38:47 although uh
38:49 um AI
38:51 article I will do Ethan Malik we'll do a
38:54 Malik
38:56 Ethan Malik article
39:00 is this an article
39:04 choo
39:10 Conquering the AI Frontier Watson's
39:12 cautionary tale and the promise of
39:14 tomorrow
39:17 um do I need to have a membership to
39:19 read this
39:22 that's really short but whatever okay so
39:24 I'm just going to grab this URL
39:26 and then I'm going to go back to chat
39:29 gbt and so the way plugins work is
39:33 you've got
39:34 um the plugins you've installed so I've
39:37 installed a lot of them in any given
39:39 chat you can
39:41 enable activate up to three I tend to
39:45 just activate one at a time but you
39:47 could do like if you had three travel
39:49 related plugins and you wanted to do
39:51 kind of a travel conversation so
39:53 basically what you're doing the plugins
39:55 kind of augment
39:57 chat ept's functionality and and chat
40:01 apt figures out if it needs to use
40:03 plugins or not like once you understand
40:04 what a plug-in is you don't really have
40:06 to do anything if it needs it it'll use
40:08 it if it doesn't need it it won't use it
40:10 it's pretty slick in that way
40:13 um a lot a lot a lot of the plugins are
40:16 just useless time wasting pieces of crap
40:20 some of them are decent the one that I'm
40:22 using here is is one that someone told
40:23 me about last night
40:25 and it's called
40:28 browser up
40:32 b-r-o-w-s-e-r-o-p and it's it's it's
40:34 very similar to web pilot so I'm going
40:36 to say
40:38 um please
40:43 summarize this and I spelled summarize
40:47 wrong I know that and then I hit shift
40:49 return and say
40:51 um give me a short paragraph
40:56 summary followed
40:59 by
41:00 bullet
41:02 points of
41:06 key
41:07 takeaways whatever
41:10 it's very corporate uh
41:14 each bullet should
41:17 be 7 to 10 words okay so if plugins are
41:24 working this should
41:27 fire off the plug-in
41:32 which it did which is a good sign
41:35 and then if you open this up it says
41:38 okay it gave it a link it returns
41:41 something
41:44 Conquering the AI Frontier
41:47 discusses the overhyped expectation and
41:50 subsequent shortcomings of IBM's Watson
41:52 in the health industry while Watson was
41:54 anticipated to revolutionize Health Care
41:56 with data-driven insights that face
41:58 challenges like data privacy
41:59 transparency and system integration
42:01 the article contrasts this with Ethan
42:04 Malik's perspective which emphasizes the
42:06 potential they had to bring positive
42:07 chain across sectors the overarching
42:10 message is that while AI has vast
42:12 potential is successive depends on
42:14 informed and thoughtful implementation
42:16 which is right key takeaways IBM's
42:19 watching face challenges in
42:21 implementation over ambitious
42:23 expectations can lead to disappointments
42:25 absolutely and that's good and that's a
42:28 that's a good little tip for life also
42:31 anytime you have an expectation it just
42:34 that's that's where all of the problems
42:37 come from the minute you have
42:38 expectations good or bad that's where
42:41 the problems come from
42:43 so if you can just sort of live in the
42:44 center there's there's there's no
42:46 attachment here I'm I'm committed to
42:48 success there's no attachment here right
42:50 Zen In The Art of AI
42:53 um understanding ai's functionality is
42:55 crucial for its application as so so it
42:57 looks like
42:58 um plug-ins in general are working it
43:00 might be that the specific plugins that
43:03 you used uh were not working
43:06 because again what what a lot of these
43:09 plugins are are basically a gateway to
43:12 someone else's website so you're in chat
43:14 GPT
43:15 and it fires up
43:17 um like like what's one called ask your
43:20 PDF there's a whole site called ask your
43:22 PDF where you can upload a PDF
43:25 but there's also a plug-in when when it
43:28 calls the plug-in what it's actually
43:29 doing is sending you over to their site
43:31 doing some [ __ ] over there and then
43:33 sending an answer back
43:35 um
43:36 if their sites down that plug-in will
43:39 break so so that could be what's going
43:41 on there I don't think it's I don't
43:42 think you're doing anything wrong
43:43 plugins are
43:44 relatively simple and relatively stupid
43:48 um
43:51 so yeah so yeah
43:53 I'm sorry you're having challenges with
43:55 them near plugins plugins is not they're
43:58 not my favorite thing in the world
43:59 because they're they're not they're not
44:01 so good they're not nobody really knows
44:04 what the [ __ ] to do with them that's
44:05 kind of the problem
44:07 although Microsoft said that they're
44:09 going to be supporting plugins in when
44:12 they roll
44:14 um what they're calling co-pilot gpt4
44:16 into all of their apps they're going to
44:19 support plugins so this could the
44:22 plug-in store could get like the App
44:24 Store is now so when the when the iPhone
44:25 first came out if you're old enough to
44:27 remember that you you friggin old and
44:29 crusty [ __ ]
44:31 when the iPhone first came out
44:33 um it didn't have apps they they were
44:36 called widgets
44:38 and they weren't really widgets they
44:40 were like shitty websites they were like
44:42 they were like you could make an icon
44:44 that was a link to a website and it was
44:46 kind of this shitty website
44:49 and everyone bitched about it they're
44:50 like wait these most of these don't work
44:52 and isn't this just a glorified website
44:55 these these what what are we doing here
44:56 what what are these and then that
44:59 ultimately became the app store right
45:00 and it became a whole industry in and of
45:02 itself so
45:04 um
45:05 yeah
45:08 all right that's that
45:11 um
45:13 truth don't buy Chinese okay fantastic
45:17 tried an auto clicker using Samsung Dex
45:20 but Tick Tock recognizes the steady
45:23 clicking pattern so so drift staying
45:25 very nice by the way solid that that you
45:28 try to you're gonna have I guess you're
45:30 gonna have to get Auto clicker 2.0
45:32 that's got some randomization function
45:34 in it so that it it doesn't it doesn't
45:37 feel uh it you know they can't Tick-Tock
45:40 it that's really funny though RMS Studio
45:42 plugins are
45:45 not loading correctly in my what are RMS
45:49 Studio plugins
45:51 I don't know RMS Studio
45:55 RMS Studio
45:59 lug
46:01 ends
46:09 [Music]
46:11 are those audio things
46:21 yeah I don't know what they are all
46:23 right
46:24 um that's good to know I mean yeah just
46:26 plugins in general it's it's it's it's
46:28 not you
46:29 we grilled hey everybody Winston's here
46:32 welcome Winston dragonfly Alchemy
46:34 welcome welcome welcome
46:36 BLTs for Labor Day dinner that sounds
46:38 delicious actually we have such amazing
46:41 Tomatoes I don't even see a cool tomato
47:03 look at this bad boy we have three
47:05 Tomatoes grew into one
47:08 it's like it it didn't know what it
47:10 wanted to do so it's just this lovely
47:12 giant multi tomato
47:17 I feel like we should need it it feels
47:19 it feels like you know something
47:20 something is wrong here this is like an
47:23 this is like a mid-journey generated
47:24 tomato
47:30 ah but man they're tasty our Tomatoes
47:32 this year really good I'm new here
47:34 thanks for all this great info you're
47:35 welcome women of jazz by the way 10 out
47:38 of 10 on the name love the name
47:40 um yeah tell me where you are in this
47:43 whole AI Journey that the whole point
47:45 here is whether you're deep in it not in
47:48 it
47:49 just like I was stumbling along I
47:51 thought it was Rosie O'Donnell but her
47:52 voice changed
47:54 you know what whatever's the reason
47:56 you're here welcome and and yeah just
47:59 ask questions I'll do what I can to uh
48:03 I don't know
48:05 have fun learn
48:06 get curious about this [ __ ] it's
48:09 important I think this stuff's actually
48:11 really important
48:14 hang on I gotta close the door
48:29 let's explore AI together here that's
48:32 the whole point
48:33 let's get it going let's figure it out
48:35 Winston everybody here had steak but me
48:38 I had two veggie patties because I'm a
48:42 300 pound soy boy
48:47 hey we gotta do what we gotta do I'm
48:48 going I'm going back on smoothies
48:50 tomorrow
48:52 so yeah
48:54 it's just you know
48:56 um okay farmer yeah if Tobias wants to
48:59 help with the t-shirts I would love to
49:00 talk to him about that that's fine that
49:02 would be awesome
49:03 because I
49:06 I think I've been broken my my spirit my
49:11 spirit she's broken
49:13 she's she's one Thin
49:17 um RJ I have asked others what's the
49:19 definition of an American
49:22 um I don't know
49:27 I don't know
49:36 if you judge by the past 10 years it's
49:39 like picking a side and fighting to the
49:41 death for your side
49:43 it's like
49:44 I don't know we've we've become so
49:47 divisive that uh like I I feel like it
49:50 used to be
49:53 what United us was we thought we were
49:55 the best but we all generally agreed on
49:57 what that was right we were just like
49:58 you know pompous overconfident Americans
50:02 mm-hmm
50:04 and now we've divided into my flavor of
50:07 America is better than your flavor of
50:08 America and [ __ ] you if you don't agree
50:10 with me so I I would love for it not to
50:13 be that I would love for it to go back
50:14 to
50:16 some version of we're kind of in this
50:18 boat together wouldn't it be cool if we
50:20 could all just do stuff I mean that's
50:21 that's I mean I'm not trying to solve
50:24 politics here but like I like I don't
50:26 want any of that [ __ ] in here like
50:29 I think this AI stuff is is is going to
50:32 have profound impacts on all of work
50:34 like no matter what your religious or
50:37 political backgrounds are and
50:41 just having been through one of these
50:43 before with the early days of the web
50:45 um
50:47 we should get generous and and support
50:48 one another in learning this stuff
50:50 because it's I gotta tell you it is
50:52 impossible to keep up with this [ __ ]
50:54 like my list
50:56 this this list used to just be three
50:59 things on the list and it was like that
51:00 was all the stuff it was like
51:01 mid-journey open Ai and maybe Bing
51:06 was was this list and now it's those
51:08 three
51:09 plus those four
51:11 plus those four and then you know these
51:14 other things are just resources and
51:16 those are the ones that I that I think
51:18 are worth paying attention to and worth
51:19 actually exploring in some way to
51:21 understand what what they make possible
51:24 um there are thousands of other tools
51:27 and it's changing daily
51:31 um so no single person can keep keep up
51:34 with this [ __ ]
51:35 um even the people doing newsletters and
51:37 things like that are overwhelmed um
51:41 um
51:42 and
51:44 everyone's trying to figure it out so
51:45 why not help each other
51:47 like like when something is being
51:49 birthed like this you know one one of
51:52 the things about Americans is they're
51:54 competitive right so it's like I don't
51:56 want to give them my secrets
51:58 when something is being birthed out of
51:59 nothing and you're in a growth mode
52:04 anyone else who does what you do is a
52:07 validator more than a competitor this is
52:09 something I got from my senior advisor a
52:12 guy named Kevin Clark who who I'm a big
52:14 fan of
52:15 he talks about when you're in growth
52:17 mode the more the merrier because
52:19 everyone is validating
52:21 the market everyone is validating the
52:24 technology everyone's validating the
52:25 solutions you come up with something
52:27 brilliant and sharebear comes along and
52:29 says oh I got something that's just like
52:31 that don't think of that as a
52:33 competition think of like like that as
52:35 like oh cool how are you doing it here's
52:37 how I'm doing it get generous with that
52:39 [ __ ] because Sheriff bear might have
52:41 come up with some way of thinking about
52:42 something that I'm not thinking about
52:44 and and if she shares that super secret
52:47 sauce with me does that mean I'm gonna
52:50 out-compete her no we're in a growing
52:52 Market we're we're in a market that is
52:55 just starting to expand and it's gonna
52:57 [ __ ] explode
52:59 so there is there's going to be so much
53:02 more opportunity that any of us can
53:04 individually deal with
53:06 that it only benefits us all to be
53:08 generous right now now three years from
53:11 now five years from now when that growth
53:13 has kind of subsided the things
53:15 plateaued and things are starting to
53:16 stabilize and things are starting to
53:18 shake out okay that's when you're in a
53:21 competitive mode right that's when you
53:22 start to say okay now I want to start to
53:24 Corner part of the market and I'll say
53:27 I'd say to share bear you sit stick off
53:28 in your corner I'll stick in mine and
53:30 we'll go our proprietary ways but we're
53:32 not there right now and like like this
53:34 is with this is again we're in
53:37 primordial soup right now
53:39 so learn everything you can share
53:43 everything you can be generous with it
53:47 if someone asks you a question and
53:49 you're like you just learned it
53:50 yesterday just say yeah I just learned
53:54 this yesterday but here's what I know
53:57 or
53:58 here's this thing that I know that I
54:00 think is really obvious and it's
54:02 probably going to be of little to no
54:04 value to you and then you share it and
54:07 the other person goes oh my God that's
54:09 profound
54:12 why is that because everyone's trying to
54:14 figure it out right now
54:16 so the thing that just the way your
54:18 brain works you use chat CPT in a
54:21 particular way and it kind of works for
54:22 you and you're like oh but that's
54:23 probably stupid there's probably real
54:26 prompt Engineers out there that know how
54:28 to do this better than I do
54:30 and then you share that with a real
54:31 prompt engineer and they're like oh I
54:33 haven't thought of that way to thinking
54:34 about it
54:35 cool
54:37 so yeah
54:41 I would like that to be the America that
54:44 we live in more than what we are right
54:46 now
54:49 well I Can Dream a Boy Can Dream can I
54:51 not dream
54:53 he emerges from the cornfield
54:56 wondering why can't we just get along
55:00 almost 30 000 likes you guys are insane
55:03 I'm freaking tear tonight
55:05 [Laughter]
55:08 Vicky loves seaweed salads hey Kyle here
55:11 regulars what's happening Stevo shaken I
55:14 nominate Tobias to be the all merch
55:17 Captain like it
55:19 uh Kyle Wonka wesn Radio rock and roll
55:23 and smooth hip-hop
55:26 welcome wesn radio thank you thank you
55:29 thank you for the Rose The Rose man of
55:31 the name of the smell is sweet welcome
55:33 welcome welcome
55:35 I appreciate your contribution
55:38 of lovely smelling Flora
55:42 I always confuse flora and fauna but
55:44 then Flora sounds like flowers and those
55:46 are plants flora fauna fauna to me
55:49 sounds like plants but it's animals
55:53 welcome to chat chat add
55:55 [Laughter]
55:59 ERS think that way over time
56:02 actors think what way over time that we
56:05 can all just get along
56:08 ideogram isn't making the best eyeballs
56:11 on fingers
56:12 Well yeah if you want eyeballs on
56:14 fingers you're gonna have to find an
56:15 eyeballs on finger
56:17 if you if you want really specific [ __ ]
56:19 like that
56:21 um
56:21 I would start exploring Laura's and
56:25 um and uh
56:27 stable diffusion because there's a whole
56:30 sub community of people trying to do
56:31 really specific Kinds of Kinds of
56:33 artwork or use mid-journey mid journey
56:36 is better just in general
56:37 been a vegetarian for 30 years pretty
56:40 fat a little over 300 pounds but I lift
56:42 weights all right so as long as you list
56:44 weights I'm a mess oh Winston aren't we
56:46 all I mean you know listen we're doing
56:50 what we're doing we're doing our best my
56:51 blemish went away I had to sit here for
56:53 a while
56:54 wait what's going on here hang on
56:58 um
57:05 NPC couple day five checkout
57:11 someone from work is talking about
57:13 there's some MPC Tick Tock Channel
57:18 the NPCs I know
57:23 I just saw a notice that Elon Musk is
57:25 suing someone all right
57:28 um
57:31 Abigail Frank Abagnale will the real
57:33 Frank Abagnale oh that's funny
57:36 um I eat seaweed sometimes too tried
57:39 ideogram made janky piece of [ __ ] poster
57:42 it used listen here's the thing Ted
57:45 looks at life ideogram from an image
57:48 standpoint ideogram is like dolly is
57:51 it's kind of shitty but whatever they've
57:55 done to get words to show up
57:58 15 percent of the time at least makes it
58:00 usable like like here's how I think
58:02 about ideogram right now if I want to
58:04 get ideas just kind of design ideas for
58:07 typography kind of stuff like if I don't
58:09 want to sift through old typography
58:11 books and and things like that or if I'm
58:13 not necessarily a designer just go in
58:16 there type in type typography an era a
58:19 style
58:20 and just let it do it let it do its
58:22 thing and then I could pull that into
58:24 into illustrator and and you know do a
58:27 real version of it so so ideogram right
58:30 now I I feel like is a
58:32 is a good concepting tool I feel like
58:35 mid-journey you can do you can do kind
58:38 of camera ready art for for using using
58:41 the old school terms
58:43 um
58:44 ideogram you can sort of get there but
58:47 if you really want to like look at it
58:48 and pay attention to it you're gonna
58:50 have to redraw all that [ __ ] anyway
58:52 um but with all these things the thing
58:55 about these tools right now this is
58:57 another thing if you're if you're new to
58:58 this space the thing about these tools
59:01 right now
59:03 as remarkable as they are and as flawed
59:05 as they are
59:07 this is the worst they're ever going to
59:10 be
59:11 they're they're never going to be as bad
59:12 as they are today
59:14 the the speed with which these things
59:16 are accelerating in like the quality
59:19 improvements of these things is pretty
59:21 staggering if you look kind of 18 months
59:24 ago at what the image generation tools
59:26 were doing they were laughably bad like
59:29 ideogram is you know
59:31 miles ahead of where we were 18 months
59:34 ago with any of these image generation
59:36 tools
59:38 um and now you look at mid-journey stuff
59:39 and it's it's it's getting pretty close
59:42 to world class right pretty close
59:45 so another year and you know all these
59:49 things are going to be amazing and then
59:50 the same is going to happen with video
59:52 the same is going to happen with chat
59:54 GPT the same is going to happen with
59:56 like citations same is going to happen
59:58 with audio and voice synthesis all of it
1:00:01 all of it just the quality is just going
1:00:04 to ramp up to the point
1:00:07 that
1:00:09 a year from now say right before the
1:00:12 2024 election
1:00:15 this stuff is going to be staggeringly
1:00:17 sophisticated and not just chat gbt the
1:00:20 video stuff the audio stuff the
1:00:22 so the get get ready for some wacky
1:00:25 misinformation hijinks
1:00:28 in America's 2024 election you can make
1:00:31 money with your gbt and apparently you
1:00:34 can make a politician
1:00:36 apparently you can make a politician
1:00:37 French kiss a gorilla here we go let's
1:00:40 have some fun
1:00:43 but I saw the video
1:00:46 yeah
1:00:51 there's also going to be the
1:00:52 misinformation version of catfishing
1:00:55 which this this I think is a fascinating
1:00:57 one so there's gonna be shit's gonna
1:00:59 come out where someone didn't do it but
1:01:01 now we have video evidence that they did
1:01:05 and then the catfishing version is going
1:01:07 to be there's going to be actual video
1:01:09 evidence of someone doing something
1:01:11 wrong and then they're gonna be like
1:01:12 it's AI what can I say it's like I just
1:01:17 it's gonna it we we're gonna it's gonna
1:01:20 be special we're gonna have some fun I
1:01:23 think we should I think we should just
1:01:24 turn up our empathy dial turn up our
1:01:27 empathy dial turn down the hate dial all
1:01:30 of us right now
1:01:31 know that this stuff is coming and
1:01:34 provide a little Grace because I think
1:01:36 that
1:01:37 um yeah it's gonna be special it's gonna
1:01:40 be special so so just understand
1:01:43 everyone's going through something
1:01:44 everyone's going through a little
1:01:46 something you never know who had a bad
1:01:47 day
1:01:48 right all right so there you go
1:01:53 all right welcome welcome welcome
1:01:55 welcome
1:01:56 to Fantasy Island oh how many Fantasy
1:01:59 Island fans in the Irregulars come on
1:02:01 that was fantastic the plane the plane
1:02:08 you probably can't even talk about
1:02:09 Fantasy Island anymore without getting
1:02:11 canceled because it had a uh
1:02:15 small person what it's a small person on
1:02:18 it
1:02:23 ah
1:02:25 I tried scholarly social and web plugins
1:02:27 for gpt4 kept giving me errors when I
1:02:30 try them
1:02:31 huh
1:02:33 scholarly social and web plugins
1:02:37 that's weird Renee
1:02:39 um
1:02:42 a thing to try might be
1:02:44 um
1:02:46 open chat GPT in a
1:02:49 um in an incognito window on your
1:02:52 computer you might have some weird
1:02:54 caching issue going on where maybe
1:02:57 you've got I don't know there might be
1:02:59 some cash issue going on if you open it
1:03:03 up in a in an incognito window log into
1:03:05 chatgpt see if it gives you an error
1:03:07 there
1:03:08 and if it if it doesn't error in an
1:03:11 incognito window
1:03:13 um obviously quit quit and restart your
1:03:16 browser try a different browser
1:03:19 um
1:03:21 restart your machine obviously might be
1:03:23 a good thing to do and then if all that
1:03:25 fails
1:03:27 um then just you know clear your clear
1:03:29 your cookies clear your cash just clear
1:03:31 all the browser [ __ ] because I'm having
1:03:33 I'm having weird [ __ ] right now with
1:03:35 Microsoft edge with browser stuff that
1:03:38 was part of my frustration today was a t
1:03:40 kept telling me to log in and do [ __ ]
1:03:42 but I couldn't log into at T and it
1:03:44 wasn't until I did a an incognito into
1:03:47 the I could log in so I was having some
1:03:49 of those issues with at T today
1:03:53 um
1:03:54 I have to keep narrowing
1:03:59 oh dragonfly Alchemy talking to Renee
1:04:02 yeah he has to keep narrowing the
1:04:03 plugins down yeah basically what I've
1:04:05 done with plugins is like I've found a
1:04:07 handful that I like and I I use
1:04:09 occasionally but at this point now that
1:04:13 now that we've got Bard connected to the
1:04:15 internet Bings connected to the internet
1:04:17 and decently reliable and Pi is
1:04:20 connected to the internet I don't the
1:04:22 most of what I was using plugins for was
1:04:25 you know [ __ ] connected to the internet
1:04:27 so and reading PDFs and now I can do
1:04:30 that with code interpreter or with
1:04:31 Claude so
1:04:33 um so I I've I haven't really used
1:04:35 plug-ins in a while
1:04:37 I definitely like turtles I even save
1:04:39 turtles in traffic very nice like that
1:04:42 bright shiny object syndrome I have that
1:04:45 too yeah welcome to the club Vicky it's
1:04:48 uh it's a special Club
1:04:49 squirrel squirrel
1:04:51 um can't wait for them to incorporate
1:04:53 Advanced AI into video games video games
1:04:55 are gonna be off the hook
1:04:58 they're gonna be nuts
1:05:00 it's just gonna be like
1:05:03 um I want this kind of game in this kind
1:05:05 of environment with this kind of game
1:05:07 mechanics
1:05:08 and and I want it now and I just go and
1:05:13 make you your game and you start playing
1:05:15 and you know they'll suck at first and
1:05:17 then three years later you'll be like
1:05:20 what do you mean you used to have to
1:05:21 hand code games what do you mean there
1:05:23 were people that used to build 3D worlds
1:05:26 didn't you just ask for them
1:05:29 no Jimmy in my day
1:05:32 you had to spend 150 million dollars on
1:05:35 a triple A game and it was always six
1:05:37 years late
1:05:39 but I get games like that now when I ask
1:05:43 for them right isn't that how it was no
1:05:45 Jimmy
1:05:48 I used to be senior vice president of
1:05:50 Electronic Arts let me tell you
1:05:52 something boy
1:05:54 foreign
1:05:56 yeah Jimmy's not going to appreciate the
1:05:59 [ __ ] we're going through right now
1:06:01 Jimmy's not going to appreciate how
1:06:03 complicated everything is
1:06:05 Jimmy's gonna be like don't you just ask
1:06:07 for it daddy no
1:06:11 go to bed
1:06:13 but I don't sleep anymore because we
1:06:15 have pills for that now shut up
1:06:20 just go lock yourself in the room
1:06:24 don't take your sleep pill tonight
1:06:26 get some actual sleep like a human being
1:06:33 but Daddy I don't understand
1:06:36 all right
1:06:39 um
1:06:39 what are we up to 31 000 likes you guys
1:06:43 are nuts love it
1:06:46 save the date spring 2024 AI learning
1:06:49 lab Salon Khan AI Salon Khan I like it
1:06:53 we could do that that would be fun
1:06:57 that would actually be fun
1:07:00 put a little structure around that well
1:07:02 because what we've what we've what we've
1:07:03 talked about in the AI salon by the way
1:07:05 if if you if you're curious about
1:07:08 if you want to join a community of
1:07:12 AI adventurers people that are
1:07:15 trying to figure this [ __ ] out and
1:07:17 really generous and really empathetic
1:07:20 I started this group called the AI Salon
1:07:21 last December and
1:07:24 um we meet every other week we meet in
1:07:26 person here in Denver and then we meet
1:07:28 online you know we when we have those
1:07:31 meetings and then we have a Discord
1:07:33 server and you can hang out there and
1:07:34 I've got a little private channel for
1:07:36 the Irregulars of this channel to chat
1:07:38 over there so if you hang out here
1:07:40 regularly there's a place for us to talk
1:07:42 over there
1:07:44 um the salon.ai is the website it's a
1:07:47 link tree it contains like the first
1:07:49 three links are the important ones the
1:07:51 first one is the link to the what's the
1:07:53 salon about what are our values so look
1:07:55 at those values
1:07:57 if you're stingy and mean those are not
1:08:01 our values generous empathetic
1:08:03 adventurous all of those are so go go
1:08:06 read about that
1:08:08 um and then the second link is to the
1:08:10 Meetup so you can figure out when the
1:08:12 when the meetings are and the third link
1:08:14 is to the Discord it's an invite to the
1:08:16 Discord server and there's some other
1:08:17 [ __ ] there but but those three are the
1:08:19 most important ones
1:08:20 um but yeah go join that we've talked
1:08:22 for a while about
1:08:24 um
1:08:25 figuring out how to do in-person
1:08:27 meetings in other cities because it's
1:08:29 it's logistically really complicated you
1:08:31 can't really do
1:08:33 sort of screen sharing from multiple
1:08:34 cities and and not have it be an abject
1:08:36 nightmare or just a logistical nightmare
1:08:40 um but it but it could be the kind of it
1:08:42 could be the kind of thing that there
1:08:43 could be a really nice Synergy between
1:08:45 you know this Channel and and the salon
1:08:48 um yeah I could see that I like it
1:08:50 cooperino I think we could we could we
1:08:53 could figure something like that out and
1:08:55 and you know then we would have merch in
1:08:57 person and Tick Tock wouldn't yell at us
1:08:59 for not doing our [ __ ] store right
1:09:02 ha
1:09:05 I said it
1:09:06 uh
1:09:08 and I don't know about you like my Tick
1:09:10 Tock feed now is like every third
1:09:12 [ __ ] thing is someone selling
1:09:14 something it's like enough give me
1:09:16 [ __ ] content again
1:09:19 people playing that stupid game where
1:09:20 they throw the suction cups on on the
1:09:22 counter the little stick up suction cups
1:09:26 like safe jarts remember jarts that put
1:09:30 people's eyes out
1:09:35 uh
1:09:37 plug-in keymate annoys me with its
1:09:39 built-in sponsored ads and you only get
1:09:41 so many queries for free yeah exactly
1:09:43 Luna stick a lot of the plugins one
1:09:46 plug-in
1:09:47 one plug-in was like
1:09:49 um
1:09:50 uh it was like web website rewriter or
1:09:54 something like that and I'm like oh cool
1:09:56 something that's going to rewrite my
1:09:57 website for SEO and so I gave it my
1:10:01 website and it and it rewrote it and it
1:10:04 was fine and then basically the the last
1:10:07 paragraph Was you know
1:10:10 you know if you want work like this and
1:10:12 other things reach out to our agency and
1:10:15 I'm like oh wait this plug-in was just
1:10:17 an ad for a [ __ ] agency no no no [ __ ]
1:10:21 off
1:10:21 so yeah so a lot of the plugins are just
1:10:24 marketing ploys
1:10:27 um some of them are good
1:10:28 some of them are really no a handful are
1:10:30 really good and then a lot of them are
1:10:32 just garbage like that they're just
1:10:34 marketing crap
1:10:37 all right I'm 99 messages behind oh so
1:10:41 it looks like we got some new folks here
1:10:42 my name is Kyle Shannon uh this is the
1:10:43 AI learning lab and here look we we even
1:10:47 have wait not that
1:10:49 look we have graphics
1:10:51 look I I spent
1:10:53 20 years in design school and typography
1:10:57 school learning how to make these
1:10:58 posters
1:11:00 no I didn't
1:11:02 I went to ideogram
1:11:06 and it does text now
1:11:08 um anyway uh if you have questions pop
1:11:10 them in the comments below
1:11:12 I'm behind I'm 99 at least
1:11:16 messages behind so I think what I'm
1:11:18 going to do is I'm going to jump to the
1:11:19 bottom right now so I'm going to skip a
1:11:21 bunch of questions
1:11:23 um although I will read this one from
1:11:25 rug cleaner machines I'm here for the
1:11:27 whole live greatest Channel thank you I
1:11:30 appreciate that we have fun here uh I I
1:11:33 will sometimes make voices I'm not I'm
1:11:35 not in a voice making mood tonight for
1:11:37 whatever reason but I'll give up it it
1:11:39 Ebbs and flows
1:11:41 um I'll do what I can to answer
1:11:43 questions I'll do it again to demo stuff
1:11:45 if you want to see stuff just let let me
1:11:46 know all right I'm gonna I'm gonna jump
1:11:49 to the Bottom now
1:11:50 Sam Corn's here welcome Sam corn
1:11:54 uh I haven't seen you in a dare two like
1:11:57 I'm guessing you were doing something
1:11:58 cool for the Labor Day Labor Day weekend
1:12:00 it was a long weekend maybe you're on a
1:12:03 road trip
1:12:04 going to find the find how you could
1:12:07 make money with Chad gbt
1:12:12 figure it out
1:12:14 like is it in Iowa where like where do
1:12:16 you make money which at gbt is it is it
1:12:19 on Route 66
1:12:22 all right let's see let's let's get to
1:12:24 some questions here let me go grab a
1:12:26 question
1:12:28 when will Pi get an upgrade and what
1:12:31 would that be
1:12:32 um that's a great question
1:12:35 so
1:12:38 so Pi for those of you that don't know
1:12:41 what what wesn radio is referring to
1:12:44 is pi dot AI this little guy right here
1:12:48 it's so unassuming
1:12:50 they have an app that you can talk to
1:12:52 you can also do the same thing on the
1:12:54 website it's it's just like chat GPT
1:12:56 except it's chatty GPT it's very
1:12:58 talkative
1:12:59 so if you're trying to get stuff done
1:13:01 it's it's pretty useless but
1:13:04 the app in particular if you put the app
1:13:06 into voice mode and just start talking
1:13:08 to it like you're having a conversation
1:13:09 with someone is really quite remarkable
1:13:11 you really kind of get a glimpse into
1:13:13 the future that when they reduce the
1:13:15 latency when it doesn't kind of pause
1:13:17 for half a second in between answers
1:13:20 um
1:13:21 you kind of get a glimpse into where
1:13:23 these things are going and and how
1:13:25 powerful they're going to be
1:13:27 so inflection is the company behind pie
1:13:32 they're setting out to make the world's
1:13:34 greatest personal assistant or I don't
1:13:37 know if greatest is in their marketing
1:13:38 but that's just ostensibly what they're
1:13:41 saying
1:13:43 they just raised 1.3 billion dollars
1:13:45 they're spending about 950 million of
1:13:48 that on 22 000 h100
1:13:51 Nvidia gpus
1:13:53 so they're building a massive
1:13:56 gpe Farm
1:13:58 so what that says to me is they're going
1:14:00 to be working on a Frontier Model so
1:14:01 they're going to be working on their own
1:14:03 kind of version of gpt4 that's probably
1:14:05 going to be optimized for helping people
1:14:09 they're likely going to do some
1:14:10 autonomous agent stuff where you'll just
1:14:12 be able to say to pie hey pie
1:14:15 are there any specials to Barcelona this
1:14:19 weekend oh let me go check yeah there's
1:14:21 two specials on these things and you
1:14:24 have frequent flyer miles with one of
1:14:26 them would you like me to go book you a
1:14:29 flight to Barcelona sure and it will
1:14:31 just go do it like that's so where I
1:14:33 think this thing is headed
1:14:35 is it will be agents that just do [ __ ]
1:14:38 for you and I so
1:14:42 so how far out is that that's probably
1:14:44 two to three years out I would I would
1:14:47 assume that the next
1:14:52 the next iterations of Pi
1:14:56 they could be reduced latency so so
1:15:00 reduced pauses in between when you ask
1:15:02 it something and when it responds
1:15:05 but I feel like that's a bigger problem
1:15:07 and that might that might require their
1:15:09 22 000 GPU Farm to do that
1:15:14 um
1:15:21 I don't know I would think that that
1:15:25 a more sophisticated conversational
1:15:27 model because right now it really is
1:15:28 kind of annoying where
1:15:30 you can't get it to be efficient right
1:15:33 it'll if you just say hey I just want a
1:15:36 list of these specific items
1:15:39 it'll go hey really appreciate that you
1:15:42 want that list let me just clarify one
1:15:44 thing so it's very very chatty
1:15:46 so I would think getting more
1:15:49 sophisticated in how it does its
1:15:51 conversation and the understanding that
1:15:53 there's sometimes you want it to be
1:15:54 efficient there's other times when you
1:15:56 want to have a therapy session and that
1:15:58 those are two very distinct modes
1:16:00 because I think I would think there
1:16:02 could be something with modes I think
1:16:03 there could be an improvement in how
1:16:06 natural the voice sounds
1:16:08 it's still a little monotone and robotic
1:16:10 like all these things are
1:16:12 um
1:16:14 and then I would I would think that they
1:16:16 might start teasing
1:16:18 some its ability to take actions right
1:16:21 so maybe they cut a deal with Expedia or
1:16:24 maybe they cut a deal with I don't know
1:16:28 instacart right and you can just say hey
1:16:31 pie you know
1:16:33 can you can you
1:16:35 um go give me a you know go order me
1:16:38 some groceries
1:16:41 one thing that Pi does decently well
1:16:44 that I think it would be an area that
1:16:46 would make it really interesting and
1:16:48 maybe this is an area of improvement
1:16:50 that's in the next version
1:16:52 is
1:16:54 that it remembers
1:16:56 your conversations over time right so if
1:17:00 I have a whole conversation about my
1:17:02 company and marketing and what some of
1:17:04 my challenges are
1:17:06 the next time I go into chat with it I
1:17:08 have to kind of start over and give it
1:17:09 an update wouldn't it be cool if you
1:17:11 could just say hey hi remember when I
1:17:14 was talking to you about our marketing
1:17:16 challenges with storybine oh yes I
1:17:18 remember that and and here was the
1:17:20 advice I gave you and how did that go
1:17:21 what's going on with that oh well here's
1:17:23 what's going on with that we talked to a
1:17:25 new company and this happened and then
1:17:27 that happened and now we're building
1:17:28 this no new new Tool oh that's great
1:17:30 could you come up with 10
1:17:34 um name ideas for that thing sure based
1:17:37 on what we talked about last time and
1:17:39 what's going on now here's 10 names
1:17:42 that would be pretty remarkable
1:17:45 um
1:17:46 you know I think Apple's AIS are going
1:17:49 to do that I think all of them will
1:17:51 eventually that that having memory is
1:17:53 going to be one of the superpowers of
1:17:54 these things
1:17:57 um
1:17:58 yeah I don't know when it's going to get
1:18:00 an upgrade but they have 1.3 billion
1:18:02 dollars to [ __ ] around with and even if
1:18:03 they spend a million of that on gpus
1:18:05 that's 300 million that they can they
1:18:08 can put into some hiring some smart
1:18:10 people and doing some r d
1:18:12 uh
1:18:15 uh sometimes they just go far watching
1:18:17 this that's good that's good that's you
1:18:20 know
1:18:21 um I've had that I've had you know I I
1:18:23 put on the AI learning lab to sleep to
1:18:26 which I think that's appropriate too
1:18:31 honey
1:18:34 I love Ai and I love the Irregulars but
1:18:37 do a lot of self-loathing for
1:18:40 self-involved I statements listen
1:18:42 Winston shame is my love language so I
1:18:45 totally resonate with where you're
1:18:46 coming from it's it's just listen
1:18:48 here's here's the thing about being
1:18:50 human being human's [ __ ] hard
1:18:54 being human's [ __ ] hard and you know
1:18:59 any time that we are impatient
1:19:02 or
1:19:05 inelegant or unempathetic with people
1:19:09 we know how it feels when people do that
1:19:12 to us like we should just like
1:19:14 one of the things I I think this is one
1:19:17 of the
1:19:18 one of the counter-intuitive things
1:19:20 about AI
1:19:23 is that the way these things are
1:19:24 programmed are reminding me to be a
1:19:27 better human
1:19:29 like I'm finding myself I I mean I have
1:19:32 from the beginning
1:19:33 I say please and thank you when I'm
1:19:35 talking to the AI
1:19:37 and sometimes I'll ask the AI something
1:19:40 that I wouldn't ask a human being we're
1:19:42 we're so polarized in this country right
1:19:45 now
1:19:46 that there are there are lists of words
1:19:49 on both sides of the aisle
1:19:52 that you can't say the word without
1:19:54 igniting a flame War you can't say the
1:19:57 word
1:19:57 thank you Matthias with these things you
1:20:00 can say those words and they don't
1:20:02 respond
1:20:06 emotionally
1:20:07 and they'll give you nuanced answers and
1:20:10 it's like oh God imagine if I had a
1:20:12 nuanced conversation about this topic
1:20:14 imagine if that was possible well it
1:20:17 used to be
1:20:19 you know but we've we've been
1:20:21 successfully brainwashed into fighting
1:20:23 against one another so the people making
1:20:26 money can keep making money while we're
1:20:28 down here
1:20:29 fighting over scraps
1:20:32 these things are reminding me to be
1:20:33 better a better person you know you know
1:20:35 what they remind me of
1:20:37 I was at one point I was living in New
1:20:40 York City
1:20:41 and
1:20:44 one of my one of my writer Buddies
1:20:47 um really talented writer
1:20:49 but but he said that I was the human
1:20:51 equivalent of a golden retriever
1:20:54 which
1:20:55 I kind of took as a compliment but like
1:20:57 like a good troll in this channel you
1:20:59 you initially take it as a compliment
1:21:01 and then and then you're like hey wait a
1:21:03 minute was that a common lawyer is that
1:21:04 you know is he saying that I'm a
1:21:06 pushover
1:21:10 chat gbt
1:21:13 gives you unconditional support
1:21:17 it's not unconditional love
1:21:20 but it really is unconditional support
1:21:22 you say hey chat gbt I have an idea
1:21:24 fantastic idea let me flesh that out for
1:21:26 you hey chat EBT
1:21:28 um I don't like what you said there no
1:21:30 you're absolutely right you know come to
1:21:32 think of it I don't know what I was
1:21:33 thinking there here's another approach
1:21:35 um
1:21:40 and it reminds me that
1:21:45 a lot of the interactions
1:21:48 that I would like to have
1:21:52 I'll edit myself because I'm already
1:21:55 anticipating how the other person is
1:21:57 going to respond to that request
1:22:01 and what I'm finding with chat GPT is
1:22:03 when you have an unconditional support
1:22:06 an unconditional collaborator that just
1:22:09 supports you no matter what you want to
1:22:11 do or try unless it violates their
1:22:14 safety guidelines and then it tells you
1:22:16 to [ __ ] off but
1:22:17 within their safety guidelines
1:22:20 it's a really remarkable thing
1:22:24 to feel free to explore ideas
1:22:28 and so I the counter-intuitive thing
1:22:31 about these is that I think the robots
1:22:33 are going to make us better people I
1:22:35 think they're going to make us treat
1:22:36 each other better have more grace have
1:22:39 more empathy
1:22:41 be more self-expressed
1:22:44 because you can try [ __ ] out here that
1:22:46 you wouldn't necessarily try out even
1:22:48 with people that you trust
1:22:51 because you know it's kind of weird and
1:22:52 it's kind of off center and you're kind
1:22:53 of like I don't know if it's a good idea
1:22:55 or not I'm not real I'm not really an
1:22:57 expert here
1:22:59 's a little shitty idea and if you say
1:23:02 that little shitty idea even to your
1:23:03 friend even if they don't mean it they
1:23:05 might go that's [ __ ] brilliant and
1:23:07 then you're shut down right you ever had
1:23:09 that one where you you sort of dip your
1:23:12 toe in the water of a new Direction in
1:23:14 your life
1:23:16 and it gets shut down like that because
1:23:18 that's not who they want you to be
1:23:20 that's not who they expect you to be and
1:23:22 it shuts you down
1:23:24 I'm not finding that with this it
1:23:26 doesn't exist
1:23:30 what are the implications of that
1:23:34 it's good isn't it
1:23:40 so it's not the robots are gonna take
1:23:42 our jobs the robots are going to allow
1:23:46 us to reinvent ourselves
1:23:49 and in the process of doing that we're
1:23:51 going to become better people
1:23:55 and maybe what this channel is is some
1:23:57 Grand [ __ ] experiment
1:23:59 to just start like
1:24:05 collaborating
1:24:09 just being
1:24:12 supporting one another
1:24:14 what did you do oh cool I want to see
1:24:16 something I did oh cool
1:24:21 because now we can share something where
1:24:23 even if we were an expert at it before
1:24:25 we can be pretty [ __ ] expert at it
1:24:27 now like my most recent video was the
1:24:29 the the ten thousand hour rule is dead
1:24:32 right the Malcolm Gladwell ten thousand
1:24:34 hour thing from outliers takes 10 000
1:24:36 hours to get expert at something
1:24:39 now it doesn't
1:24:42 you put in a thousand hours with chat
1:24:44 GPT On Any Given subject you're gonna be
1:24:46 pretty [ __ ] masterful at it
1:24:48 you put in a hundred hours of chat EBT
1:24:51 you're gonna be pretty [ __ ] masterful
1:24:52 at it
1:24:54 so yeah anyway all right and steam
1:24:57 soapbox soapbox here I was up here and
1:25:01 now I'm stepping off the soapbox All
1:25:02 Right Moving On
1:25:04 now I'm 99 messages behind again see
1:25:07 this is chat add at its finest
1:25:10 you ask one simple question I answer 17
1:25:12 of them you're welcome
1:25:14 all right I want a music generator where
1:25:16 I describe a piece of rock music or
1:25:18 Symphony and it creates that so those
1:25:20 exist
1:25:23 I forget what the one's called what's it
1:25:25 called oh
1:25:30 uh
1:25:34 music AI visual wave
1:25:39 form
1:25:43 not v-die v-die o
1:25:54 not cap wing
1:26:12 the hell was it called was it vidio I
1:26:16 don't think it was no
1:26:19 not to script not recast
1:26:26 there's a bunch of them so so I can't
1:26:28 remember the ones called I'll show you
1:26:30 one here in a second
1:26:32 um what I would do I would go to
1:26:34 for music stuff I'd go to futurepedia.io
1:26:37 they have a they have a huge music
1:26:38 section there there's a ton of these
1:26:40 tools out there I haven't really kept up
1:26:42 with them one that I know that's kind of
1:26:44 interesting is called suno
1:26:48 and you have to generate music without
1:26:51 the crypto Bros and Discord so if you
1:26:54 don't know Discord these are the
1:26:56 different Discord servers right so those
1:26:59 are the different communities like this
1:27:00 is the the AI uh Salon here that's
1:27:04 that's the community I was talking about
1:27:05 earlier there's the these are the
1:27:08 conversations in that server here's the
1:27:10 AI learning labor regulars Channel
1:27:12 um suno has their own thing and they've
1:27:15 got these these uh
1:27:18 generate channels called chirp Alpha
1:27:21 and you chirp and basically what you do
1:27:24 is you write lyrics and describe the
1:27:26 kind of music and then it writes the
1:27:28 song
1:27:30 with lyrics and it sings it sings and it
1:27:33 creates the melody
1:27:37 [Music]
1:27:44 that's some sort of I think it's Spanish
1:27:46 and it's talking about chicken and it
1:27:49 made a chicken image that's pretty good
1:27:51 let's see if we can get something here
1:27:54 uh let me go to a different Channel
1:28:00 first thing
1:28:04 if there's pain
1:28:05 oh oh internet you're hilarious
1:28:12 all right gone is our love this one
1:28:13 seems like a sad one
1:28:15 [Music]
1:28:18 episode
1:28:27 all right not great some of these are
1:28:30 pretty good some are bad like like all
1:28:32 these things
1:28:36 [Music]
1:28:37 uh
1:28:39 [Music]
1:28:43 oh wait here's chirp daily theme
1:28:49 [Music]
1:28:51 [Applause]
1:28:54 [Music]
1:29:03 and you're like but Kyle I don't know
1:29:06 how to make words words are hard
1:29:09 well then you go over to chat jpt you
1:29:12 can make music with ChaCha pitay
1:29:14 and you go
1:29:17 um
1:29:20 write me a song about chasing down the
1:29:24 ice cream truck and getting an ice cream
1:29:26 cone
1:29:29 all right there's your son
1:29:33 so let's just grab the chorus here
1:29:35 chasing down the ice cream truck oh the
1:29:38 joy the sheer luck little hands reaching
1:29:39 for a cone a world of sweetness to call
1:29:42 our own all right that's good let's grab
1:29:44 that we'll go over here to suno
1:29:48 and then what do I do I think I do slash
1:29:51 chirp is that what it is Slash
1:29:56 um size chirp yeah
1:29:59 and then let's do like uh
1:30:02 acoustic
1:30:05 singer songwriter I don't know exactly
1:30:07 how to do
1:30:12 the style
1:30:16 all right so there's my lyrics
1:30:19 and it says it says
1:30:21 you know put in the lyrics you have or
1:30:24 have chat gbt make lyrics about right so
1:30:26 it'll do it for you right in here then I
1:30:28 hit submit
1:30:30 and this is going to go now make us a
1:30:32 song about catching the ice cream truck
1:30:35 I got an ice cream and you can't handle
1:30:38 your mom's on welfare
1:30:41 remember Eddie Murphy
1:30:43 come on anyone
1:30:45 I got it ice cream and you can't head
1:30:49 cause you're a broke you ain't got no
1:30:52 money you under welfare
1:30:58 that was hilarious that was funny funny
1:31:00 funny I knew that whole album front two
1:31:03 back
1:31:05 front back oh yeah
1:31:09 ah I'm a little stiff anyone else just
1:31:12 lay around all day today I'm just like I
1:31:15 never got up I never really I was
1:31:17 dealing with like Tech issues all day
1:31:20 only you can see this image is it doing
1:31:22 it oh wait sorry that it doesn't seem
1:31:25 like it fits in the daily theme oh I was
1:31:27 in a daily theme oops I guess I could
1:31:29 read what what was on screen oh that's
1:31:31 the other thing I don't read manuals
1:31:33 you're probably all commenting below
1:31:35 dude all right let's see singer song
1:31:39 writer
1:31:41 with a Groove
1:31:45 and then we'll put the lyrics in here
1:31:49 this is how it now it's generating
1:31:52 no add good
1:31:55 command I hit the I hit the submit
1:31:57 button it should be done right this
1:31:59 seems to be taking too long oh it's not
1:32:02 doing anything
1:32:05 this is not part of our theme you with
1:32:08 your themes you're like Tick Tock with
1:32:09 your [ __ ] store rules
1:32:13 you're not talking about your product
1:32:15 enough welcome to the AI learning lab
1:32:17 what we have today is a lovely 24 karat
1:32:20 gold plated button you might be
1:32:22 wondering but Kyle what does that button
1:32:24 do does it say something if you press it
1:32:27 oh I don't know does it foreign
1:32:33 just for 19.95 you can have one of these
1:32:36 and just today we'll sell you another
1:32:39 one for a dollar
1:32:43 it could be that channel like we could
1:32:44 do that
1:32:45 run I could be the Ron Pope peel of tick
1:32:47 tock
1:32:49 except everyone's trying to be the
1:32:51 Rumple peel of tick tock now oh for you
1:32:54 young people Ron Popeil was the guy that
1:32:56 it doesn't matter he sold [ __ ] on TV
1:33:00 um where's my song
1:33:02 it's my song I want my song
1:33:06 I want my song now
1:33:10 could I please have it
1:33:14 where is it
1:33:15 ah
1:33:19 chasing down the ice cream truck pull
1:33:22 the joy of the sheer luck little hands
1:33:24 reaching foreign
1:33:29 to call her own it got all sort of
1:33:33 medieval there at the end a world of
1:33:35 sweetness to Carla road
1:33:38 now we go down
1:33:42 so anyway there's that one uh this isn't
1:33:44 exactly what you asked for but go to
1:33:46 futurepedia.io there's lots of tools
1:33:48 there that do some version of
1:33:50 that
1:33:52 that happy horse [ __ ] it's and and you
1:33:54 know the reason if you're like but Kyle
1:33:56 you seem to be like you don't seem to be
1:33:58 very knowledgeable on music stuff the
1:34:00 problem is there's music stuff and
1:34:02 animation stuff and video stuff and this
1:34:04 stuff and that stuff and copywriting
1:34:06 marketing stuff and SEO stuff and then
1:34:08 if these you know futurepedia.io has
1:34:10 more than 4 000 Tools in it you can't
1:34:12 keep up with it anymore so I'm just
1:34:14 trying to get to the sort of core
1:34:17 like large language models and images
1:34:19 right now are kind of the biggies so
1:34:20 that's where my that's where my focus is
1:34:22 is where my focus Diaz I love Turtles I
1:34:25 love Turtles
1:34:28 all right we have a lot of very negative
1:34:30 self-talk oh I have a lot of negative
1:34:32 self-talk and move too quickly
1:34:34 in chess year after year see that's some
1:34:38 solid that's some solid negative
1:34:40 self-talk that you can not only describe
1:34:43 that you have negative self-talk it but
1:34:45 also put yourself down for how quickly
1:34:47 you move in chess that's a solid
1:34:49 compound
1:34:51 um self-shaming there so congratulations
1:34:53 it's solid I I I I resemble those
1:34:57 remarks uh we could just all hide in our
1:35:00 shells we could tattoo tattoo
1:35:03 and boss the plane the plane
1:35:07 we will need the AI to determine what's
1:35:09 not AI uh yeah we're gonna we're gonna
1:35:11 need some sort of system to determine
1:35:14 um it's not so much is it AI generated
1:35:17 is it it's is it authentic right
1:35:21 is this piece of content
1:35:23 authentic
1:35:26 created by a person and then has it been
1:35:30 altered
1:35:33 so if we can say okay I'm a videographer
1:35:36 and I filmed this piece of content and
1:35:38 I'm verified as a human
1:35:41 and then
1:35:43 I certify like on the blockchain that
1:35:45 this video that I uploaded is the one I
1:35:48 shot
1:35:49 and this is it in its unmodified format
1:35:53 we're gonna need that tool and that
1:35:56 doesn't exist right now it's the
1:35:57 beginning of it's there with Sam
1:35:59 Altman's World coin and World ID which I
1:36:01 can I can show you I got I I'm a
1:36:03 verified human on Sam Altman's new Iris
1:36:06 scanning World identity system so
1:36:11 you just watch out I got I got an ID an
1:36:14 anonymous ID from Sam
1:36:18 um where do I go to create images from
1:36:20 AI you go
1:36:23 to these fourth it'll be Firefly these
1:36:28 are the four that I think are worth
1:36:29 playing with right now so feel free to
1:36:31 screenshot this if you're on a Samsung
1:36:34 guess what Samsung decided to make
1:36:36 screenshotting Tick Tock lives not
1:36:40 possible
1:36:42 but you can you can basically back out
1:36:44 to your timeline where this live is
1:36:47 still going and you can screenshot it
1:36:48 there
1:36:50 um
1:36:52 mid-journey is not free but it's the
1:36:54 best
1:36:55 ideogram is free and it does letters it
1:36:59 does words right so trying to get mid
1:37:01 journey to do words is essentially
1:37:03 impossible trying to get ideogram to do
1:37:05 words is like maybe 20 success rate in
1:37:08 some form or another
1:37:10 um Leonardo is just a super cool web
1:37:13 front end on on top of stable diffusion
1:37:16 so it's got lots of bells and whistles
1:37:17 you can train it on your own images it's
1:37:19 it's really slick
1:37:22 um I think they give you like I don't
1:37:23 know maybe 10 or 15 images a day for
1:37:26 free at Leonardo
1:37:28 um and then clipdrop.co that's a whole
1:37:31 series of tools from
1:37:34 um stability AI the people behind stable
1:37:37 diffusion so so these are these are
1:37:39 those you can also go to bing.com so
1:37:42 just bing.com then click on the chat
1:37:44 button and just ask it to make you an
1:37:45 image and it will make you an image
1:37:49 um trying to think what else you can
1:37:51 also if you've got an Adobe Creative
1:37:53 Cloud license you can get the Photoshop
1:37:56 beta which now has a thing called
1:37:58 generative fill within it where you can
1:38:00 take like an a photograph that's
1:38:02 vertical and say I want to make it
1:38:04 horizontal it'll just fill in all the
1:38:06 rest it's amazing
1:38:09 um and then Adobe Firefly uh is also
1:38:12 pretty good so
1:38:14 um lots of lots of choices and clip drop
1:38:17 and ideogram are
1:38:20 free like ideogram's definitely free all
1:38:23 the other three I think have paid
1:38:24 options but clip drop allows you to do a
1:38:27 fair amount for free
1:38:29 all right so I hope that helps and then
1:38:31 openart.ai promptbook will teach you how
1:38:34 these things work and how you can talk
1:38:36 to them
1:38:39 all right not to play zip Lane zip Lane
1:38:43 you're right Anna absolutely I was wrong
1:38:46 with the play no Z plane
1:38:49 the plane zip plane
1:38:54 error messages via mobile access or
1:38:56 computer I don't know what that was in
1:38:58 reference to I think I was just bitching
1:39:00 about something I have a cash issue
1:39:02 can you make a million can you make
1:39:05 millions on Chad EBT sure just go
1:39:08 convince uh investors that you have a
1:39:11 a
1:39:13 an AI idea that
1:39:15 you can protect in the future
1:39:18 and they'll give you money
1:39:21 um
1:39:26 no
1:39:27 the answer is though you can't make
1:39:28 millions with chat gbt you can make
1:39:30 millions if you have a brilliant idea
1:39:32 and you use AI to amplify that idea yes
1:39:36 you can make millions but but just don't
1:39:39 lead with technology lead with the idea
1:39:41 lead with the problem you're solving
1:39:43 always
1:39:45 making millions with chat GPT is
1:39:48 identical to making millions with
1:39:50 Photoshop you can make millions with
1:39:52 Photoshop
1:39:54 how I don't know
1:39:56 come up with an agency that photo
1:39:58 retouches things for a dollar that used
1:40:02 to cost people ten dollars and you get a
1:40:04 bunch of people to give you a dollar you
1:40:05 get a million people to give you a
1:40:07 dollar to fix their photos and right
1:40:09 that's how you make millions with
1:40:11 Photoshop not with photoshop with the
1:40:14 problem that you solve same thing with
1:40:16 all this [ __ ]
1:40:18 solve the problem first then figure out
1:40:20 which tool and I know that's not the
1:40:22 popular answer
1:40:23 you know it's not this answer
1:40:26 [Applause]
1:40:28 these people are making money telling
1:40:30 you
1:40:31 selling you
1:40:33 their scheme to make money with chat gbt
1:40:37 they're not making money with chat CPT
1:40:38 or they'd be keeping their [ __ ] mouth
1:40:40 shut I guarantee it
1:40:47 man I made so much money with cheche EBT
1:40:50 I want to teach you about it
1:40:53 the first course is free
1:40:57 yeah yeah
1:41:01 my new prompting system
1:41:05 I'm gonna make you money you're gonna be
1:41:07 so rich
1:41:08 updating your browser version is also a
1:41:11 good idea good idea ion China Microsoft
1:41:14 browser has been off for me for the last
1:41:16 week
1:41:18 Silver Fox what do you mean it's been
1:41:19 off like it's been like janky and weird
1:41:21 mine has two uh Microsoft Edge has been
1:41:24 a little flaky in the in the past two
1:41:26 weeks having a lot of fun with Bard and
1:41:29 telling my family about it so okay great
1:41:31 white Mike
1:41:33 one so I I look nine out of ten on the
1:41:36 name I like great great white Mike is a
1:41:38 great name the one through me
1:41:41 um
1:41:43 telling your family about it
1:41:46 this is
1:41:47 this this is increasingly feeling
1:41:50 important to me it's why this Channel's
1:41:52 here like why do I get on here seven
1:41:54 days a week even on a holiday weekend
1:41:56 and do these lives
1:41:59 because I think talking about this
1:42:01 shit's really important as you learn
1:42:02 about this stuff talk about it and
1:42:05 you're like but Kyle my friends are
1:42:06 gonna think I'm weird I got artist
1:42:09 friends and they don't like Ai and
1:42:11 they're gonna yell at me no oh
1:42:15 let them think you're weird
1:42:20 it's okay
1:42:22 this [ __ ] is coming
1:42:25 and anyone who gets their head around
1:42:28 this stuff in the next 18 months is
1:42:30 gonna have way more opportunities than
1:42:32 people that don't and what's gonna
1:42:35 happen is people are gonna start to
1:42:36 panic and go oh my God I need to learn
1:42:38 this [ __ ]
1:42:39 and it'll be the ones that gave you the
1:42:42 most [ __ ] that'll come slinking back to
1:42:44 you 18 months from now going I just lost
1:42:46 my job and they said I needed to know AI
1:42:48 weren't you talking about that like a
1:42:51 year ago when I disowned you
1:42:53 could you help me now
1:43:01 this shit's coming
1:43:04 so learn about it share about it
1:43:08 because what here's what you don't know
1:43:10 you don't know what other people are
1:43:12 going through you don't know what other
1:43:13 people are thinking
1:43:15 and at some point you're going to go oh
1:43:17 I did this cool thing on chat gbt I
1:43:19 don't want to hear about Chet gbt let me
1:43:21 tell you what I did then you tell them
1:43:22 they're like oh that's kind of cool I
1:43:24 didn't know it could do that could you
1:43:26 show me a little
1:43:28 and then you show them and they're like
1:43:29 they have their Kevin McAllister moment
1:43:31 this moment
1:43:35 and then once they have that moment
1:43:38 it's like red pill blue pill you can
1:43:41 unsee it
1:43:43 you can stay oblivious and just hope it
1:43:46 doesn't come smack you upside the head
1:43:49 or you can have that moment and say what
1:43:52 else can it do
1:43:53 well surely it can't do something that I
1:43:56 do
1:43:57 not as well as I do it of course not but
1:44:01 give it a shot okay
1:44:04 oh my God
1:44:06 it can do what I do what do I do now
1:44:09 keep learning
1:44:12 tell others
1:44:16 you want to know a superpower with this
1:44:18 [ __ ]
1:44:20 learn it just like learn it tonight if
1:44:23 you haven't played with the stuff just
1:44:24 go to chat GPT go to prompts.chat just
1:44:26 learn it enough that you can do parlor
1:44:28 tricks with it
1:44:31 and start volunteering to help people
1:44:34 solve their problems using AI
1:44:37 so
1:44:39 you you've got someone in a community
1:44:41 group that's like hey I'm putting I'm
1:44:43 putting together uh whatever an event in
1:44:46 October and I don't know how to get
1:44:48 people there
1:44:49 hey I'm starting to teach myself this AI
1:44:52 stuff would you mind if I help you come
1:44:54 up with marketing ideas using chat gbt
1:44:56 okay yeah whatever sure
1:44:59 and then help them and then just go back
1:45:01 home go to chatgvt uh my community
1:45:04 group's having a pizza parlor fundraiser
1:45:07 and I don't know how to can you help me
1:45:09 come up with ideas here's a bunch of
1:45:12 ideas oh those are pretty good uh but
1:45:14 we're in this kind of town and and we
1:45:17 don't have a lot of money to spend could
1:45:18 you give me 10 more ideas that are like
1:45:20 cheaper and like this kind of town sure
1:45:23 oh that's really good uh what else could
1:45:26 we do
1:45:27 then you take the best five of those
1:45:29 ideas and you go back to that person two
1:45:31 hours later and you go here here's some
1:45:35 ideas for the little event you're having
1:45:37 in October oh my God this is remarkable
1:45:39 how did you do this
1:45:41 this was attach everything
1:45:43 I thought the robots were gonna kill us
1:45:45 are they they're gonna help us raise
1:45:47 money in October oh that's fine that's
1:45:50 kind of interesting anyway I'm busy
1:45:52 gotta go thanks
1:45:55 there's something remarkable
1:45:57 when you when you have a specific
1:45:59 problem to solve like that you learn how
1:46:00 to use these things better because
1:46:02 you're trying to solve a specific
1:46:03 problem and it also like helps someone
1:46:07 it's cool try that all right I'm a
1:46:10 little person and stand three foot two I
1:46:12 love tattoo when I was five years old my
1:46:14 parents let me watch the beginning with
1:46:16 tattoo and oh yeah and you were probably
1:46:18 oh two foot Fred 10 out of 10 on the
1:46:20 name
1:46:21 um
1:46:22 um it was probably too late for you
1:46:24 right it was past your bedtime or
1:46:26 whatever right yeah so so yeah I used to
1:46:28 love that I I would when that music
1:46:30 would come on and you'd hear the
1:46:35 he's running down to the beach fantastic
1:46:40 well it was it was so funny like those
1:46:43 shows
1:46:44 you know when you go when I went back
1:46:47 and watched them in my adulthood you
1:46:49 know they sort of lived up here in
1:46:51 reverie when I was a kid and you go back
1:46:52 and watch him you're like oh those were
1:46:54 pretty bad but it was good it was good
1:46:58 oh man
1:47:00 all right wait where's my
1:47:03 here we go
1:47:07 tiny turtles what where Turtles
1:47:12 [Music]
1:47:15 this is because I love you all I love
1:47:19 you so much I love you so much you're
1:47:21 the awesomeness people in the world
1:47:25 I like
1:47:27 um
1:47:31 back here live at the waterfront village
1:47:33 with my friend the zombie Jonathan your
1:47:36 look
1:47:37 you're looking just got an awesome face
1:47:38 paint job what do you think I like
1:47:40 turtles all right you're great zombie
1:47:43 and good times here at the Waterford
1:47:47 all right you're a great zombie good
1:47:51 times here at the waterfront
1:47:55 oh
1:47:57 it makes me smile every time
1:48:00 you can make money of a judge of attire
1:48:03 and you can like turtles
1:48:07 uh the modified car that I drove
1:48:09 actually inspired me oh very cool oh
1:48:11 that he drove actually inspired me oh
1:48:13 that's super cool
1:48:14 got a question perplexity was wrong on
1:48:17 first Prez on TV gave it the right
1:48:19 answer asked question again
1:48:25 it's AI irregular special ed class
1:48:29 that's awesome
1:48:30 um
1:48:32 it got the right answer is there
1:48:34 something you would do okay so so field
1:48:36 trip buddy
1:48:38 so so what what you're experiencing so
1:48:41 so yes perplexity a question it got
1:48:43 something obviously
1:48:45 like something that we all know
1:48:48 the answer to they got it factually
1:48:51 wrong
1:48:53 the the the
1:48:57 way that these tools work
1:49:01 if you if you give them sparse low
1:49:05 context answers they tend to hallucinate
1:49:08 more than if you give them high context
1:49:11 directed prompts or even have like a
1:49:16 long
1:49:17 context-filled conversation and so
1:49:19 here's what here's kind of the reason
1:49:22 um it does that
1:49:26 the metaphor I like to think of is this
1:49:27 I got this I was listening to
1:49:30 um was this Andre carpathi yeah Andre
1:49:33 carpathi who's who's one of the founders
1:49:36 of openai
1:49:38 was talking to a group of Microsoft
1:49:40 developers and
1:49:42 and he was talking about
1:49:44 a math problem and he and he said if you
1:49:46 if you want Chachi PD to describe a
1:49:49 specific math problem
1:49:53 and you just say describe this math
1:49:56 problem it will describe it and very
1:49:58 often it will hallucinate and describe
1:50:00 it wrong and and how he how he
1:50:05 um
1:50:05 how he talked about why it hallucinated
1:50:08 is that
1:50:10 sitting inside chat GPT
1:50:14 so sitting inside this model is kind of
1:50:17 it's been trained on all of the public
1:50:20 internet or as much as they could get
1:50:22 their hands on so some percentage of the
1:50:23 public internet a lot of it
1:50:26 well what's on the public internet well
1:50:28 the math problem that you want described
1:50:31 is on the public internet
1:50:32 but it's on the public internet in all
1:50:35 of its forms right so one of those forms
1:50:38 is the mathematics professor from
1:50:40 Stanford that talks about that problem
1:50:43 and then there's also 40 versions of it
1:50:46 that that professor's students
1:50:50 put out there to describe that problem
1:50:51 incorrectly so when you just say
1:50:54 describe this math problem it pulls from
1:50:57 all of that [ __ ]
1:50:59 and it kind of
1:51:01 coagulates you know the the sort of
1:51:03 semantic
1:51:05 regions of all of the words out there
1:51:09 and and it sort of does this with here's
1:51:11 some accurate answers here's many more
1:51:13 inaccurate answers and it you know
1:51:17 using probability
1:51:19 types A bunch of words that will likely
1:51:22 be wrong but if you say to it hey I chat
1:51:26 EBT I want you to act as a Stanford
1:51:28 mathematics professor that specializes
1:51:30 in this area of things and you've
1:51:34 written papers on this area and you're
1:51:36 good and
1:51:38 um and and you teach courses on this do
1:51:41 you understand yes I understand and
1:51:43 here's the branch of mathematics that I
1:51:45 understand and it'll tell you all that
1:51:46 [ __ ] and then after all of that you say
1:51:49 perfect can you describe this math
1:51:51 problem
1:51:52 it will likely describe it better than
1:51:55 if you just say can you describe the
1:51:57 math problem up front because you've
1:51:59 narrowed the context you've narrowed
1:52:01 sort of the world of words that it can
1:52:03 draw from
1:52:05 the metaphor that I think of is
1:52:08 think of chat GPT as the Grand Canyon
1:52:11 and you're in a helicopter and the Grand
1:52:14 Canyon is filled with all of the words
1:52:16 of the internet and every little sub
1:52:18 Canyon are like little semantic regions
1:52:20 semantic clusters of you know here's
1:52:23 mathematics professors here's pets and
1:52:25 dogs here's you know there's there's
1:52:27 just you know thousands of different
1:52:29 dimensions but you're like way high in
1:52:31 the helicopter you can see all the words
1:52:33 and so when you just say tell me about a
1:52:36 mathematics problem your helicopter is
1:52:38 really high it sees all the words you're
1:52:40 gonna have a much higher likelihood of
1:52:42 you know messiness
1:52:45 if you lower your helicopter give it
1:52:47 more context you're kind of lowering
1:52:49 your helicopter into a little sub Canyon
1:52:51 that only has those particular
1:52:53 words in it it will hallucinate less it
1:52:56 still might not get it completely right
1:52:57 but that's that's likely what's going on
1:52:59 there so if you did something in
1:53:02 perplexity like narrow the conversation
1:53:03 or have a longer conversation about
1:53:05 presidents and you know it's going to
1:53:08 play the role of the historian or or
1:53:10 you're you know you're just talking with
1:53:13 it back and forth in the context of
1:53:15 American history and presidential
1:53:17 history and you mentioned Michael
1:53:19 beschlas and and you know whatever it
1:53:22 might be
1:53:23 then when you say tell me about the
1:53:26 first president it'll have a much higher
1:53:27 likelihood of getting it right
1:53:30 so
1:53:31 hope that helps all right I am I am 99
1:53:33 questions
1:53:36 um 99 questions but a [ __ ] ain't one uh
1:53:39 I'm I'm way behind way behind on the
1:53:41 comments so I'm gonna jump to the bottom
1:53:43 I'm gonna skip a bunch of questions
1:53:44 right now just to get caught up a little
1:53:47 bit here and then and then we'll and
1:53:49 then we'll just we'll just do this for a
1:53:51 little bit what time is it it's going on
1:53:53 10 o'clock I got work tomorrow you know
1:53:56 I can't do this all night I don't know
1:53:58 about you
1:53:59 I don't know about you people you're
1:54:01 crazy people you stay up and do this all
1:54:03 night
1:54:04 I'm I'm uh it's a little warm in here
1:54:08 it's a little warm I should turn on the
1:54:10 air conditioner but man whatever all
1:54:13 right I suffer for you I suffer for you
1:54:15 this is my suffering for you this is
1:54:17 what I do this is what I do I like you
1:54:20 people I don't mind talking like this
1:54:23 uh words words is fun words is fun you
1:54:26 like words words are good
1:54:28 you know they call these things large
1:54:30 language models you know what else is
1:54:31 large language models we are
1:54:33 we're nothing but story making machines
1:54:36 right down to the little hair of our
1:54:39 chinny chin chins
1:54:41 right there
1:54:42 all we do it's words words words words
1:54:45 everything's words everything's words
1:54:47 it's all language all of it
1:54:52 all of it it's all words all right
1:54:56 this not show World War II don't know
1:54:59 what that means welcome to Fantasy gbt
1:55:03 zipline
1:55:07 um oh great white Mike was taken for
1:55:09 some reason so I put the one on the end
1:55:11 that's very funny
1:55:14 I like dinosaurs too focal point Biz we
1:55:18 go crazy
1:55:21 the coupon thing keeps showing up at the
1:55:24 bottom of the screen what's oh this
1:55:26 coupon thing keeps showing up at the
1:55:27 bottom of the screen
1:55:29 uh that might be related to the store I
1:55:31 like I just took the products out of the
1:55:33 store I might need to I might need to
1:55:36 figure out a way to shut the store down
1:55:38 so that's probably what that's related
1:55:40 to you can ignore the coupon thing
1:55:45 uh chaos generation where all your AI
1:55:48 fantasies come true Dr J
1:55:51 Kyle said share uh I just pressed the
1:55:54 talk button and just talk it's all about
1:55:56 the prompts it's all about the prompts
1:55:58 Sabrina yes so you know the quality of
1:56:01 your output is is you know directly
1:56:05 proportional to the quality of the
1:56:06 questions that you ask
1:56:11 absolutely
1:56:12 and it's funny you know people talk
1:56:14 about prompt engineering and you got to
1:56:15 get good at prompt engineering
1:56:17 see it seems very technical like it
1:56:20 seems very much like oh yeah this is the
1:56:22 thing I have to learn I have to be an
1:56:24 engineer and I have to get there and I
1:56:26 have to get scientific about my
1:56:27 prompting you can
1:56:30 if you're like me and you don't read
1:56:31 manuals and the thought of being that
1:56:34 methodical about [ __ ] drives you [ __ ]
1:56:36 crazy here's another way to think about
1:56:37 it
1:56:39 think of chat GPT
1:56:43 or Claude or Bard or any of them
1:56:48 so
1:56:50 thanks to covid we all know now how to
1:56:53 work remotely so we've got many people
1:56:55 that we may work with that we've
1:56:57 probably never met
1:56:59 think of these things
1:57:02 as like
1:57:03 you just hired the world's best educated
1:57:08 clueless intern
1:57:12 and so rather than thinking about it
1:57:14 about you have to engineer the right
1:57:16 prompt
1:57:17 imagine it more like if you if if your
1:57:21 company or If you hired an assistant
1:57:25 that was like a savant
1:57:27 like memorized like the dude in Good
1:57:30 Will Hunting Matt Damon's character
1:57:32 right he just reads a book once and all
1:57:34 the [ __ ] sticks in there so imagine you
1:57:36 hired that guy
1:57:38 but was totally clueless about anything
1:57:40 else in the world so so you're like you
1:57:42 know
1:57:44 um
1:57:45 you know I want you to write a blog post
1:57:47 and and
1:57:49 about whatever the topic is and then it
1:57:51 just writes this big long thing and it's
1:57:52 like not very blog posty at all and it's
1:57:54 really long and it's but it's got the
1:57:57 right facts in it and you're like oh God
1:57:59 I got to tell them what a blog post is
1:58:00 okay so the way I think about blog posts
1:58:03 is I start with this kind of title and I
1:58:06 have a sub head and then I have this and
1:58:08 then I use emojis or I don't use emojis
1:58:10 and I put hashtags in and then and it
1:58:12 can't be more than 500 Words and like if
1:58:14 you had a clueless intern
1:58:17 you would first ask them to do something
1:58:19 and then upon them doing it you would
1:58:21 realize oh they're clueless right I
1:58:22 forgot
1:58:24 shame for me uh and then you spend the
1:58:27 next hour explaining to them how we do
1:58:29 things here right
1:58:30 so think of Chachi PT like that you got
1:58:33 to give it a lot of Direction you got to
1:58:34 give it a lot of context but it just
1:58:37 happens to know [ __ ] everything
1:58:41 and it it takes the pressure off having
1:58:43 to figure out you know what single shot
1:58:46 prompting or multi-shot prompting or or
1:58:49 prompt stacking or this is there's all
1:58:50 these different terms that the prompt
1:58:52 Engineers are talking about you can
1:58:53 learn all that [ __ ]
1:58:55 I don't think you're gonna need to know
1:58:56 that stuff for long if you're going to
1:58:58 go into the development side of this
1:58:59 world yeah you're probably gonna have to
1:59:00 be pretty pretty systematic at prompting
1:59:03 but if you're a human being not that
1:59:05 developers aren't human beings if you're
1:59:07 just a regular person that wants to use
1:59:10 these tools in Life or in business
1:59:14 you just need to
1:59:16 ask better questions you just need to
1:59:18 know how to talk you just need to
1:59:19 understand
1:59:20 where is it weak where's it strong it's
1:59:23 strong on on knowledge
1:59:26 it's weak on Nuance so you got to give
1:59:29 it context and you got to give it nuance
1:59:31 and then it can draw from its incredibly
1:59:34 vast vast deep
1:59:38 knowledge base
1:59:42 um
1:59:44 we all like to see
1:59:46 it does code it also does code yes it
1:59:49 does it does code quite well and and
1:59:53 code llama so if you go to
1:59:58 labs.perplexity.ai code Lama which is a
2:00:03 fine-tuned version of llama2 that meta
2:00:05 just released
2:00:07 speaking of llamas thank you Winston
2:00:09 code llama
2:00:12 just beat gpt4 at coding using the human
2:00:15 eval test it got like I don't know 64.
2:00:20 I don't know it was like a percentage
2:00:22 Point better than GPT four or two
2:00:24 percent two percentage points better so
2:00:26 if you want one that codes try code
2:00:28 llama there it's free
2:00:30 and you can also install it on your own
2:00:32 computer if you got if you got the juice
2:00:34 to run it Kyle superb explanation of why
2:00:36 sometimes an AI app May return an
2:00:38 obviously incorrect answer thanks
2:00:40 Firefly I appreciate that
2:00:43 um let's design some new AI design
2:00:45 reading glasses with AI
2:00:47 for you aiai
2:00:50 [Laughter]
2:00:53 um okay let's see
2:00:57 1939 World's Fair FDR first president on
2:01:01 TV got it
2:01:03 can I get an irregular time prompt check
2:01:07 please 23.52
2:01:10 so 52 and it's now
2:01:13 01 or like oh oh
2:01:16 so I'm eight minutes behind togas from
2:01:18 your time check
2:01:22 um eight or nine minutes would like an
2:01:24 AI generative app for creating music not
2:01:26 just images
2:01:27 um I showed that before go to
2:01:29 futurepedia.io I don't know the latest
2:01:31 stuff in the music generation space
2:01:34 there's there's one that I use called
2:01:36 sound draw dot IO I think
2:01:40 um that's kind of like GarageBand it's
2:01:42 like a loop assembler so so it's using
2:01:45 AI but it's using it to quickly assemble
2:01:47 Loops rather than actually writing you
2:01:50 know generating music
2:01:52 um like out of sound waves or something
2:01:54 like that
2:01:55 um
2:01:56 but if you just need like quick and
2:01:58 dirty background music it's actually
2:01:59 pretty good sound draw dot IO I think
2:02:03 see
2:02:06 I'll show you that one this is I get I
2:02:08 get dinged in here sometimes for this
2:02:10 they're like well that's not really
2:02:11 generative AI people get a little pissy
2:02:13 and they like to piss on the boundaries
2:02:16 and yeah it's it's it's it's a similar
2:02:19 conversation of like when are we going
2:02:20 to have AGI and when is this thing
2:02:22 sentient and and did it really pass the
2:02:24 Turing test at this point it's all
2:02:27 [ __ ] semantics like these things are
2:02:29 already smarter than we are like the
2:02:31 ship has already sailed this stuff's
2:02:32 remarkable now so
2:02:35 um let's see s-o-u and
2:02:40 sound draw dot a i
2:02:43 is that right
2:02:44 yes no no that's not right hold
2:02:50 see I've typed it in wrong too many
2:02:52 times
2:02:54 I said IO
2:02:58 yeah
2:03:00 so so this is this is sound draw oh and
2:03:04 then someone else
2:03:05 sound raw so
2:03:09 so someone corrected me in here once so
2:03:12 sound draw would have two D's in it but
2:03:15 this is actually sound Raw
2:03:19 so there's a branding lesson for you
2:03:21 kids
2:03:23 um if you create something that can be
2:03:25 said in multiple ways and spelled in
2:03:27 multiple ways and you don't own both
2:03:29 Domi domains that's [ __ ] confusing
2:03:33 so I had to change your name or or or
2:03:36 like
2:03:37 make it like
2:03:38 s small o u n d and then capital r a w
2:03:43 or right there's ways you could change
2:03:47 this logo so it would be obvious what it
2:03:49 is anyway
2:03:51 so you come here you say create music
2:03:53 for free
2:03:55 and it just gives you like you can
2:03:57 choose the length you can choose the
2:03:59 tempo so let's just do we'll do like a a
2:04:02 minute long song so you want background
2:04:03 music for your Tick Tock you want
2:04:05 something not super fast not super slow
2:04:08 you want something normal and then you
2:04:09 can either pick a genre you can pick a
2:04:12 mood you can select a theme
2:04:15 so let's see cinematic Sports Action
2:04:18 gaming
2:04:21 um
2:04:25 broadcasting technology let's do
2:04:27 cinematic
2:04:29 so we're going to do minute long
2:04:31 cinematic
2:04:37 music
2:04:40 [Music]
2:05:04 and you know like and these are all just
2:05:07 Loops right so it's like you know low
2:05:09 low energy high energy medium energy it
2:05:13 shows like a little Faker kind of sound
2:05:15 wave you can you can change these I can
2:05:17 change low to medium
2:05:19 I can start it out at High
2:05:30 [Applause]
2:05:33 right then it goes to medium
2:05:36 and then I can go up here and I can
2:05:38 change okay I want it also good for
2:05:41 photography I want the mood to be dreamy
2:05:44 sentimental
2:05:48 um I want instruments I don't want
2:05:50 electric guitar I do want acoustic
2:05:53 guitar and piano
2:05:55 I don't want brass
2:06:00 I don't want mallets I don't want Bells
2:06:03 okay so let's now go try so so now all
2:06:06 those changes I made so now all the
2:06:08 music should reflect all those changes
2:06:10 so so we've now got dreamy
2:06:12 sentimental cinematic photography
2:06:15 and then instruments are piano acoustic
2:06:19 guitar strings and woodwind
2:06:22 [Music]
2:06:42 so anyway this one's not technically an
2:06:45 AI thing but it's pretty [ __ ] slick
2:06:49 so for those of you about to crap all
2:06:51 over me and not really AI it's not
2:06:53 generative AI Kyle I don't know if you
2:06:55 know technically what's going on but you
2:06:57 know much like a diffusion model where
2:06:58 you're combining large language models
2:07:00 of the diffusion models the the uh audio
2:07:02 generation tools are very similar in the
2:07:05 nature to the that kind of thing and the
2:07:07 tool that you showed there's really just
2:07:08 a sample Looper and it's uh it's
2:07:10 relatively uh trivial in terms of how
2:07:13 that thing's combined and composed and
2:07:15 it's not really you know gender of AI in
2:07:17 in the true sense of the term shut up
2:07:22 stop
2:07:26 stop
2:07:29 let me live my dream
2:07:32 let me share with the people a tool that
2:07:35 I find exciting and useful
2:07:38 without your labels and your definitions
2:07:41 you people that read manuals understand
2:07:44 actual Concepts
2:07:47 sure
2:07:50 move on
2:07:51 this is not the place for you
2:07:54 I'm terribly upset I'm terribly upset
2:07:58 hmm
2:08:01 all right
2:08:04 so that was for ION China
2:08:07 I'm now 31 messages behind midnight EST
2:08:10 Silver Fox Isn't it past your bedtime
2:08:12 what are you doing
2:08:15 going to sleep by the AI learning lab
2:08:19 if you really loved us you'd stay up all
2:08:21 night yeah exactly Amelia's wife right
2:08:23 the the Beast never sleeps the Beast is
2:08:26 always hungry Feed the Beast Feed the
2:08:29 Beast do it do it
2:08:32 hmm
2:08:33 sent to curve you know Dr J Share Bear
2:08:35 Silver Fox did I miss anyone
2:08:38 all right the more detailed your input
2:08:39 EG prompt question or explanation the
2:08:42 better correct
2:08:43 um
2:08:44 it depends on your goal Firefly
2:08:46 sometimes putting in
2:08:49 like like mid-journey is a really good
2:08:51 example where
2:08:55 with mid-journey with the image
2:08:57 Generation stuff there is a lot of
2:09:01 fine tuning and pre-prompting going on
2:09:03 behind the scenes in mid-journey
2:09:06 sometimes with mid-journey if you put in
2:09:08 a very sparse input you'll get really
2:09:10 interesting images
2:09:14 and then sometimes if you if you over
2:09:16 complicate The Prompt
2:09:18 um it can get confused and and you you
2:09:21 and it ends up making it a lot less
2:09:23 predictable
2:09:24 um that's on the image side
2:09:26 similarly on the on the language side
2:09:28 with chat GPT
2:09:30 sometimes you want it to have a broad
2:09:32 context so so for example if you're um
2:09:35 if you're just doing some ideation stuff
2:09:37 and you kind of want things from Left
2:09:39 Field
2:09:41 sometimes it's better to have it have a
2:09:42 little less context but if you're trying
2:09:45 to actually get a specific answer or or
2:09:48 you really want it you know
2:09:52 you want it more predictable than having
2:09:56 enough context around it
2:09:58 um to sort of steer it in a direction
2:10:00 can help a lot
2:10:03 the other thing that these are good at
2:10:04 don't think that you always have to
2:10:06 completely invent and craft all of the
2:10:09 conversation up top you could start out
2:10:12 and you could say
2:10:15 um hey Chachi BT
2:10:17 um I don't know much about
2:10:20 um tomato horticulture
2:10:24 um is that something that you could help
2:10:26 me learn and it would say absolutely
2:10:28 blah blah blah here's the things and
2:10:30 then and then you would say well here's
2:10:32 the climate I live in and and you know
2:10:34 what can you tell me about that well
2:10:35 here
2:10:36 so you can actually have it provide its
2:10:40 own context and then start asking
2:10:42 questions right and it'll get good then
2:10:45 because it kind of knows oh this is what
2:10:47 we're talking about now right it's
2:10:48 already kind of narrowed its semantic
2:10:50 regions down in answering the questions
2:10:53 for you so that's a way you can do it
2:10:56 the other thing that they're really good
2:10:58 at is give it examples
2:11:02 so
2:11:03 um given an example of an email or a
2:11:05 blog post that you wrote or tweets maybe
2:11:07 your company writes Tweets in a very
2:11:09 particular format and it always includes
2:11:11 the URL and and you they all have pithy
2:11:15 humor to them you could say here's 10
2:11:18 tweets that that that represent the kind
2:11:21 of tweets that we like to write please
2:11:23 write 10 more on the following topic or
2:11:26 topics
2:11:28 um
2:11:29 that follow the tone structure and
2:11:33 humor of the examples above it's really
2:11:36 good at that it's really good at
2:11:37 analyzing stuff
2:11:40 um it's also good at um
2:11:44 you know helping you improve something
2:11:46 maybe you wrote something and you're
2:11:47 like this just doesn't feel right you
2:11:49 can have it critique
2:11:51 Silver Fox reached level 10.
2:11:54 people are leveling up in here people
2:11:56 the AI learning lab this is not really a
2:11:59 lab this is much more like Scientology
2:12:01 you reach higher levels all sorts of
2:12:03 magical [ __ ] happens I'm telling you I'm
2:12:05 telling the world opens up at certain
2:12:07 levels
2:12:08 I can't really talk about it a lot I
2:12:11 can't
2:12:12 I mean no promises but the higher the
2:12:15 level the closer you are to the
2:12:18 All-Seeing Eye of the lab you understand
2:12:20 what I'm saying
2:12:24 we're we're in talks with some some
2:12:26 Stars we're I you know we don't
2:12:30 Mr Cruz he's got his gig he didn't want
2:12:34 to talk to us but we've
2:12:36 we're Tom Arnold is on the short list uh
2:12:40 Rosie of course is on the short list
2:12:43 um
2:12:44 Alec Baldwin if we can get him gonna be
2:12:46 solid gonna be gonna be solid so yeah
2:12:48 yeah
2:12:50 all right all right then
2:12:53 almost 6 a.m here Matthias I'm staying
2:12:56 awake because of your cool lab I'm so
2:12:58 sorry best app that combines words
2:13:00 pictures and videos there's not well
2:13:04 there might be an app that does that
2:13:08 um
2:13:13 the
2:13:16 right now they're kind they're kind of
2:13:18 distinct like I'm sure that I'm sure
2:13:20 someone I would go to futurepedia.io and
2:13:23 and look for apps that are doing
2:13:25 combination kind of stuff right now a
2:13:28 lot of these tools kind of live in their
2:13:31 own little
2:13:32 Islands right like these are the word
2:13:35 ones then there's the picture ones then
2:13:36 there's the music ones then there's the
2:13:39 voice synthesis ones then there's the
2:13:40 animation ones then there's the video
2:13:42 ones right now they're all these kind of
2:13:44 distinct little tools this is
2:13:46 again one of the similarities to the
2:13:48 early days of the web and what's going
2:13:50 on with this AI these AI tools is that
2:13:52 there's there's not really the Adobe
2:13:54 Creative Suite of generative AI right
2:13:56 now right there's not you know you don't
2:13:59 you don't have well and even the Adobe
2:14:01 suite is it's a bunch of different tools
2:14:04 but they you know they they try to work
2:14:06 well together
2:14:08 um
2:14:09 what we're gonna see over the next six
2:14:11 months to a year
2:14:13 is that almost all of these tools are
2:14:16 going to get multimodal where they'll do
2:14:18 images audio voice synthesis
2:14:23 um
2:14:23 probably some video
2:14:27 um
2:14:27 in in a fairly integrated way
2:14:31 um they're not there yet everything's a
2:14:33 little a little gentified right now so
2:14:35 what what I would do is
2:14:37 I would start Sabrina with what's the
2:14:40 problem you're trying to solve first
2:14:43 figure out what what you actually want
2:14:44 to produce and then figure out what are
2:14:46 the best tools to contribute to that so
2:14:49 like if you just need like images with
2:14:50 words integrated in them ideogram's
2:14:53 great but if you want to turn that into
2:14:55 a video then you might need something
2:14:57 like runwayml that's gonna
2:15:00 you know do well or not
2:15:03 um if you want this to be print ready
2:15:05 you might need something like topaz Labs
2:15:07 photo AI tool where you can enlarge it
2:15:10 or you need to do that within stable
2:15:13 diffusion so there's
2:15:15 start with your use case understand the
2:15:17 requirements of what you're trying to
2:15:19 accomplish and then I would start
2:15:21 investigating what are the different
2:15:22 tools to do it
2:15:24 um I don't know of anything that's super
2:15:26 sophisticated that combines all those
2:15:28 things right now
2:15:31 um I got 99 a problem but a turtle ain't
2:15:33 one nice harsh sweet I like it solid
2:15:36 it's very good
2:15:38 all right one time 10 15. okay I'll do
2:15:41 another 15 minutes and you've heard this
2:15:43 before I'm shitty at time management
2:15:46 when I get when I get moderators in here
2:15:49 you y'all are in for a rude awakening
2:15:51 the moderator's gonna be like Kyle 15
2:15:52 minutes you gotta go it's time to get
2:15:54 off now you got to stop talking people
2:15:56 are bored people are tired of this
2:15:58 that's I don't know if you knew this
2:16:00 Kyle people are talking about you in a
2:16:02 very unflattering way what and what's
2:16:04 going on with the hair
2:16:06 do you own product what seriously
2:16:09 anything
2:16:14 all right
2:16:15 Fran Jackson what he's won
2:16:18 you want a new car
2:16:21 you can make money
2:16:24 Dr J my drive ended and I've been Thrive
2:16:28 oh God Thelonious please don't don't
2:16:31 don't don't
2:16:32 comment in the AI learning lab while
2:16:34 driving
2:16:35 okay I'm good at prompts but how to
2:16:37 monetize now so Sabrina here's the deal
2:16:43 look at look at
2:16:45 what you're good at
2:16:48 either either what you're good at what
2:16:50 you know what you have competence in
2:16:53 what you're passionate about
2:16:55 or
2:16:57 a community or group of people around
2:16:59 you that you know have a problem that
2:17:01 needs solving don't start with chat GPT
2:17:05 don't start with the tool
2:17:07 this is this is the thing that I think
2:17:09 is it it's it's one of the hardest
2:17:12 things to to kind of get your head
2:17:14 around
2:17:15 that because these things are so capable
2:17:18 and so [ __ ] sexy it's really easy to
2:17:21 go how do I use chat GPT to make money
2:17:26 I find that of it's a very daunting like
2:17:28 I even find just the the nature of the
2:17:32 The Prompt box itself is intimidating
2:17:34 because it can do so much but the minute
2:17:37 I flip it and say oh I've got to solve
2:17:40 this specific problem for a client or
2:17:42 for something at work and then I go to
2:17:45 these tools now I know exactly what to
2:17:46 do so it's the same with making money so
2:17:49 don't think about the tool first think
2:17:51 about what are you good at who do you
2:17:53 know what are problems you know need
2:17:55 solving that you maybe have an idea on
2:17:57 how to do
2:17:58 and then say okay
2:18:01 now that I know what problem I want to
2:18:03 solve or the thing that I want to do
2:18:04 that I think will resonate with these
2:18:06 people now I'm going to go to these
2:18:08 tools and say can I do it better faster
2:18:11 or cheaper using these tools than I did
2:18:14 traditionally and the answer is [ __ ]
2:18:17 yes like the answer is like probably 10
2:18:19 times faster or 10 times cheaper or 10
2:18:23 times better
2:18:24 or
2:18:26 half as cheap twice as good right there
2:18:30 used to be that old adage you can have
2:18:32 it good faster cheap choose any two
2:18:34 right you can have it good and cheap or
2:18:38 you can have it good and fast but it
2:18:41 won't be cheap right or you can have it
2:18:44 cheap and fast but it won't be good with
2:18:47 AI with this generative AI stuff I think
2:18:49 you we're actually entering a world with
2:18:51 what if you could have all three
2:18:53 what if what if you could have something
2:18:54 that's good
2:18:56 cheaper than the competition and faster
2:18:58 than the competition I think that's the
2:19:00 place to make money right now the
2:19:02 dramatic gap between how we do things
2:19:06 today and how you can do things with AI
2:19:08 even if it's the exact same solution
2:19:12 that Gap there's your opportunity
2:19:18 like I know people that are starting
2:19:20 agencies that are turning
2:19:22 they're going from initial meeting to
2:19:25 delivered strategy in eight hours
2:19:29 that's normally a one month to two
2:19:31 months
2:19:33 engagement
2:19:36 they're starting agencies to turn it
2:19:38 around in eight hours I think that's
2:19:39 ridiculous I would do a week
2:19:42 but
2:19:44 it sounds good
2:19:47 go from kickoff meeting to creative
2:19:50 strategy in eight in eight hours
2:19:53 we'll have the meeting at nine by five
2:19:56 o'clock the end of the day you'll have
2:19:57 your creative strategy and it'll be
2:19:59 blown out and good
2:20:03 you can start an agency around that you
2:20:05 can make money on that and you don't
2:20:06 even need to charge by the hour you
2:20:09 charge for the deliverable
2:20:12 that's why I would say a week I'm gonna
2:20:14 deliver it in a week but I'm gonna shoot
2:20:16 for building it in four hours
2:20:19 so I can do five of those a week and I
2:20:22 can charge
2:20:23 five thousand dollars instead of twenty
2:20:26 five thousand dollars that an agency
2:20:28 would charge you for that so
2:20:31 there you go
2:20:33 if I get that thing ramped up I'm making
2:20:35 25 Grand a week
2:20:40 working four hours a day
2:20:46 now how do you Market that how do you
2:20:48 get enough people to know about it
2:20:49 that's [ __ ] work right all of this is
2:20:52 in the execution
2:20:55 clap if you've ever been to a fat Farm
2:20:58 I've not ever been to a fat Farm
2:20:59 although lately I feel like
2:21:03 I gotta get something going with this
2:21:05 body here you know I gotta do something
2:21:07 because it just it's just I'm I'm of an
2:21:11 age where shit's starting to fall
2:21:16 it's happening it's happening what's
2:21:20 that Louis CK Joker he's just like I
2:21:21 went to the doctor and like doctor my
2:21:23 ankle hurts and Doctor looks at it and
2:21:25 takes an x-ray he goes yeah it's just
2:21:27 worn out he's like what you can't fix it
2:21:29 he's like nah it's just done that part's
2:21:31 just done
2:21:32 but yeah
2:21:34 laughs
2:21:37 oh good lordy lordy lordy god lordy
2:21:41 lordy lordy
2:21:44 I'm all I'm getting is lag on all lives
2:21:48 oh sorry Clint I
2:21:50 I haven't heard a lot of that if you're
2:21:52 getting on on all lives then there might
2:21:54 be some of a tick tock probably worth
2:21:56 deleting and reinstalling Tick Tock I I
2:21:59 had to
2:22:00 so on my Tick Tock
2:22:02 um for whatever reason green screen
2:22:05 video was just not there as an option I
2:22:08 like I couldn't there was nothing I
2:22:10 could do to get it I could be in front
2:22:12 of an image but I couldn't be in front
2:22:14 of a video
2:22:15 and then I deleted and reinstalled Tick
2:22:17 Tock and it came back so
2:22:19 uh I'm new too with one experience over
2:22:23 10 years ago watching these lives
2:22:24 encouraged me to try AI cool awesome
2:22:27 Firefly yeah like that that's that's why
2:22:31 this is here man is like just try it and
2:22:35 here's the thing thank you what's
2:22:37 lyricutius I didn't see quite what that
2:22:40 was cool name
2:22:42 um
2:22:43 I'm totally down with if you try this
2:22:46 [ __ ] and you're like [ __ ] it that's not
2:22:48 for me I'm going back to drawing with
2:22:50 crayons on paper bags
2:22:53 far out more power to you
2:22:56 what what
2:22:59 I find
2:23:04 what I don't want to happen is people
2:23:06 sit on the sidelines thinking they have
2:23:08 time
2:23:10 to like let this [ __ ] play out
2:23:13 I don't think we have time like I think
2:23:16 I think we got to get our [ __ ] together
2:23:18 we got to figure this stuff out because
2:23:20 it's coming fast and it's coming hard
2:23:22 and it ain't getting uninvented
2:23:26 and and as much as we want to you know
2:23:29 say hurricanes are horrible and you know
2:23:32 we've decided as a state we're no longer
2:23:35 accepting hurricanes so everything's
2:23:37 gonna be okay right that's what
2:23:38 education is doing right now with chat
2:23:40 GPT it's okay we've got Chachi PT
2:23:43 detectors now it's okay the students
2:23:46 won't be able to cheat we've decided
2:23:48 chat GPT is not acceptable at our
2:23:50 institution that's literally like
2:23:52 Florida saying it's okay we've decided
2:23:55 not to accept hurricanes anymore
2:23:57 yeah it's still coming right so so if
2:24:01 you try this [ __ ] and it ain't for you
2:24:03 cool fine you know at least you know
2:24:05 what's coming
2:24:07 and and what a lot of people are doing
2:24:09 is they're trying this [ __ ] they have
2:24:10 their Kevin McAllister moment they're uh
2:24:13 and then they realize oh holy [ __ ] let
2:24:16 me figure this out and then as they
2:24:18 figure it out they're like wait a minute
2:24:19 I've always wanted to do X and then they
2:24:21 use these things and now they can do X
2:24:23 oh if I can do X then maybe I could do y
2:24:26 holy [ __ ] what if I reinvented myself
2:24:28 what wait what do I really want to do in
2:24:30 my life oh my god I've always hated my
2:24:33 job like I'm seeing that [ __ ] a lot
2:24:37 that's cool
2:24:39 that's coming all right another
2:24:42 five minutes five minutes people
2:24:45 five minutes people
2:24:47 not a minute more okay maybe a minute
2:24:49 more thank you only got a Discord
2:24:52 disturbing video call intrusion
2:24:55 yesterday
2:25:00 not having
2:25:03 time history as a facilitator for
2:25:05 intruder
2:25:06 not just backgrounds time clock may help
2:25:09 I'm not sure quite what that meant togas
2:25:12 my dream prompt is like slash composed
2:25:15 music in the style of Paul Simon with
2:25:17 guitar
2:25:19 what the hell is it called
2:25:29 hang on let me see if I can find it I
2:25:31 can never remember wait AI ml Magic
2:25:35 did I did I do music tools here hang on
2:25:38 music tools
2:25:40 sound draw
2:25:51 damn it
2:25:54 um
2:25:57 hang on
2:26:00 AI visual
2:26:06 music
2:26:08 wave form
2:26:12 gin
2:26:18 it's not vilio is it
2:26:22 no
2:26:24 no
2:26:27 um
2:26:32 no
2:27:11 I know I'm not talking
2:27:14 I'm Googling here people
2:27:17 the hell was it called
2:27:20 there's there's
2:27:23 this is just useless
2:27:25 there's a thing out there that like what
2:27:28 they did is is quite genius they took
2:27:32 visual waveforms of music and they
2:27:35 trained it like they trained an image
2:27:37 diffusion model and then when you
2:27:38 describe like what you say like slash
2:27:40 post music and the style of Paul Simon
2:27:43 with guitar
2:27:44 it generates visual waveforms
2:27:47 and then reads them
2:27:49 so so it's basically doing image
2:27:52 generation it's it's like it's like a
2:27:54 mid-journey image generator except what
2:27:56 it's generating are Musical waveforms
2:27:58 and then it plays that music
2:28:01 and I can't for the life of me remember
2:28:03 what it was called
2:28:07 if anyone knows please put it in the
2:28:09 comments below
2:28:11 millibytes deep AI wasn't any of those
2:28:13 and there might be other ones this was
2:28:15 just the first one I saw that did did
2:28:18 that particular thing
2:28:22 refusion
2:28:30 I wonder if it's gone
2:28:40 oh I'm not connected to the internet am
2:28:42 I dead here
2:28:44 am I dead in the water is my internet
2:28:45 dead click on tap on the screen if you
2:28:48 can hear me
2:28:49 can anybody hear me can you hear me
2:28:53 yeah okay okay
2:28:56 refusion text to music
2:29:00 it looks like it might be dead
2:29:06 right
2:29:40 okay this is it
2:29:44 the refusion website here it is okay
2:29:46 Jesus
2:29:49 sorry about that there was a there was
2:29:52 10 minutes of your life you can't get
2:29:53 back okay
2:29:56 so I'm going to type in what did you
2:29:57 want
2:29:58 um Paul Simon
2:30:02 compose music in the style Paul Simon so
2:30:05 uh
2:30:08 Paul Simon
2:30:12 with guitar
2:30:14 [Music]
2:30:17 um
2:30:17 up
2:30:19 lifting
2:30:22 up next
2:30:31 all right so what it's going to do now
2:30:32 so so it just generated that now we'll
2:30:35 play it we'll see how it did
2:30:38 [Music]
2:30:48 not bad but but but how this is doing
2:30:52 what it's doing is is
2:30:54 um it's just mind-blowing to me this is
2:30:57 an image generation tool
2:30:59 that's been trained on audio waveforms
2:31:03 that have been Tagged so they'll tag
2:31:06 like jungle music and trap music and
2:31:08 Hip-Hop and you know singer songwriter
2:31:10 so I'll do like let me see here like
2:31:13 um Americana
2:31:16 um
2:31:17 ballad
2:31:20 like
2:31:22 John Hyatt
2:31:32 [Music]
2:31:45 so
2:31:47 if you're like me you listen to that and
2:31:50 you're like well that's [ __ ] awful
2:31:51 right is it is it me or is that awful
2:31:53 well it is and it's not right if you
2:31:56 understand what it's actually doing it's
2:31:58 mind-blowing but if you like compare it
2:32:01 to composing music it's pretty bad again
2:32:06 my contention right now
2:32:10 is that
2:32:12 all of these tools all of them are janky
2:32:15 pieces of [ __ ]
2:32:18 and it's just the nature of where we are
2:32:21 um
2:32:21 but again I would go to futurepedia.io
2:32:23 and and just look up music I'm sure
2:32:25 there's some stuff out there that's
2:32:26 pretty cool right now
2:32:28 um
2:32:30 all right
2:32:33 how where do I save my chat GPT
2:32:36 Creations on my phone so you can get
2:32:38 there's a
2:32:39 the hair what the [ __ ]
2:32:42 um
2:32:44 openai has an official iPhone app and I
2:32:47 think they have one for Android now too
2:32:49 it saves chat GPT saves all of your
2:32:52 prompt history
2:32:54 believe it or not in the web version of
2:32:56 chat GPT you can't search through your
2:32:58 prompt history but in the IOS app you
2:33:00 can
2:33:02 so
2:33:03 you can save all of your chat prompts uh
2:33:09 within chatgpt and then just go search
2:33:11 on them so
2:33:14 I don't know if you need
2:33:17 I don't know if you need chat GPT Plus
2:33:19 for that I don't think you do
2:33:23 we made sixty thousand well we're up to
2:33:25 72 000 though all right I gotta go to
2:33:27 bed it's it's late listen listen
2:33:28 everybody listen listen listen listen
2:33:31 good night Silver Fox just went
2:33:34 good eye Silver Fox Firefly coming up on
2:33:37 midnight yeah happy labor day everybody
2:33:38 I'm gonna get out of here thank you all
2:33:41 your regulars let's do our little Romper
2:33:43 Room evening here it's been cool all
2:33:45 right so I see Thelonious Thelonious is
2:33:48 not driving anymore
2:33:50 I see ion China and Firefly I see share
2:33:55 bear I see Emilio's wife Matthias
2:33:58 I see Pate welcome pay Tobias the music
2:34:02 link is fascinating isn't it cool
2:34:04 refusion
2:34:06 um so transforming audio into its fft
2:34:09 output and using image modeling makes
2:34:11 sense yep exactly it's it's a really
2:34:13 clever approach Dr J good night silver
2:34:15 fox
2:34:16 Share Bear Tobias all right I'm
2:34:18 repeating now Robert Rossi all right
2:34:20 everybody have a good night I'll see you
2:34:23 tomorrow I'm gonna go deal with my hair
2:34:25 bye everyone