Title: aiLL 10/13/23 - The New Era of AI is Here: What You Need to Know

Description

Chapters

Transcript

0:04 [Music]
0:15 [Applause]
0:27 all right are we going are we going it
0:30 looks like we going welcome welcome
0:33 welcome pick and stick
0:36 LLC thanks for
0:39 joining welcome to the AI learning
0:43 lab happy Friday
0:47 night it's like you got an option on a
0:50 Friday night go on a hot date or you
0:52 could hang out in the AI learning
0:56 lab guess where we
0:58 are a hot day today learning
1:05 [Music]
1:07 lab I think there's a lot of allergies
1:09 here I'm not generally bothered by
1:11 allergies but I got a little
1:16 sniffle welcome everybody do me a favor
1:20 if you would share the live tap the
1:23 likes do the engagement stuff uh so we
1:26 can get some more folks in here we'll
1:28 get this show on the road we is all
1:31 started oh yeah down in the bou Bou
1:38 [Music]
1:42 Bob that's a cool a cool picture of the
1:45 flat irons up in
1:48 Boulder but behind it they put a whole
1:52 town that doesn't
1:57 exist and it's pretty cute
2:02 [Laughter]
2:09 Corey darling Mentor what's
2:12 happening hey Dr J how are you Apple
2:16 user yeah I'm here and you too we're all
2:21 here what are you GNA do Friday night I
2:24 got plans with the tick
2:29 talk
2:31 oh
2:34 yeah look at
2:37 that would I have that if it weren't a
2:40 lab for learning no we got a llama we
2:45 got a turtle we got typography that's
2:48 sort of sort of
2:52 right it's about is it right right as it
2:55 would have been in the'
2:58 40s me hey hand cut things out of
3:06 myar all right oh life's tough it's
3:10 tougher when you're
3:12 stupid Tik tok's just got to make those
3:15 things last longer two second like what
3:17 is it like a second and a half just not
3:19 it's just not right
3:20 people just not right well welcome
3:23 everyone my name's Kyle Shannon this the
3:25 AI learning lab if you're uh sitting
3:27 there thinking to yourself what the heck
3:28 do you do what is this place
3:30 doesn't look like a lab where's your lab
3:32 coat are you a scientist do you have
3:36 qualifications no no that's not what we
3:39 do here we have
3:41 fun we talk about AI but not like the
3:44 wonky AI the AI like it can do that cool
3:47 [ __ ]
3:49 too what are we going to do with our
3:51 lives
3:53 run it's more like that t-shir shirt
3:56 Story please I don't know it's just a
3:58 cool pattern of
4:01 a sliver moon in waves over the ocean I
4:06 suppose um I don't know if it's a fancy
4:08 brand or not but it's nice it feels nice
4:12 I've had it for a
4:13 while and uh what else do we do here u a
4:16 little bit about my background I'm an
4:17 entrepreneur I'm the co-founder and CEO
4:20 of a uh video technology company called
4:24 storyvine and I've
4:26 been rather pathologically obsessed
4:29 about this AI stuff for about 2 years
4:32 now
4:34 um and uh and so much so that I started
4:38 this year Channel I started a thing
4:40 called the AI Salon I was also an
4:43 entrepreneur back in the early days of
4:44 the worldwide web I was co-founder of a
4:47 company called agency.com which is one
4:48 of the first digital
4:50 agencies and we you know we were there
4:54 for the way before it was the dot bubble
4:58 they were they were like 200 people in
5:00 New York City and like you know 300
5:03 people in San Francisco and you know
5:04 handful of people in other cities that
5:06 were like holy [ __ ] this stuff's cool
5:09 and
5:11 uh and it took the rest of the world a
5:13 while to figure it out and uh we're in
5:16 the same place with this AI stuff right
5:17 now although it's significantly more
5:21 profound and it's moving a lot faster so
5:23 uh so it's wild wild wild times so the
5:26 way this works I'll walk you through
5:27 some some guidance if you would do me a
5:29 favor share the live so we can get the
5:31 love out there get more people in let
5:33 Tik Tok know like just pretend like you
5:35 enjoy it and then after an hour or so
5:38 when I bludgeon you to death with with
5:40 stupid [ __ ] you'll just go I guess I
5:43 like it and it's like one of those fake
5:46 it till you make it things so
5:48 just show some engagement here we'll get
5:50 some more people in and then we'll uh
5:52 it'll be good okay um what are we going
5:56 to do okay if you are new to generative
5:58 AI if you're just scrolling in Tik Tock
6:00 and you're like is that Rosie oon let me
6:03 see what she's up to um it's
6:07 me she she's up to learning AI um you
6:12 should try chat
6:15 GPT and let me get this here
6:20 oh hold let's see got to dim that a
6:24 little bit so it looks nice there we
6:28 go all right so if you are brand new to
6:32 AI to the to the fascinating world of
6:35 artificial intelligence and machine but
6:38 where Kyle there chat.
6:41 open.com that's the official chat GPT
6:43 website if you have not played there go
6:46 play there now just go play just go play
6:48 go play go play go play but Kyle I don't
6:51 want to I'm not very technical doesn't
6:53 matter just go there okay chat.
6:56 openening ey.com what do I do ask it to
6:59 write you a poem ask it to write you a
7:02 country song about your pet or your
7:05 significant other or your ex that broke
7:07 your heart um ask it to write you a
7:10 Python program ask it to write you a
7:13 project plan for the pizza business you
7:16 never
7:17 started a business plan for the pizza
7:20 business you never started just go there
7:22 ask it [ __ ] anything oh I swear here
7:25 too by the way hey Sherry Banks thank
7:27 you for the ru the ru by any other
7:32 name Siri it's a little early for that
7:34 don't you
7:37 think Siri's been much more insecure
7:40 lately Siri Siri Siri go
7:45 away
7:55 okay oh
7:58 Siri
8:00 oh she's so cute
8:03 okay all right these other things if you
8:06 have played with chat gbt and you
8:07 haven't played with these other things
8:10 um Claude in particular is worth
8:12 spending some time on Bing and Bard are
8:15 both okay but they're both connected to
8:17 the internet
8:20 um and they're well I don't know if
8:22 Bing's going to get better Bing's got a
8:24 lot of weird Microsoft nness to it right
8:26 now so it's it's it's pretty crappy but
8:29 there there's
8:30 good stuff there Bard's about to get
8:32 really better whenever they release
8:34 Gemini claude's really really good um
8:37 what else is good to look at here um Pi
8:41 is really interesting although chat
8:44 gpt's official app just added
8:47 conversational AI to its app so now pi
8:50 and chat GPT the official app you can
8:54 just put it into conversation mode you
8:55 don't need to hit submit you don't need
8:57 to type you just talk to it it listens
9:00 and
9:00 response um in fact I can just show a
9:04 little demo oh I can show show a demo of
9:06 both of these so we'll do chat gbt
9:15 first can you help me understand how
9:18 internal combustion engines
9:23 work
9:28 absolutely here's a conc
9:30 explanation internal combustion engines
9:32 are engines that generate power by
9:34 burning fuel within the engine itself
9:36 rather than external so I just saw a
9:38 thing so Pi was based on chat GPT no it
9:41 wasn't so Pi let me let me show you Pi
9:44 really quick so here's pi and then in
9:47 the bottom well you can't see it there
9:49 but put it in talkie
9:52 mode what is going on
9:55 here hang
9:58 on
10:02 it looks like Pi just freaked out
10:05 hold poorly behaving robots
10:09 Okay can you tell me how an internal
10:12 combustion engine
10:16 [Music]
10:18 works sure thing here's a basic rundown
10:22 air and fuel are drawn into the engine
10:25 cylinders the mixture is compressed by
10:27 the engine's piston yeah I don't care
10:30 okay so um Pi is created by a company
10:34 called
10:36 inflection they raised
10:38 $1.3 billion dollar in funding um and
10:46 billion and they are setting out to make
10:49 the world's most powerful personal
10:52 assistant that company right there um
10:56 and it's a it's a wholly distinct
10:58 company from open AI um chat GPT was
11:02 kind of the first of these but there's
11:03 there's a lot of them so this is all of
11:05 these companies are using technology
11:07 that was invented in 2017 by Google
11:09 called the
11:10 Transformer and so a lot of these things
11:13 have been in development for a lot of
11:14 years but chat GPT accelerated the
11:16 release of lots and lots and lots of
11:18 toys or tools or toys that are like
11:21 tools or tools that are like toys um and
11:24 that's what we're playing with and
11:25 that's what we play with in here
11:27 um this little bad feel free to
11:30 screenshot this this little batch are
11:32 what I think are the image generation
11:34 tools that are are worth your time right
11:36 now there there's probably another five
11:41 Beyond this but just for now just get
11:45 started if you go to bing.com create or
11:48 I saw silverfox you got access to dolly3
11:51 inside chat GPT I'll show you that in a
11:53 second it's remarkable Dolly 3 is really
11:55 really amazing um and and generating
11:58 images inside of a a chat within chat
12:02 gbt is just it's a completely different
12:05 experience than going to something like
12:06 mid Journey or stable
12:08 diffusion um Kyle when will I be in DC
12:12 DC next uh I don't know I get to the
12:14 East Coast a fair amount but I haven't
12:15 been to DC in a while so um I don't know
12:19 probably probably sometime in
12:23 2024 um let's see mid journey is the one
12:27 that's sort of the highest quality art
12:31 ideogram um was the little darling for a
12:34 while cuz it did text but then Dolly 3
12:36 came out and then Firefly from Adobe
12:39 Adobe just announced a whole pile of new
12:42 AI tools they're going all in on AI um
12:45 and they're also starting to release
12:47 some of their artist attribution tools
12:49 and things like that and they just
12:50 announced Firefly 2.0 which is
12:52 apparently pretty good I haven't even
12:54 had time to play with it yet so lots
12:57 going on on the image front
12:59 this is some interesting video apps so
13:02 Runway ml will make videos out of
13:04 nothing or out of a picture heyen will
13:07 translate your video of you talking into
13:11 you talking in other languages it is
13:13 [ __ ]
13:14 remarkable um block a Labs you describe
13:17 a scene and it'll make you a 3D scene um
13:20 did Will synthesize a voice and make an
13:24 animated talkie picture of it um so
13:27 those are pretty cool
13:31 these two things futurepedia doio and
13:34 there's an AI for that are like big old
13:35 list of tools and then these three
13:37 people Ethan mullik David shap Shapiro
13:40 and Paul rater from the marketing AI
13:42 Institute are worth uh following um
13:46 Ethan mik's A Wharton business school
13:48 Professor David Shapiro is working on
13:50 AGI he is wicked smart and building a
13:52 really cool open source project to make
13:55 um AI smarter than people um and and
13:59 he's he's you know really understands
14:01 this stuff in a deep deep way so if you
14:03 want to feel if you if you want to feel
14:06 intellectually inferior go watch David
14:08 Shapiro actually all three of these guys
14:10 are pretty [ __ ]
14:14 smart um and then the other thing this
14:17 thing in the middle here the AI Salon
14:19 the salon. is the link tree um this is a
14:23 community I created started last
14:25 December we we do meetings every other
14:27 week and we have an online Discord
14:29 Community where we can hang out and ask
14:32 questions and make pictures and stuff
14:34 like that a lot of times the the folks
14:36 that hang out and hear a lot called the
14:38 Irregulars are in there making pitches
14:41 and making jokes and all that sort of
14:44 stuff which speech to text wait wait
14:48 speech text AI do you recommend um I
14:51 assume you mean text to speech and I
14:54 assume you also mean voice synthesis
14:57 like which one if you want to take your
14:59 voice and make a model of it so you can
15:02 just write text and then you start
15:03 talking um 11 Labs right now is is the
15:07 one um for two reasons one their models
15:11 better like like apple released uh a
15:15 thing called personal voice in iOS 17
15:19 which I thought was going to be amazing
15:20 cuz it's apple and it took like 20
15:23 minutes to train it you had to answer
15:25 150 questions it's just [ __ ] awful
15:29 and then it was not good it was not good
15:33 it was not
15:34 good 11 Labs is pretty good not great
15:38 what all of them Miss right now is any
15:40 kind of Dynamics with you know
15:43 intonation and Dynamics with you know
15:46 emotional regulation basically but there
15:50 are a lot of tools that have them that
15:54 you can kind of manually set so we're
15:56 not going to be that long till those are
15:58 going to get good but right now now it's
15:59 11
16:00 Labs um and then this tinyurl.com
16:03 kyes AI tools um one of the Irregulars
16:07 here named Theon has put that together
16:09 that's a Google sheet that's got every
16:12 time I mention some tool or some person
16:15 on the live he's been capturing it in
16:17 that spreadsheet so super generous of
16:19 him to share that um and that's part of
16:22 the deal here is as you learn stuff if
16:24 you start hanging out here a lot um you
16:27 know great I would love that um share
16:29 what you're learning share what you know
16:31 share what's going on at work share you
16:33 know as you're learning this stuff share
16:35 it with people even if they think you're
16:36 weird because who gives a [ __ ] who gives
16:40 a [ __ ] they think you're weird you know
16:42 it's going to happen a year from now
16:43 they're come back hey could you teach me
16:44 that AI stuff you were talking about me
16:46 me me so so think of the AI learning lab
16:50 as as as your anonymous group for for
16:54 being curious and excited about AI uh
16:58 can my boss tell if I use chat gbt in my
17:01 writing if you have to ask that question
17:02 I guarantee you he can't no one can tell
17:05 in fact Ethan mullik the the Wharton
17:07 business school Professor um even the AI
17:11 detectors don't work um in fact they
17:14 don't work so bad that open AI the
17:17 people behind chat BT pulled theirs off
17:19 the market because it just didn't
17:22 work um they also did a thing where they
17:25 got business professors to try to uh
17:28 guess which writing was AI generated and
17:31 which was human generated guess what
17:34 they failed miserably so no your boss
17:36 won't be able to tell now if you're
17:39 shitty at prompting and you just create
17:43 driil and you don't proofread it and
17:45 it's got [ __ ] in it then your boss
17:47 will probably just think you're an
17:50 idiot and probably won't accuse you of
17:53 using chat gbt but might accuse you of
17:55 being a dumb Dum so so so learn these
18:00 tools well enough to understand where
18:01 the boundaries are what are they good at
18:04 what are they bad at and
18:07 and don't lose the most important thing
18:11 your [ __ ]
18:13 brain use your big boy or girl
18:17 brain no I meant text to voice to
18:21 listen when we can read lengthy PDFs
18:27 oh um
18:30 [Music]
18:35 I don't
18:37 know I there are probably a lot of tools
18:39 out
18:41 there just just put it into voice mode I
18:45 I I I would think that just
18:48 accessibility if you have an iPhone just
18:50 your accessibility tools
18:53 just pop it in and like cuz I know um OS
18:58 X on the Mac can read
19:01 things um I I honestly don't know that's
19:04 a that's a good question zindo good
19:06 question I I I have no answer for you um
19:09 it might be worth going to something
19:11 like futur
19:27 pedia.com
19:29 tonius I so it's funny like two nights
19:32 ago someone was in here bitching about
19:34 ah this is going to take more energy
19:35 than life
19:37 itself uh Jim Ross Kyle thank thanks so
19:39 much for today's V Virtual Office hours
19:41 you're welcome Jim uh Jim so I've got a
19:45 an Friday afternoon office hours on
19:47 LinkedIn that I do and Jim hangs out in
19:51 here and he's been excuse me he's been
19:54 coming to that um really good guy really
19:57 doing interesting things um has a
19:59 property management business he is not a
20:02 technologist at all although he's pretty
20:04 good um but he's been using this AI
20:09 stuff to like it so so Jim hope you
20:13 don't mind that I talk about you as as
20:15 if you're not here Jim um but Jim is
20:20 um he's it's a small operation and he's
20:24 been using AI to like punch way above
20:28 his weight like doing all sorts of
20:30 really cool content marketing things and
20:32 social media things and things that he
20:34 wouldn't have done otherwise like he he
20:36 wouldn't have spent the money to hire an
20:38 agency to do that and all of a sudden
20:40 he's he's operating like a five or six
20:43 or 10man team just because of these
20:45 tools so really interesting guy um so
20:48 good to see you thanks thanks for coming
20:49 nanoel electric device performs
20:52 real-time AI classification without
20:54 relying on the cloud oh interesting yeah
20:57 so so that's that's um I mean Apple
21:00 Apple's last two announcements it's
21:02 they've made it pretty clear that
21:03 they're they're building Hardware
21:06 designed to be able to run these large
21:08 language models locally on device so I
21:11 think we're going to see more and more
21:12 of that it's still going to take a [ __ ]
21:14 ton to train them but if you can run
21:16 these things if they can get them
21:17 efficient and they can run these things
21:19 primarily locally with an occasional
21:20 call to the web that's pretty big deal
21:23 pretty big
21:26 deal and the other thing is you know the
21:28 AI might just solve the the uh the
21:31 energy situation and help us get the
21:34 fusion faster or who knows what
21:37 else um so yeah very
21:42 cool now
21:44 amazing um what's your LinkedIn my
21:47 LinkedIn is I think it's just Kyle
21:50 shamon my company is storyvine so if you
21:53 look
21:54 for Kyle Shannon Story
21:57 vine
22:06 uh linkedin.com
22:09 Inky Shannon yeah just that just just my
22:13 name it's my name you know I go radical
22:17 some people go with the moniker or the
22:20 Avatar I go with uh what would be called
22:23 your
22:24 name cuz I'm
22:27 old when I started getting into
22:29 blockchain and nfts I'm in I'm in these
22:31 servers these Discord servers with all
22:33 the these nft Bros and they're like
22:35 don't dox yourself man I'm like too late
22:38 man it's been like 30
22:40 years I've been docks for like 30 years
22:45 man
22:49 dude yeah what are you gonna do what are
22:52 you going to do it was funny I was I was
22:55 doing all that that nft stuff and one of
22:58 the nfts I
23:00 had briefly for about for about a
23:03 six-month period was worth like
23:05 significant six figures and I was like
23:09 oh [ __ ] this this is why people don't
23:12 want to be docked if you're sitting
23:15 there with you know the the cost of a
23:18 you know of a small house on a USB
23:23 stick Kyle what is the difference
23:25 between Advanced statistics
23:29 and
23:31 so I don't know what the rest of your
23:32 sentence is so
23:36 so
23:38 so um Advanced statistics and I don't
23:43 know what's the rest of that but if we
23:45 get the rest of that Stardust astrology
23:47 I'll do my best to answer
23:50 it Advanced statistics and advanced
23:53 algebra for algorithms I don't I don't
23:56 have any idea so so just a little bit
23:58 about this channel my background I'm not
24:00 a mathematician I'm not an engineer so I
24:03 know generally how this stuff works I
24:05 don't know the math of it at all I have
24:07 a degree in
24:08 acting
24:11 um my my passion for technology is at
24:15 the inflection point where it shifts
24:17 from the domain of the scientist to to
24:20 the domain of the rest of us so with the
24:23 worldwide web it took the internet which
24:26 was primarily command line and research
24:28 and it brought it to the masses by
24:30 making it point and click so that
24:32 inflection point was when I got involved
24:34 with that um with this AI stuff similar
24:37 sort of deal where it's been around for
24:39 decades November 30th 2022 chat GPT
24:43 launches and that for me is the
24:45 inflection point that says
24:47 Ah these tools are amazing but now the
24:50 rest of us can play with it
24:53 right
24:55 um you know there were some things a few
24:58 years you know a year or two before that
25:00 that were that were that were hints of
25:01 it coming but chat gbt really is the
25:04 thing that that comes from thereafter so
25:06 I don't I don't know the difference uh
25:10 between those two things and I would I
25:13 would refer you to chat GPT which can
25:16 probably answer that it's like that uh
25:19 let me Google that for you let me GPT
25:21 that for you so I don't want to do that
25:24 because there's enough mansplaining
25:26 going on in this channel as it is
25:29 my degree is in religion and WooHoo for
25:32 Liberal Arts Majors
25:34 exactly we are in what I call revenge of
25:38 the liberal arts major right now where
25:40 um if you specialized in something um
25:46 that specialization is getting
25:47 democratized for every
25:50 specialization so more and more people
25:52 are going to be able to do more and more
25:54 things that they were not trained for
25:56 and it's pissing off the special
25:58 specialist but the horizontal thinkers
26:01 people that can think across vertical
26:06 expertise they're going to have some
26:08 major skill um they're going they're
26:10 going to there's going to be major
26:12 demand for that those kind of
26:14 thinkers um so people like liberal arts
26:18 Majors
26:23 yay you know the ones that all of the
26:26 Specialists used to make fun of
26:28 [Laughter]
26:31 yeah good luck making any money with
26:33 that acting degree
26:36 buddy why didn't I get the new features
26:39 in chat GPT yet I have
26:42 Pro did you go turn them on so let me go
26:45 show you something here
26:47 Justin if you paid your 20
26:51 bucks in the old chat
26:54 JPay by the way you can make money with
26:57 chat J
27:09 uh I got some allergies going on I got
27:13 postnasal dripper something anyway okay
27:15 so when you're at chat GPT if you paid
27:17 your 20 bucks you'll have your little
27:18 plus sign and then you should have some
27:22 things down here actually some things
27:24 are missing for me oh no they're not
27:27 okay um
27:28 um so what you have to do is go to your
27:35 settings and then within settings
27:39 there's a thing called beta features go
27:41 there and make sure they're turned on so
27:44 if you're not say seeing browse with
27:46 Bing plugins and Advanced Data
27:49 analysis Barbara art school survivor
27:52 that's
27:55 awesome that's great um then you'll see
27:59 these things here now Dolly 3 is um that
28:04 even though it's in beta it doesn't seem
28:05 to be showing up in people's beta list
28:08 like these other things do um but
28:10 they're just kind of rolling that out a
28:12 number of people people in here that I
28:14 know have gotten it in the past two days
28:17 I got it like five days ago so the dolly
28:20 3 the image Generation stuff is is
28:22 really brand
28:24 new um but but let me show you while
28:26 while I'm in here why don't I show you
28:27 some some of the advanced stuff that's
28:29 that's super
28:30 cool so I'm just going to go with the
28:32 default model right here and there's a
28:34 new little icon
28:36 here and it allows you to upload images
28:40 and then it can interpret images
28:50 so all right that's a good one so this
28:53 was I did a workshop Wednesday night um
28:56 was the second Workshop of a thing thing
28:58 called the AI
28:59 Launchpad and so this was a schedule
29:01 that I wrote on the Whiteboard I'm like
29:03 how long was the
29:10 workshop come on come on
29:14 you so it
29:17 uploads this thing I don't have anywhere
29:20 how long the workshop was but I do have
29:22 time ranges for the little
29:25 sections and let's see the workshop was
29:28 from 5: to 7:00 so it was 2 hours long
29:32 going to
29:35 say what was it
29:46 about looks like I finally got image
29:48 upload I did I got it maybe two days
29:52 ago and it's also in the the chat GPT
29:55 app now is fairly fully function fun it
29:58 doesn't have plugins in it but it's and
30:00 it doesn't have data analysis which
30:02 makes sense for for an
30:04 app um but it's got do it's got browse
30:07 with Bing it's got image upload and it's
30:10 got conversation
30:12 mode so anyway so so look this thing
30:15 just analyzed what was on the Whiteboard
30:17 um Let me throw up another one that that
30:19 I think is is another cool
30:22 example um is this it yeah so so this
30:26 was a strategy session I did with my
30:28 co-founder and you know we wrote all
30:30 this crap down on Post-it notes and then
30:32 this just been sitting up on
30:34 my my wall for like a year or two I
30:37 don't know we did this ages ago and um I
30:42 never sort of got around a typing it in
30:44 right so I can go what was going on
30:51 here
30:55 oops mid Journey has some amazing new
30:57 features coming out by the way when
30:59 replicant are are there have they
31:01 announced them are there any previews
31:03 are there any like um like a video or
31:07 examples they talking about it on the
31:11 Twitter I mean honest to God mid Journey
31:13 better come out with something cuz dolly
31:16 dolly Came Out Swinging hard and then
31:20 Firefly 2 is supposed to be pretty
31:22 amazing um and and the fact that mid
31:25 journey is still in [ __ ] Discord the
31:27 fact that they don't have a website or
31:28 an app right now I think is is really
31:31 stupid um they their CEOs a bit of a
31:36 crypto bro we're oh we're testing now
31:38 David announced some hints in office
31:40 hours that's awesome replicant okay good
31:43 so so are so you it sounds like you work
31:46 there so if you work there it like I'm
31:49 telling you the fact that chpt inside of
31:53 or uh Dolly 3 inside of chat GPT the
31:57 fact that you don't need to learn all of
31:58 the commands and things like that I
32:01 think it's going to be a big big uh
32:05 challenge to Mid Journey um to Mid
32:07 Journey scaling um so that's really good
32:10 to hear I'm excit I'm excited to hear it
32:12 because I still think mid journey is the
32:14 best in terms of quality but there's a
32:17 convenience thing and the fact that um
32:20 that dolly3 does text so well and can
32:23 kind of do things within the context of
32:25 a conversation is a big deal anyway way
32:27 so so this
32:30 image says it says I said what was going
32:33 on here was all I asked that the image
32:35 you've shared is a wall of sticky notes
32:36 grouped under different headers it
32:38 appears to be a brainstorming or
32:39 planning session commonly used in
32:41 business or project management here's a
32:43 breakdown of the headers
32:44 goals um there's also mention of a 1 to
32:47 twoyear exit implying a strategy or goal
32:50 to potentially sell or
32:52 transition unjustified assumptions
32:54 choices so I can even say um
33:00 can you give me a bulleted list of the
33:02 Post-it notes under the unjustified
33:04 assumptions
33:11 section I'm testing the Alpha app now as
33:14 well replicate you were in here like two
33:17 months ago as I recall right and I think
33:19 because I think I was bitching about
33:21 Discord back then and you said the the
33:23 the uh an app was under the work so the
33:25 fact that you're Alpha Testing it right
33:27 now great if you want an outside Alpha
33:29 tester let me know because I'm if if you
33:33 know anything about this channel you
33:35 know that these lists that I make are
33:38 like ain't nobody paying me to make
33:41 these lists right these are not
33:42 sponsored lists these are the things
33:44 that I think are generally in order of
33:47 preference the most valuable things
33:49 right Dolly 3 I have at the top just
33:51 because it's incorporation into chat but
33:54 mid journey is still there mid Journey
33:56 comes out with an app can get your [ __ ]
33:58 together with text and you know still
34:01 maintain the quality like like that's a
34:03 huge deal so anyway Dolly E3 has great
34:07 un language understanding though yeah I
34:09 know it yeah it it really does the and
34:11 so let me let me show some of that so so
34:14 anyway like like here's the the details
34:17 of what what was written on those on
34:19 those sticky notes I have shitty
34:21 handwriting right and it it got this
34:23 stuff one of the other things that
34:25 people are doing with this
34:27 this image Vision stuff is they're
34:30 they're they're architecting software or
34:33 websites like in flowcharts on a
34:35 whiteboard and then just telling chat
34:37 GPT to read the flowchart and code the
34:40 code the components of the app or code
34:42 the website and it just does it so
34:46 that's pretty wild so now let's flip
34:47 over let me go to a new
34:54 um come on what's going on there we go
34:57 let me go to a new
34:59 chat and I'm going to go to Dolly 3 so
35:02 let's go look at what this thing does so
35:05 so I'm still in chat gbt oh wait another
35:07 thing from replicant mid Journey does
35:09 have a user based testers for web it's
35:11 for users that have created 10,000
35:13 Images
35:14 Oh I probably am under that I'm probably
35:18 about at that with stable diffusion but
35:21 I'm probably about yeah I'm probably
35:23 under that so all right
35:26 whatever
35:29 I uh you know it's not it's not what I
35:32 do full-time so I get it um okay so
35:36 we're in chat GPT so one of the things
35:38 that that excites me about having image
35:41 generation within the context of chat is
35:44 you don't have to start out necessarily
35:46 thinking about oh I want to make an
35:48 image right you could just
35:51 say I've got a dog grooming business
35:55 that focuses on on chubby black and
35:59 white dogs only and I'd like you to help
36:03 me come up with a name for it
36:10 period Panda Pooch Parlor Oreo K9 Oasis
36:14 Chubby Checker grooming we're going with
36:16 Chubby Checker
36:17 grooming okay Chubby Checker grooming is
36:20 awesome let's go with that now can you
36:23 come up with a tagline for my
36:26 business absolutely where plump pups
36:30 where plump pups get primped I think
36:32 we're going with the thing that's
36:34 impossible to say for the Pud for the
36:36 pudgy Pooch with
36:41 panach check in Chubby check out Charmed
36:45 oh that's good too wait what's the name
36:46 of the company oh Chubby Checker
36:49 grooming Chubby Checker grooming for the
36:51 pudgy poop with pooch with panach the
36:55 pudgy poop okay so let's go
36:58 okay perfect let's go with number
37:02 four now I'd like you to come up
37:05 with four
37:09 different marketing strategies for me
37:12 that involve social
37:18 media of
37:22 course before and after showcase boring
37:25 interactive Chubby Checker challeng
37:27 yeah okay
37:29 maybe um pudgy Pooch tips and
37:34 tutorials and Chubby Checker
37:37 stories Facebook and Instagram stories
37:40 heartwarming
37:42 stories okay I like number four can you
37:45 write a heartwarming story of a chubby
37:48 black and white Pooch that used to be
37:51 sad because its hair was all matted and
37:54 now it's happy because it's got
38:00 panach did I choose the panach one I
38:02 think I did what did I choose
38:07 yeah the tale of pudgy Parker in a quiet
38:10 neighborhood there was a chubby black
38:11 and white dog named
38:14 Parker okay so so we're gonna whatever
38:17 that story is it's
38:19 awesome okay that's great but change the
38:22 dog's name to
38:24 Champ and then make me four social media
38:29 images that have the name of the company
38:32 the
38:33 tagline
38:35 and a tag on the dog's call collar that
38:38 says
38:40 champ the image should illustrate the
38:44 story you just
38:47 told and so
38:50 now it will go make those
38:52 images better get cracking Kyle what did
38:55 I do what happened what happened
38:58 what
39:01 happened um oh better get cracking on
39:04 Mid Journey images I know man I I just
39:07 thought maybe replicant would throw me a
39:09 bone cuz you know I'm I'm uh I'm I'm in
39:11 here talking them up although I was kind
39:14 of crapping all over it I just listen
39:17 here's the deal I love mid Journey it's
39:20 it's like it from a quality standpoint
39:23 it's the best one out there but uh
39:26 regular people that don't want to
39:28 tolerate Discord are you know you know
39:32 you know what I'm saying all right does
39:34 do any of these okay that says champ oh
39:37 this is the before Story the kids are
39:39 laughing at him this is pretty
39:41 good so that's the before here's the
39:45 after Chubby Checker
39:47 grooming Chubby Checker grooming that
39:50 looks kind of like champ and actually
39:52 they got the mirror I did a I did
39:54 another image that had a mirror in it
39:56 and the mirror was a [ __ ]
39:58 disaster like this has got reflections
40:01 of bottles right here but there's no
40:02 bottles right here like it's not super
40:04 good at
40:06 mirrors that's a cute
40:09 one for the pudgy well for the pub gy
40:14 pug with
40:16 panach
40:20 champ um so so let's
40:25 see let's see if we can do something
40:29 [Applause]
40:31 here I like one and two as before and
40:34 after pictures can you combine them into
40:36 a single wide layout that's got before
40:40 on the left and after on the right with
40:43 the name of the
40:44 business spanning both
40:47 images And the tagline on the
40:51 bottom let's see let's see if it if it
40:54 can I mean one of the things that um
40:57 that dolly3 has is a thing called prompt
41:02 coherence where it's pretty good you can
41:04 add a lot of objects into an image and
41:06 it for the most part will include them
41:09 um I found that it's a little weird with
41:11 like it while it makes text the idea of
41:14 put this text at the top and that text
41:16 at the bottom it it tends to get a
41:18 little confused with that
41:22 stuff but we'll
41:25 see very
41:32 cool those kids are so
41:35 [Laughter]
41:43 mean how do we let's
41:55 see
41:58 this this is not this is not a dog
42:01 grooming ad this is a horror
42:05 film and that's not chubby come on
42:08 there's a difference between chubby and
42:09 morbidly obese doggies but he is looking
42:12 better over
42:13 [Laughter]
42:15 here um but anyway that's um that's
42:20 Dolly 3 within chat GPT just you know
42:23 the fact that you can just you know have
42:25 this long conversation about you know
42:28 whatever you're working on a story
42:31 you're working on a marketing thing and
42:32 say hey just make me an image for
42:34 that it's pretty pretty pretty dang
42:38 something um this is not how you write
42:42 image gen prompts correct that's what
42:45 I'm saying this is not writing image gen
42:50 prompts chat G or Dolly 3
42:53 inside um chat GPT what what I'm Dem
42:56 demonstrating is that you don't have to
42:58 write image prompts now could you could
43:01 we do a
43:04 um a full-on image prompt here and put
43:07 it in here let's go play with that so
43:08 let me jump over to
43:10 PO
43:13 Poe and I'm going to play with a bot
43:15 here called mid
43:17 journey and I'm going to say
43:22 um picture of the flat irons mounted
43:27 feature in Boulder Colorado at
43:31 Sunset after a light
43:34 snow what
43:36 this what this um plug-in does is
43:41 Right a very like a long specific prompt
43:45 for Mid Journey it's weird that it wrote
43:47 it as a poem but that's all right let's
43:49 just take
43:50 that and we'll go over to chat GPT and
43:55 without any context just pop in a mid
43:58 Journey prompt and see how it
44:00 does should be interesting what's what's
44:03 going on here Pudge the judge what's
44:05 going on here is behind the scenes and
44:09 and chat GPT used to show this behind
44:11 the scenes cat GPT is actually writing
44:13 four distinct prompts so it's taking
44:15 whatever whatever was written up above
44:17 but but it's also taking it in the
44:19 context
44:20 of the the larger conversation GP gp4
44:26 vision will not talk about people in
44:28 photos though yeah that's correct Ted so
44:31 everything that I've seen of these uh
44:34 image reading tools so
44:39 um um Bard reads
44:43 images GPT chat GPT now reads images and
44:46 there was a third one none of them will
44:48 deal with the
44:51 people
44:53 um yeah those are not bad I I want
44:56 wonder why it's it's weird that they're
44:58 they're always um from the side like
45:05 that actually let's let's go ask
45:12 it those are okay style-wise but they
45:15 are from the side and I'd like to see
45:17 the flat irons from the front also can
45:20 you make the image landscape instead of
45:24 portrait and so again wait do I see a
45:27 difference between gp4 and Po and open
45:30 AI I haven't y used the the gp4 within
45:35 po a lot I would imagine it's different
45:40 the Bing chat which is also based on
45:44 calls to to GPT 4 is very very different
45:47 than open ai's chat
45:50 GPT um for me right now chat gbt 4
45:54 within open AI within chat GP PT is the
45:57 best one um but I haven't done enough uh
46:01 I haven't done enough playing in po to
46:04 know if it's if it's different or not
46:07 mid Journey describe will describe
46:09 people a bit oh yeah yeah it will
46:14 actually so that's sort of more from the
46:17 front but they're still kind at this
46:18 weird side angle which is it's funny
46:21 because there's a lot of pictures of the
46:24 flat irons in Boulder like it's it's
46:26 like what everyone takes pictures of so
46:29 there's got to be lots of source images
46:30 of that it's weird that they're off to
46:32 the side like that but whatever all
46:34 right all right all right all right all
46:37 right all right right right right we
46:39 having fun this this Friday night listen
46:41 listen people let me ask you something
46:43 are you enjoying yourself here if you're
46:45 enjoying yourself here go ahead and
46:47 click the like the tapy Taps to make the
46:50 likes uh follow the channel I would
46:53 appreciate a follow if you want to see
46:55 this kind of content uh um whatever else
46:58 you do to support oh share the live let
47:00 let other people know we're here what
47:02 about the copyright concerning all those
47:04 generated images so Nicholas there's the
47:07 copyright stuff is a very very very very
47:10 very very very very very very very very
47:14 very contentious and gray area it is not
47:20 clear one of the things that's the
47:23 biggest misconceptions about the way
47:25 generative AI works
47:26 is that it's copying and pasting
47:29 copywritten work it's
47:32 not when when these models are trained
47:35 they're trained on copywritten work
47:37 right hence that's where the gray area
47:39 is but the works as they started out are
47:45 not what gets embedded into these models
47:48 um both on the word side and the image
47:51 side all this stuff the G in GPT stands
47:54 for generative generative pre-trained
47:56 Transformer it's actually generating
47:58 kind of from nothing using this training
48:01 it's it's conceptually more similar to
48:04 someone going to school for four years
48:07 and learning all about Andy Warhol and
48:09 loving Andy Warhol and when they get out
48:11 of school all their art looks like Andy
48:13 Warhol it's more akin to that now the
48:16 problem is if you prompt these things
48:19 correctly you can get stuff that looks
48:21 exactly like Andy Warhol and if you
48:24 invoke artist names and things like that
48:27 you can you can get that so so there are
48:29 all sorts of issues is it copyrightable
48:31 is it copyright infringement um where's
48:34 the attribution um where's the
48:36 compensation all these issues are to be
48:39 worked out now if you're super concerned
48:42 about this let me flip over
48:46 here if you're super concerned about
48:48 this stuff then pay attention to what
48:51 the companies are doing Adobe for
48:53 example has not trained their image
48:57 models on any copywritten work it's been
48:59 trained on um public domain and Adobe
49:02 stock images that they own the rights to
49:05 and now they're layering in allowing
49:08 artists to opt in or opt out of being
49:10 trained and the ones that opt in they're
49:12 coming up with compensation and
49:14 attribution models for
49:18 so if you want to avoid those kind of
49:21 issues altogether you know stick with
49:23 something like Adobe now what's the
49:26 downside of their approach the downside
49:28 is that their their images aren't as
49:30 good because they've been trained on a
49:32 much smaller
49:33 set um but I would also learn how these
49:37 tools work and and what I've seen some
49:40 artist friends of mine uh guy named
49:42 Kenny fredman who's a really brilliant
49:44 creative director and really into this
49:47 image stuff he personally will not use
49:50 any other artist's name when he's
49:52 writing his
49:53 prompts and so he just figures out if he
49:56 wants a certain style of an artist he
49:58 just works until he can get it there and
50:01 you know that's a choice you can make um
50:04 but it's it's a really great area it's
50:06 going to be my guess 20 years of Court
50:08 battles till we sort all this [ __ ] out
50:11 because it's a lot a lot of people are
50:14 comparing it to the to the 80s and 90s
50:16 with hip-hop when hip-hop artists
50:18 started sampling thank you for the heart
50:21 thank you for the heart oh and a
50:22 lemonade Sher Banks thank you so much up
50:25 to level 11
50:26 bang
50:30 um with with sampling with with people
50:34 sampling music for um for hip-hop they
50:39 were literally copying and P pasting
50:42 right they're taking the piece of music
50:45 and sort of wrapping it into some other
50:48 clips and mixing them and throwing
50:51 effects on them they were copying and
50:53 pasting music That's not how these tools
50:56 work so it's it's a it's a much more
50:58 nuanced technical argument and and
51:01 that's the the defense for the mid
51:04 journeys of the world and the open a
51:06 open AIS of the world and the stable
51:08 diffusions of the world stability AI is
51:11 they're going to say the while they were
51:14 trained on this work the work itself did
51:17 not it looks nothing like that in this
51:19 thing and it's more akin to people
51:21 learning something than it is about
51:23 copying and pasting so we'll see my dad
51:26 had four remotes me Lynn Sue and Ruth
51:30 exactly that's that's some Gen X right
51:33 there we were the
51:36 remotes change a channel I don't want to
51:39 change a
51:42 channel that was the equivalent of this
51:44 change a
51:47 channel put on seven put on seven put on
51:51 seven Kyle Kyle Kyle Kyle put on seven
52:05 um you gave some good advice about
52:07 images we make document everything you
52:09 do oh yeah that's a good a good point as
52:10 well forever
52:12 hooked so it looks like what the what
52:16 the copyright office if you want to
52:18 copyright images you make with AI what
52:21 what what's going to likely happen is
52:24 this
52:26 people that are creating art that they
52:28 want to copyright it's probably not
52:29 going to be pure AI right it's probably
52:32 going to be like they'll do some work in
52:33 AI then they'll take it into Photoshop
52:35 they'll do a little manipulation then
52:37 they'll take it out maybe they use that
52:38 as an inspiration for another AI then
52:41 then they put some typography on it
52:43 and in design or
52:49 illustrator oh very cool happy Friday to
52:52 you thank you very
52:54 much um
52:57 if if you're sort of going down a path
52:59 where you want to you want to make a
53:01 case for copyrighting images what what
53:03 the copyright office is saying is you've
53:05 got to prove human contribution what
53:08 they're saying is if you just type in a
53:09 prompt and it makes an image that's not
53:12 copyrightable if
53:14 you spend a month writing 6,000 prompts
53:20 and you combine you know 400 images to
53:24 end up with this thing that you then
53:25 take into Photoshop and do that document
53:27 all that [ __ ] and then that way if you
53:30 want to copyright it you can go to the
53:31 copyright office and say while AI was
53:34 used as a tool in the process this was
53:37 ultimately my
53:38 creation um that's the way I see most
53:41 people working anyway maybe not to that
53:43 degree um how do you access dolly3 so if
53:47 you go to
53:48 bing.com
53:49 create that's Dolly 3 it's not as good
53:54 as Dolly 3 inside chat GPT but if you if
53:58 you are a chat GPT plus member if you
54:00 pay the 20 bucks a
54:02 month they've been rolling it out over
54:04 the past week or two so just be patient
54:07 if it's not there now but pay your 20
54:10 bucks to chat GPT and then you'll have
54:13 Dolly 3 in the context of a chat GPT
54:16 session and it's really really good it's
54:17 better than Bing Bing Bing's
54:19 implementation of it's not great but to
54:22 be clear bing.com create is an image
54:26 generation tool for free that Microsoft
54:29 hosts and that is dolly3 so it does text
54:32 well it only does Square images which is
54:35 bizarre I can't believe they didn't add
54:36 a landscape and portrait button but
54:39 whatever I'm not on the product team at
54:42 Microsoft
54:43 obviously um but there you go that's it
54:46 full free fall free I just want to tweak
54:49 a photo I take and then paint from there
54:54 aou you've got roast
54:56 [Laughter]
54:59 beef oh a jumping off place
55:04 um what I would probably do in that case
55:07 Mary Mary is is take the image into um
55:12 into Photoshop and do generative fill
55:15 right so just select parts of the image
55:17 maybe you extend the width of it or you
55:19 change the change the sky to a stormy
55:22 sky instead of a sunny Sky something
55:25 like that
55:26 eclipse tomorrow is messing around with
55:28 a lot of us oh what's going
55:32 on
55:33 ah that
55:43 stinks
55:47 uh feel better Silver Fox that sucks
55:53 um aoy Irregulars don't forget to drink
55:56 water thank you Ron it's a very good
55:58 call
56:03 CH that's tasty Robert Rousy that's a
56:05 lovely orange a lovely orange indeed
56:07 good day to you sir good day sir may I
56:10 inquire about something go on your
56:15 qualifications yes what are they my what
56:20 your qualifications what are your
56:22 qualifications sir sir is it not clear
56:26 to what I am
56:28 inquiring aoya regulars don't forget to
56:30 drink water wake me up when I can get
56:33 one of the AI what was that Sherry Banks
56:35 to push buttons on my Android or PC apps
56:38 or programs Karen uh
56:42 vigilantism it's
56:46 coming I would say
56:48 probably I'm guessing probably about a
56:51 year until we see any sort of autonomous
56:54 autonomous agents taking
56:58 action I
57:00 suspect but I could be really wrong that
57:03 apple is going to do it first that
57:05 Apple's going to do it first best that
57:07 they're going to they're going to I
57:09 think Apple's going to kind of redefine
57:12 generative AI
57:15 as agents that do [ __ ] on our behalf but
57:19 I could be wrong oh that's a cool one
57:21 look at looky what looky what Ted Ted
57:24 made for us
57:26 that's really cool I love the colors
57:28 it's like a it's like a magazine
57:30 illustration magazine
57:31 cartoon that's cool
57:42 um subject matters indeed the sound
57:46 keeps going and in and out anyone else
57:49 oh that sucks even like is it right
57:52 after I switch screens or is it just
57:55 going in and out in
57:57 general it could be Siri eating my
57:59 microphone too tell me the best way to
58:02 get learn wait to get learned
58:07 what I lost that comment dang
58:18 it that's why you wear glasses sitting
58:21 close to the
58:22 TV one of seven turns of the do
58:25 oh tell me the best way to get learned
58:27 on
58:28 Photoshop
58:31 um Photoshop is you know photo
58:35 photoshop's one of those things where
58:37 you can learn the basics in an hour and
58:39 spend a
58:41 decade mastering it it took I I remember
58:45 I was I did desktop publishing when I
58:47 was an actor in New York I did desktop
58:48 publishing is my day job I was working
58:51 with Photoshop
58:54 for like two and a half or three years
58:57 when it finally sunk in the relationship
59:01 with dots per inch and size of
59:03 image it's cuz it's it's just it's still
59:07 a really convoluted concept to get in
59:11 your head but like when I got it in my
59:12 head I was like oh but it's like some of
59:15 the concepts take
59:18 forever um
59:21 Ron when it's a gerent tocy any thoughts
59:24 on government regulation of these tools
59:26 when it's a gerontocracy yeah let me
59:28 talk to that in a second let me go let
59:29 me go show some photoshop
59:36 stuff
59:43 Photoshop I don't
59:47 I I
59:50 think
59:52 that this Photoshop has oh this is a
59:56 beta okay so this will definitely have
59:57 the stuff in
1:00:03 it okay so let's
1:00:05 grab I'll just grab that
1:00:08 image a new version is available update
1:00:19 now let me just open Real Photoshop I
1:00:23 assume it's got generative fill in it
1:00:30 [Music]
1:00:36 now let me see if I can get you up to
1:00:38 speed here Mary
1:00:40 Mary I'm not weird I'm
1:00:46 irregular cute right okay so let's close
1:00:52 this don't save close
1:00:56 that all right let's open this image
1:00:59 again
1:01:01 okay are we going there we go so there
1:01:04 we've got an
1:01:07 image that was made in stable diffusion
1:01:11 this was an image that that kind of blew
1:01:13 my mind this this was one of those
1:01:14 things where when I saw when I saw this
1:01:17 I was like oh my God that's the kind of
1:01:18 thing I would make if I had the skill
1:01:21 and the talent but I don't but now I do
1:01:24 this was one of those revelatory things
1:01:26 for me okay
1:01:28 so here's the original image and I just
1:01:31 grabbed the crop tool and the crop tool
1:01:33 you can either crop in or you can crop
1:01:34 out so I just expanded the crop tool
1:01:37 wider so the image is now going to be
1:01:40 wider than what the original is but
1:01:43 there's this what they call a contextual
1:01:45 menu here that says generative expand so
1:01:48 I can click on that and and I can type
1:01:51 something in um or I can just say
1:01:54 generate so I'm not going to type
1:01:56 anything in I'm just going to let it
1:01:57 analyze this image and then fill in the
1:02:00 background so literally just say just
1:02:02 click on the generate
1:02:04 button um where is my mouse there it is
1:02:09 and so what it's going to do is it's
1:02:11 going to go do its AI [ __ ] what however
1:02:13 the hell it does
1:02:16 that and then it'll just fill out the
1:02:19 image right so it extended this little
1:02:21 you know thing down here and it'll
1:02:23 create three versions of it so there's
1:02:26 one with more more of that
1:02:28 [ __ ] and there's one where they kind of
1:02:30 look like wings that one's kind of cool
1:02:32 but I
1:02:34 think H that one looks more organic that
1:02:37 one looks more stylistic and artistic or
1:02:40 stylized okay so now we're going to do
1:02:42 this now I'm going to go and I'm going
1:02:44 to draw a
1:02:45 little selection right
1:02:48 here and I'm going to
1:02:51 put
1:02:53 spooky
1:02:56 eyes
1:02:59 spying and let's see if we can get
1:03:02 someone off in the shadows to be spying
1:03:05 on the big allseeing
1:03:09 eye sort of it's creepy but not quite
1:03:12 right I don't know what that
1:03:15 is yeah that's pretty creepy that's like
1:03:17 a dude's face all right that's
1:03:20 good and then over here we're going to
1:03:23 put um
1:03:31 zombie let's see zombie
1:03:35 um deer
1:03:40 skull we're going to take this image
1:03:42 from creepy to downright disturbing is
1:03:44 what we're going to
1:03:51 do no that's no not even close
1:03:56 no those are bad all right so we're not
1:03:59 going to do zombie deer skull we're
1:04:00 going to do um let's do
1:04:08 um um
1:04:10 spooky ghost
1:04:15 face hopefully this
1:04:17 will be less shitty than that but anyway
1:04:21 this is this is how you can do it so you
1:04:22 can extend images you can change the
1:04:24 shape you can you can just highlight the
1:04:28 sky and tell it to put in a different
1:04:30 kind of Sky you could highlight the
1:04:32 mountains and tell it to make it a plane
1:04:34 you could highlight a car and tell it to
1:04:37 change the color of
1:04:39 it okay that is not a spooky
1:04:42 ghost nor is that nor is that all right
1:04:46 all these are horrible so I'm going to
1:04:48 get rid of those so
1:04:50 anyway um
1:04:52 yeah but that's that
1:04:55 I mean it is kind of it's Halloweeny
1:04:57 it's just not particularly
1:05:00 good um but that's that's Photoshop
1:05:03 generative film um so hopefully that
1:05:07 helps you a little
1:05:09 bit um the other thing that you can try
1:05:13 is um Adobe Firefly has got all the
1:05:17 generative fill and expand stuff as well
1:05:20 it's a little more simple than Photoshop
1:05:22 photoshop's kind of like you know an F1
1:05:25 car if you're just trying to do
1:05:26 something simple it's it's a lot it's a
1:05:29 lot it's a lot to
1:05:34 learn quick Creative Cloud AI learning
1:05:37 lab someone asked about the regulation
1:05:39 and
1:05:40 and um how do I feel about government
1:05:43 regulations when our elected officials
1:05:48 can't even [ __ ] understand what WiFi
1:05:51 is and how it works um good question I
1:05:54 have some thoughts champ get up there
1:05:57 all right you good all
1:06:01 right um some of the things that I've
1:06:04 seen that are that are heartening
1:06:06 are the players from the big AI
1:06:09 companies open Ai and Google and
1:06:11 Microsoft and anthropic and those guys
1:06:15 is that you know they're actively
1:06:18 getting in front of those people I think
1:06:21 that the fact that Jeffrey Hinton quit
1:06:24 Google so that he could educate people
1:06:27 about this stuff I would imagine that
1:06:30 they will create the equivalent of the
1:06:33 FAA or you know some regulatory
1:06:36 body
1:06:39 that looks at the strength of AI and
1:06:43 regulates it differently for different
1:06:46 kinds of companies and this is this is
1:06:49 something so so the big companies got
1:06:51 together and they created this thing
1:06:53 called the front to your model Coalition
1:06:56 or Foundation or something like
1:06:58 that basically what they're saying is if
1:07:01 you can afford 20,000 gpus to dra to
1:07:04 train one of the frontier models the
1:07:06 base model like
1:07:08 gp4 you need to be dealt with on a
1:07:10 different level you need to be
1:07:12 regulated if you're doing something
1:07:14 significantly Beyond gp4 in terms of
1:07:17 capability that should probably be
1:07:20 licensed like you license a nuclear
1:07:22 power plant like you license you know
1:07:24 you know the ability to fly and build
1:07:26 airplanes that if you're building things
1:07:29 that are significantly more powerful
1:07:31 than than what we have today which are
1:07:33 already pretty [ __ ] powerful that
1:07:35 probably looks like licensing if you can
1:07:37 build one of these Frontier models that
1:07:39 probably looks like regulation and then
1:07:41 what they're basically saying is below
1:07:43 that don't regulate too much because
1:07:45 you're going to stifle um you're going
1:07:47 to stifle Innovation right that's always
1:07:49 the balance you strike there so I think
1:07:52 the fact that you've got the industry
1:07:55 aggressively saying hey this [ __ ] is
1:07:59 serious take it serious and and you know
1:08:02 you have people like Jeffrey Hinton and
1:08:04 Ethan mullik floating around that um are
1:08:08 resources that the government can tap
1:08:11 into
1:08:13 um
1:08:15 promising but are they going to [ __ ] it
1:08:18 up before they get it right
1:08:23 probably probably probably probably
1:08:26 probably Tik Tok is lagging so badly
1:08:29 tonight I've been doing a lot of
1:08:30 switching back and forth I have a
1:08:32 feeling they need an F FTC type of body
1:08:35 for cloud storage too oh yeah that's
1:08:37 interesting yep yeah so
1:08:40 so
1:08:43 um the I mean the the thing that's kind
1:08:45 of weird about it is
1:08:48 that meta and stability AI op Source
1:08:52 their their Foundation models
1:08:55 so like the the the weights the the
1:08:57 actual Secret Sauce of some major models
1:09:01 are
1:09:05 out so even if the government says hey
1:09:07 you can't make any more of those well
1:09:09 they're already out there and so you
1:09:11 know there's no safety guard rails on
1:09:13 them you can do anything you want with
1:09:14 them and apparently they're pretty
1:09:15 trippy if you if you start playing with
1:09:17 them in that
1:09:18 way um but you know I I would certainly
1:09:22 argue that I don't think G pt4 at this
1:09:25 point is is super dangerous but you know
1:09:30 there's there's a couple interviews with
1:09:31 Sam Alman where he's like oh we've built
1:09:33 [ __ ] that we will never
1:09:35 release when I hear statements like that
1:09:38 I'm like okay that shit's getting leaked
1:09:40 at some point or someone else is going
1:09:44 to you know leave open Ai and go build
1:09:46 that thing and release it so I don't
1:09:48 know what that [ __ ] is but you know it's
1:09:50 probably got to do with I don't know
1:09:52 identifying people or or you
1:09:55 know finding all the information on the
1:09:58 internet instantly about any given
1:10:00 person or some sort of crap like that um
1:10:03 I don't know I don't listen I I I am
1:10:07 not I am not well-versed enough in the
1:10:10 government to know what I'm talking
1:10:12 about at all I'm just kind of cynical
1:10:14 about it and assume that the financial
1:10:17 interests will write the laws and then
1:10:19 we will just suffer the
1:10:23 consequences learning
1:10:25 Photoshop yeah well I you know I I
1:10:28 figure I figure teaching a little bit
1:10:31 about Photoshop just because of the
1:10:32 generative fill because it's it like the
1:10:36 nice thing about
1:10:38 Photoshop having generative fill is no
1:10:41 matter what the limitation is of the
1:10:43 original image you can kind of go easily
1:10:45 fix it within Photoshop right now or
1:10:47 modify slight little things within it
1:10:49 and if it doesn't quite get it right
1:10:51 it's Photoshop just go Photoshop it
1:10:53 right just cop copy and paste in some
1:10:55 other chunk of another image and so I
1:10:57 think it's actually an important um tool
1:10:59 to use although I don't think we're
1:11:01 going to need need to be tweaking these
1:11:04 things nearly as much a year from now
1:11:06 you know
1:11:08 um so once my
1:11:11 DNA data storage startup has our first
1:11:14 proof of concept will be our own digital
1:11:19 storage is that really what you're
1:11:21 working on Ron or is that kind of a joke
1:11:23 cuz
1:11:24 cuz I I know that you can you can uh you
1:11:29 can put data in DNA right so is that is
1:11:33 that really what you're up to that's
1:11:35 some [ __ ] Brainiac [ __ ] there or
1:11:37 what's it's a joke
1:11:40 cuz that's what what was the what was
1:11:43 the uh was it a who was
1:11:48 the what was the movie where where
1:11:50 they'd go and they the guy would sort of
1:11:53 vacuum up any hair that fell off him and
1:11:56 he he would he would dust up his thing
1:11:58 so that they couldn't get his DNA for
1:12:01 exactly that
1:12:12 reason I'm certified in Photoshop I
1:12:14 never use it anymore I know I
1:12:18 know well it's funny it's funny like I
1:12:20 don't use Google anymore like I'll I'll
1:12:23 use chat GPT instead of Photoshop like
1:12:27 even if I know it's going to take longer
1:12:28 and it's probably not going to be right
1:12:30 I'm just like [ __ ] it I'll just use chat
1:12:32 gbt and I'm the same with the image
1:12:34 generation now I'm like I can go
1:12:35 Photoshop this I'm like oh [ __ ] it I'll
1:12:37 just prompt again uh Rudy SEO good
1:12:40 morning from South Africa 5:14 a.m.
1:12:44 cheers here's my
1:12:49 coffee oh what a cool image Joker check
1:12:51 this thing out that joker did
1:12:58 AI learning lab a dude with all the like
1:13:01 things in the
1:13:02 background with some misspellings but
1:13:05 most of it's pretty good like Adobe
1:13:06 Firefly chat J you can make money CH J
1:13:11 Bang mid
1:13:12 Journey that's pretty
1:13:15 slick look at all those people learning
1:13:18 on their giant wide
1:13:21 monitors that's cool image man that's a
1:13:24 cool image Joker Joker MIT Harvard
1:13:27 already do DNA data storage
1:13:35 nice you can tell Ted Ted looks at
1:13:38 Life's a little snacky I had pizza for
1:13:41 dinner tonight that's that's very
1:13:43 psychic of
1:13:46 you nice oh I just noticed that look
1:13:50 what the box says Kyle
1:13:52 delivers
1:13:58 oh man y'all are weirdos oh wait you're
1:14:01 not weird you're
1:14:02 irregular um by the way that's a new
1:14:05 something that was innovated here at the
1:14:07 AI learning lab if you get back an image
1:14:10 that's got like bad spelling on
1:14:13 it um it's not wrong it's irregular all
1:14:18 right so just my request is you adopt
1:14:21 that language just moving forward
1:14:24 any AI generated image that's not quite
1:14:26 right it's just irregular remember going
1:14:29 to the Kmart and getting the jeans that
1:14:31 were irregular and you you you had to
1:14:33 like look at them figure out what was
1:14:35 weird about them and it wasn't until you
1:14:36 put them on and they cranked on your
1:14:38 ball sack that you're like oh there's
1:14:41 the irregularity right
1:14:50 there h i do multilingual video content
1:14:53 for the federal government my job got
1:14:55 easier and more unreliable oh yeah I bet
1:14:58 Ron have you seen the have you seen the
1:15:00 [ __ ] I did have you been in here and
1:15:02 looked at the stuff I did with hey Jen I
1:15:04 assume you know about hey Jen yes
1:15:07 Ron um quick
1:15:13 time do I have pingpong yeah here's my
1:15:16 pingpong check this out Ron if you
1:15:18 haven't seen
1:15:19 it uh what's going on hang on I've got
1:15:23 it up in the cloud somewhere I do
1:15:30 believe there it
1:15:35 is so we've heard from a number of
1:15:37 clients that modular content is
1:15:39 important to them getting more efficient
1:15:41 with how we create content and how we
1:15:43 leverage
1:15:50 content uh behind me you see these
1:15:53 little green boxes um that's how we
1:15:54 think about stories at Story vine we
1:15:56 break stories down into little atoms
1:15:58 right little
1:16:09 modules um I actually took these videos
1:16:11 and I ran them back through Heen back
1:16:15 into English
1:16:18 and the Mandarin to
1:16:21 English was pretty good the Italian to
1:16:24 English was okay and the Japanese to
1:16:26 English was pretty bad so but it was
1:16:28 interesting oh is there a live going on
1:16:31 I am en raptured with my d new D3
1:16:34 nice there there's no live going on
1:16:37 there's nothing going on here you're
1:16:38 good source Camp just hang out we'll
1:16:40 we'll call you if if anything good
1:16:42 happens we'll let you
1:16:45 know I'm just going to go ahead and
1:16:46 check into rehab tomorrow I know yeah
1:16:49 Dolly 3 is amazing isn't it Source Camp
1:16:51 it's like well join the if if you
1:16:54 haven't joined the the AI Salon join the
1:16:56 AI salon and there's a show int tell
1:16:58 Channel as you're coming up with cool
1:16:59 [ __ ] with Dolly 3 just dump it into the
1:17:01 show Intel Channel or or um request
1:17:04 access to the Irregulars you're in here
1:17:06 you're in here a bit um yeah I know this
1:17:09 is
1:17:10 listen talk about check into rehab I I
1:17:14 go on Tik Tok live for like three hours
1:17:16 a night 7 days a
1:17:22 week
1:17:24 like what am I doing with my life I
1:17:27 don't know I'm going to keep doing it
1:17:29 though because I I just I'm get I'm I'm
1:17:32 giving a keynote tomorrow at at uh at
1:17:35 the the AI Peaks um the first AI
1:17:39 conference in
1:17:40 Colorado and I'm not done with my
1:17:44 presentation
1:17:46 yet but you know those of you that are
1:17:49 in here a lot understand why it's an add
1:17:52 thing but
1:17:54 um but it's also
1:17:58 like like I just stuff changes so much
1:18:01 like
1:18:03 like what it was what was important to
1:18:05 talk about 6 months ago is completely
1:18:08 different than what we should be talking
1:18:10 about today so I'm just trying to like
1:18:13 I've got the bulk of it done I've got
1:18:15 like 85% of it done and I'm just trying
1:18:16 to figure out like what do I close with
1:18:19 what do I close
1:18:20 with and I think I'm going to close with
1:18:23 basic what this channel is which is
1:18:26 play build
1:18:29 share like just get curious start
1:18:32 playing just play play play play play I
1:18:35 talk about it all the time just go just
1:18:36 go to chat gbt just start playing what
1:18:39 am I supposed to do there it doesn't
1:18:40 [ __ ] matter just play and then when
1:18:43 you realize oh I could make a this go
1:18:47 build that and it doesn't mean build an
1:18:49 app it could be build a prompt to do
1:18:52 some little piece of your business it
1:18:54 could go mean build some images it could
1:18:57 could mean go build an application using
1:19:00 zappier which is a no code tool where
1:19:03 you just string a couple of prompts
1:19:05 together and then share what you're
1:19:07 learning start a [ __ ] Tik Tok channel
1:19:09 it's funny I was on
1:19:12 um I was on my my LinkedIn Friday office
1:19:16 hours today which by the way you're
1:19:18 welcome to come hang out there if you
1:19:19 want as well and Jim Ross was in there
1:19:23 who's a who's an irregular
1:19:25 here and he was asking me hey Kyle do
1:19:28 you know anyone that's doing like like
1:19:30 small business how small businesses are
1:19:33 using AI content right now I'm like no
1:19:37 and I'm like but that's a really good
1:19:38 idea like you should go do that like
1:19:41 that's needed CU he's got credibility
1:19:43 there he's like a property manager and
1:19:45 he's not a technologist right he's just
1:19:48 like I'm learning these tools and as I'm
1:19:50 learning these tools I'm building [ __ ]
1:19:52 for my business you you know what he did
1:19:53 he wrote a 70 page 75 page
1:19:56 booklet to give to his owners or
1:20:00 investors or I don't know whatever one
1:20:01 of the work compatriot types he's he's
1:20:05 writing books he's making marketing
1:20:08 social media posts he's doing all this
1:20:10 stuff that he would not have done had he
1:20:13 not been using AI I'm like that's a
1:20:14 [ __ ] like go share that go tell
1:20:18 that so I think that's how I'm going to
1:20:20 end end the thing tomorrow is like just
1:20:22 get curious start playing start building
1:20:25 start telling people what you're
1:20:26 learning find other weirdos that are
1:20:29 doing
1:20:32 this we're not weird we're
1:20:37 irregular you can make money like
1:20:42 CH
1:20:43 yeah if you couldn't make money with
1:20:45 chat GPT would I have a gold button I
1:20:48 don't think
1:20:49 so you think that's from China no that's
1:20:52 solid gold
1:20:54 I know Parker from Gold Rush I had him I
1:20:58 had him melt this down for me that's
1:21:00 like from this from season 4 all the
1:21:03 gold from season 4 is in this button you
1:21:06 can make
1:21:07 money making so much money I can pay
1:21:10 Parker to make me a gold
1:21:15 button I know you're jealous it's
1:21:21 okay uh
1:21:24 [Music]
1:21:29 I pitched it to our decision makers but
1:21:31 they're newspaper people oh good God and
1:21:33 think I'm selling snake oil yeah Ron
1:21:37 well so here's the
1:21:40 deal there are certain industries oh
1:21:43 that's a really cool one Ted damn that's
1:21:46 really cool what was the prompt for that
1:21:48 look at that thing I want that
1:21:51 lunchbox that that's like that's like
1:21:54 somewhere between a lunchbox a Mac and a
1:21:59 trs8 [ __ ]
1:22:01 beautiful one of the original Macs
1:22:04 that's super
1:22:05 cool
1:22:08 dag dag I want to know the
1:22:13 prompt what was the
1:22:17 prompt that and that is so cute all
1:22:22 right um oh newspaper people so so my
1:22:26 company storyvine does automated video
1:22:29 storytelling and like one of the things
1:22:31 I've thought for years is that a
1:22:34 platform like storyvine could be
1:22:36 absolute Insanity for citizen journalism
1:22:39 so imagine people all over the country
1:22:42 or all over the world oh thank you very
1:22:44 much Winston I appreciate
1:22:47 that appreciate appreciate appreciate
1:22:50 imagine them being able to have a tool
1:22:53 that walks them through a structured
1:22:56 news story they just answer a series of
1:22:59 prompts and that goes up to the cloud
1:23:01 and you could have local journalism that
1:23:03 goes into a National Database for
1:23:05 everyone you know National could pull
1:23:06 Clips out for the national stories local
1:23:09 could do things for the local you could
1:23:11 get um citizen journalists in every
1:23:13 small town across the state all they
1:23:15 would need is like Wii access and a
1:23:18 smart and a
1:23:20 phone and we even did a project with one
1:23:26 of the local um ganette or what's it
1:23:30 called tegna now one of the ganette
1:23:31 Affiliates here in Denver we did this
1:23:34 small business thing we came up with a
1:23:35 whole new product that they could go out
1:23:37 and sell digital they they they couldn't
1:23:40 get out of their own way like like media
1:23:43 companies in particular are so [ __ ]
1:23:46 married to their existing business model
1:23:50 that it's it's just going to be like
1:23:53 it's just going to be like Kodak right
1:23:54 when the dude walked in from from Kodak
1:23:57 and said hey we invented the digital
1:23:59 camera and they're
1:24:00 like but we're not we're not in the
1:24:02 pixel business young
1:24:04 man we're in the silver Halli business
1:24:08 uh this is all about the chemicals and
1:24:10 the the paper it's about real
1:24:12 photographs if you know what I mean you
1:24:15 know and what was it 10 years later they
1:24:17 were out of business so we're going to
1:24:19 see a lot of that [ __ ] A lot of that
1:24:22 [ __ ] lot lot lot lot lot lot lot lot lot
1:24:26 lot lot lot lot of [ __ ] of companies
1:24:29 just having their head actively buried
1:24:31 up their ass because they don't want to
1:24:34 look at what this is bringing because
1:24:36 what it's bringing it's it's going to
1:24:40 upend [ __ ] every business model as
1:24:43 far as I can tell like I can't I can't
1:24:45 look at a single business model right
1:24:46 now and not see like oh yeah that's
1:24:50 probably going to be
1:24:51 different
1:24:54 and then what's that mean if the if the
1:24:56 business models themselves are going to
1:24:57 be different like the one that I know is
1:24:59 the advertising industry I know that I
1:25:01 know it really
1:25:02 well they make their money on time and
1:25:05 materials billing right you hire us
1:25:08 we'll put 50 Junior copywriters in a
1:25:11 bullpen and we'll have them write all
1:25:13 your social media cop copy and we'll
1:25:15 charge you
1:25:16 $120 per hour per person times 50 for
1:25:20 the month for the next three years they
1:25:23 make a lot of money like that and Along
1:25:25 Comes A Tool that's like uh what if you
1:25:27 didn't have to have all those 50 people
1:25:29 to write that same amount of
1:25:31 copy the people are super
1:25:34 important the ad
1:25:36 agencies don't care about the people
1:25:39 other than those are their billable
1:25:41 hours that's their business model so and
1:25:46 and in fact I won't say the name of the
1:25:48 agency we had an agency that we were
1:25:50 talking to that their biggest client
1:25:52 which was was which is the largest
1:25:54 retailer in the world their biggest
1:25:57 client said to them we want to increase
1:25:59 the use of social
1:26:01 video and we showed them our storyvine
1:26:03 platform and after three months of
1:26:06 [ __ ] us around they said
1:26:10 listen they kind of said it like
1:26:13 this we love what you do what your
1:26:16 technology does actually solves the
1:26:19 problem that our client has but that's
1:26:23 not our business model we make our money
1:26:26 by having our video editors sit there
1:26:28 for hours and crank out billable
1:26:32 hours good luck to
1:26:35 you
1:26:37 so like we're just one little company
1:26:40 and we're you know we're not using AI
1:26:42 automated stuff so but but like there's
1:26:45 a
1:26:46 mentality
1:26:48 that as their clients get more and more
1:26:51 educated about what what's possible with
1:26:53 these tools they're not going to
1:26:54 tolerate time and materials they're not
1:26:56 going to tolerate being [ __ ]
1:26:59 gouged so the clients are going to start
1:27:02 saying listen I don't really care about
1:27:04 your business model I want you to
1:27:06 produce the
1:27:08 same quality work or
1:27:11 better twice as fast for half as much
1:27:15 and if you can't do
1:27:17 it Jim and Sally over here that used to
1:27:20 work at your agency they just started
1:27:22 10p person shop where they're already
1:27:27 delivering at the same or higher quality
1:27:30 twice as fast and half as much as what
1:27:32 you're charging right now so
1:27:35 either you know I take them from
1:27:38 $100,000 a year to $10 million a year or
1:27:41 you cut your rates in half you let me
1:27:44 know
1:27:46 bye every business is going to be faced
1:27:49 with
1:27:50 that what what are they supposed to do
1:27:52 to do
1:27:54 well I don't know the Smart Ones are
1:27:57 going to go uh so who was that it was
1:27:59 Ron right Ron was up there pitching
1:28:02 pitching Ai and and and the powers that
1:28:04 be were like Ron Ron with his artificial
1:28:07 intelligence doesn't he know that
1:28:08 there's copyright
1:28:12 issues Jane bring me a
1:28:16 martini and at some
1:28:18 point one of those guys is going to get
1:28:20 smacked upside the head with the real of
1:28:22 this and get that Ron guy back in here
1:28:25 oh Ron quit Ron started the company
1:28:28 that's our competitor right
1:28:31 now that that guy with his stupid
1:28:37 a and then Ronald take down the big
1:28:43 publisher or the big publisher says well
1:28:48 this is going to be painful but we got
1:28:49 to figure this [ __ ] out we got to figure
1:28:51 it out now
1:28:54 now you know you know you know what I'm
1:28:58 saying do you know what I'm saying you
1:28:59 know what I'm saying you know what I'm
1:29:00 saying do a regulars remember TJ Maxx of
1:29:04 course what's the best AI generator for
1:29:06 marketing chat gbt 4 there listen there
1:29:10 there are tools like like Jasper and
1:29:14 copy Ai and I think um hubs spots
1:29:17 putting a lot of stuff in um canvas put
1:29:21 a put a bunch of AI stuff
1:29:24 in so so there are lots of platforms
1:29:27 that are incorporating AI
1:29:29 in
1:29:32 but I think it it will serve you if
1:29:39 you learn how to do what Jasper and copy
1:29:43 AI are doing like what those sites are
1:29:45 is basically a collection of pre-written
1:29:48 prompts where you push a button and it
1:29:51 makes a blog post
1:29:53 learning how to talk to these large
1:29:55 language models and
1:29:58 and keeping up with kind of what they're
1:30:00 doing and how they're moving from single
1:30:03 purpose like text only to multimodal
1:30:05 where you can do text and
1:30:08 images is going to serve you so if you
1:30:11 don't have time for that [ __ ]
1:30:16 then I would strongly recommend fat or
1:30:20 F start following a go guy named Paul
1:30:31 rer this dude from the marketing AI
1:30:34 institute.com
1:30:37 they that organization's been around for
1:30:40 like five years like they they have no
1:30:43 right being around for five years like
1:30:45 chat GPT just came out last November so
1:30:48 he started this thing I think right
1:30:51 around the time
1:30:53 it was probably gpt2 the image
1:30:55 generation tools were just coming out
1:30:58 gpt2 was
1:31:00 out because gpt3 was
1:31:02 2020 and it's a 5-year-old organization
1:31:06 so they've been thinking about this
1:31:07 stuff from a marketing perspective for
1:31:10 longer than anyone else so I would
1:31:12 follow him I would also follow Ethan
1:31:15 mullik he he leans a little more
1:31:18 into um future of education but there's
1:31:22 a little a little bit of future of work
1:31:23 there and and a lot of what he talks
1:31:25 about is applicable for um for marketing
1:31:32 so go find the tools but learn the [ __ ]
1:31:36 underneath them like learn what these
1:31:38 things are actually making possible and
1:31:40 I think it will serve you better you'll
1:31:42 be able to better evaluate those tools
1:31:44 right because those tools might be
1:31:46 remarkable I think some of the stuff
1:31:47 that canva just released is pretty slick
1:31:50 but it it seems to me that canva right
1:31:53 now is throwing like all the pasta at
1:31:55 all the walls and what that what that
1:31:58 probably is going to look like it's
1:31:59 going to be a confusing mess for some
1:32:01 amount of time from an interface
1:32:05 perspective God man I have some sort of
1:32:08 something going on I think it's uh I
1:32:11 think it's
1:32:13 allergies but I don't really have
1:32:15 allergies so but I hear it's it's been
1:32:18 pretty bad this this
1:32:19 fall already
1:32:26 um Irregulars you're up to 177,000 likes
1:32:30 by the way if you're new here welcome my
1:32:32 name is Kyle Shannon this is the a
1:32:34 learning lab this is what we do here we
1:32:35 just talk we ramble I'll occasionally
1:32:37 make a silly voice or a cuss word or
1:32:40 whatever and then occasionally you can
1:32:42 make money with thank you Sherry Banks
1:32:45 that's a cute little twey
1:32:46 bird cute little Tweety Bird I like that
1:32:49 Tweety Bird just going to go on ahead
1:32:51 and check check into rehab tomorrow yeah
1:32:54 right after right after this last
1:33:02 image I'm part of a uh I'm part of a
1:33:09 group this is a group of fishing
1:33:12 buddies and see their their hats that's
1:33:16 Ivan the dude in the red see see how it
1:33:18 says one more
1:33:21 cast one more cast just one more cast
1:33:25 [Laughter]
1:33:28 buddy same thing just one more
1:33:30 image
1:33:32 [Laughter]
1:33:37 crackhead oh that's
1:33:40 funny I learned it from watching you
1:33:42 Kyle I learned it from you oh that's
1:33:44 fantastic yeah you know what those uh
1:33:49 them Publishers with their big cigars
1:33:51 and their their uh High balls of
1:33:54 whiskey they they're going to figure it
1:33:56 out at some point I I I've probably said
1:33:59 this on here before in
1:34:03 199 five or early 96 I was invited to
1:34:07 speak at a newspaper Publishers
1:34:10 conference about the worldwide web and I
1:34:14 was a cocky 30s something and I was like
1:34:17 like I am now I was all high on on what
1:34:19 it was you're the reason I found AI oh
1:34:21 thank you ANTM
1:34:22 um I was all high on the worldwide web
1:34:25 and I'm and like I could I could see
1:34:27 what was coming it was just like there
1:34:28 there was something about the web that
1:34:29 was just crystal clear to me the a AI is
1:34:31 much less clear cuz it's it's so
1:34:34 profound and it's so big and it's all
1:34:36 over the web was just kind of a a very
1:34:38 simple sort of piece of functionality so
1:34:41 you could kind of see the future a bit
1:34:42 more clearly and so so I go to this this
1:34:46 newspaper Publishers event and I said
1:34:49 whatever I said and then I and then I
1:34:51 went and I sat on the front of the stage
1:34:54 and and I just without the microphone I
1:34:56 just said to the room I said the thing
1:35:00 that you should really consider
1:35:02 doing is understanding how to blur the
1:35:05 line between editorial and
1:35:08 advertising and the room went [ __ ]
1:35:10 dead
1:35:13 silent and the organizer of the event
1:35:16 jumped up on stage and said okay thanks
1:35:20 Kyle that was awesome and he walked me
1:35:22 off stage out of the building and said
1:35:26 thanks it kicked me out of the
1:35:28 [Laughter]
1:35:32 conference but then look what happens
1:35:35 the the line between advertising and
1:35:38 editorial is you know it's [ __ ]
1:35:43 smeared together like it's never
1:35:45 been but
1:35:47 whatever you know it's just like it's
1:35:49 just I think it's it's actually just
1:35:52 human nature it's not even business
1:35:53 nature it's human nature
1:35:55 to to fear the new [ __ ] and to fear
1:35:59 change um and I get it I mean I get that
1:36:03 it's really [ __ ] terrifying that it's
1:36:05 like existentially
1:36:08 terrifying I mean if you think about an
1:36:10 advertising holding company right they
1:36:12 own hundreds of agencies they employ
1:36:16 tens or hundreds of thousands of
1:36:19 people and their entire model is based
1:36:26 on a production method that is now
1:36:32 irrelevant like are you still going to
1:36:34 need to handcraft some stuff yeah some
1:36:37 stuff but think about a pyramid so so
1:36:41 what's going to happen with content I
1:36:43 think let me I need a shape I need
1:36:46 [Music]
1:36:48 a I ain't got nothing all right I'll
1:36:50 have to do hands so here's your here's
1:36:53 here's [ __ ] you make right as an
1:36:56 agency at the very tippy top is like
1:37:00 awardwinning Super Bowl commercials Emmy
1:37:03 award-winning TV shows right the next
1:37:07 layer down is like it's got to be really
1:37:09 [ __ ] good the next that layer down is
1:37:12 like yeah it's got to be good and the
1:37:14 and the bottom layer is like just give
1:37:16 me some [ __ ]
1:37:18 content what AI is doing is it's
1:37:21 automating from the bottom up well
1:37:24 agencies make most of their money from
1:37:27 here
1:37:28 down from here down so if this is being
1:37:33 automated from down here up there's a
1:37:36 problem right there's not enough of top
1:37:39 of the pyramid work to employ all those
1:37:43 people and there's a need for more and
1:37:45 more content cheaper and cheaper because
1:37:48 we can do it
1:37:50 right
1:37:53 like that's [ __ ]
1:37:55 terrifying but it's also one of the most
1:37:58 exciting opportunities like in the
1:38:00 history of that industry the question
1:38:03 is does that industry participate in the
1:38:07 upside or do they get their asses handed
1:38:09 to them do they become the next Kodak do
1:38:12 they become
1:38:14 Blockbuster oh we got the stores we got
1:38:16 all the tapes no innovating of course we
1:38:19 innovated we went from vide tapes to
1:38:21 DVDs you know what's nice about them
1:38:23 they're so much
1:38:26 thinner Red Box what's red box why would
1:38:29 anyone buy a d you know rent a DVD
1:38:32 outside of 7-Eleven that's
1:38:34 hilarious we sell boxes of candy you
1:38:37 don't understand it's the movie
1:38:39 experience you get to come experience
1:38:43 renting a movie at a
1:38:46 store they're out of
1:38:50 business why cuz you can go like this
1:38:52 now click oh there's the
1:38:58 movie they didn't understand what
1:39:00 business they were
1:39:02 in
1:39:04 so and and I you know like the ad the ad
1:39:08 business is is just the one I know think
1:39:10 about legal think about the law business
1:39:13 think about Consulting holy [ __ ] [ __ ]
1:39:16 are you kidding
1:39:18 me a lot of what you need Consulting
1:39:22 organizations for is to go hey I I'm too
1:39:25 busy could you tell me what's happening
1:39:27 with these three
1:39:29 Industries sure we'll get back to you in
1:39:31 a month with a big
1:39:33 report that'll be
1:39:37 $150,000 oh and first class
1:39:40 accommodations okay I guess now they're
1:39:44 just
1:39:45 like what's going on in these three
1:39:49 Industries here's some data files let me
1:39:52 stick them in Advanced Data analysis in
1:39:54 chat GPT for 20 bucks a month and run my
1:39:58 own
1:40:00 analysis I can do that in half an
1:40:03 hour Man Ted looks at life's killing
1:40:06 it this is so
1:40:10 cool these are moving over into the into
1:40:13 the uh the images here these are great
1:40:17 The Lunchbox man this one's this one's
1:40:20 beautiful
1:40:24 beautiful beautiful
1:40:31 beautiful I'm not doing like all the
1:40:34 comedy stuff I normally do I'm like
1:40:36 getting a little philosophical here
1:40:37 tonight you
1:40:41 know oh you know who I just reminded
1:40:44 myself
1:40:45 of if you watched American Idol about I
1:40:48 don't know four or five seasons ago
1:40:50 there was a a wacky woman singer she was
1:40:54 like always like she would go
1:40:57 like I'm just like oh I'm [ __ ]
1:40:59 batshit crazy
1:41:01 [Laughter]
1:41:05 too that's exactly how I was received
1:41:08 Ron this AI
1:41:13 stuff can't compete with the federales
1:41:16 but I appreciate your spirit yeah
1:41:19 exactly uh no Emoji equivalent for your
1:41:23 presentations
1:41:26 thanks H yes go Ron but you know what
1:41:30 you know what Ron what's going to happen
1:41:33 is like I I know it I know it one of
1:41:40 those blow hards at some point is either
1:41:43 going to be golfing with someone or
1:41:46 they're going to be in some sort of
1:41:48 Consulting meeting or or I something's
1:41:52 going to happen one of those blo hards
1:41:53 is going to get religion about this
1:41:54 stuff I'm like we got to get our [ __ ]
1:41:56 together with this AI stuff who knows
1:41:59 anything about this well wasn't that guy
1:42:01 Ron in here that we kicked out and yeah
1:42:04 get Ron in here Ron's now the VP of
1:42:09 AI you're going to get a promotion Ron
1:42:12 just going to be a little painful
1:42:13 getting there and you might need to go
1:42:16 start a competitor to your newspaper
1:42:18 before then but other than that
1:42:19 everything's going to be just fine
1:42:24 [ __ ] [ __ ] media companies I don't
1:42:26 get media companies I've never gotten
1:42:29 media companies what the [ __ ] is going
1:42:31 on here oh I see oh we got an ink
1:42:41 no Tik Tok telling me great job you've
1:42:44 been live for 90 minutes I know my ass
1:42:46 is numb I understand how this
1:42:49 works um true heard that telling chat
1:42:52 gp4 take a breath fixes hallucinations I
1:42:55 don't know if it fixes it um so so
1:42:58 that's from a they so Ethan mullik one
1:43:01 of the guys that I say to go follow
1:43:03 there was a um a study that they did
1:43:06 where they
1:43:08 compared human beings writing prompts
1:43:11 and chat gbt writing its own writing
1:43:13 prompts for itself and it did better
1:43:16 like the prompts that chat GPT wrote for
1:43:18 itself performed better than anything
1:43:20 the did and one of the prompts that that
1:43:23 it wrote that was the most effective the
1:43:25 one that was the highest most effective
1:43:28 one was take a breath and then answer
1:43:32 carefully or answer there's another one
1:43:34 that's like um think step by step and
1:43:38 then answer this right and so you tell
1:43:40 it to think step by step um but yeah
1:43:43 it's Again part of the art of
1:43:47 understanding these large language
1:43:48 models is that they need a lot of
1:43:50 context they they need a lot of
1:43:52 understanding if you just say write me a
1:43:55 blog post it will write you a blog post
1:43:58 but if you start talking about you know
1:44:00 this particular kind of business and
1:44:02 this particular kind of Challenge and
1:44:04 these particular kind of experts and you
1:44:07 know this kind of nested context and
1:44:10 then say write me a blog post in that
1:44:13 context it's going to write you a much
1:44:14 better blog post but the that that study
1:44:19 that you're talking about you're
1:44:21 refering to there what the main point of
1:44:22 that was is have chat GPT like give it
1:44:26 all that context and then have chat GPT
1:44:29 write the prompt for you to give to
1:44:33 it what kind of degrees do you have I
1:44:35 have a BFA in
1:44:40 acting um I've got a uh the only thing
1:44:44 resembling I I've got a uh I did a a
1:44:47 Harvard Business School like a two week
1:44:50 thing through Omnicom I sold my company
1:44:52 to Omnicom and when they were minority
1:44:54 investors I did some sort of Harvard
1:44:56 business business school thing and
1:44:59 that's it that's it I had a I had a full
1:45:02 ride to the Alabama Shakespeare Festival
1:45:04 to get my MFA in theater but uh I got an
1:45:08 equity job the summer before so two
1:45:10 weeks before I was supposed to go to the
1:45:12 Alabama Shakespeare Festival I was like
1:45:13 [ __ ] you I'm going to New
1:45:16 York but I've been a I've been an
1:45:19 entrepreneur all my life so I've started
1:45:21 12 or 15 companies so like like the I
1:45:25 started agency.com which was one of the
1:45:26 early digital agencies we grew that
1:45:29 company from two people to 2200 people
1:45:31 in 5 years so I probably have a number
1:45:34 of degrees but I just don't have pieces
1:45:36 of paper to hang on the wall so I've
1:45:38 learned a lot over my years when GPT
1:45:40 struggles with a command I ask it to do
1:45:43 its best and then it produces
1:45:44 interesting pretty cool you're a
1:45:46 millionaire um I was for a while there
1:45:48 was a while where I was worth what was I
1:45:52 worth it was worth like $200 million but
1:45:55 I didn't sell a single share of my
1:45:57 agency.com
1:45:59 stock that's fantastic Bob tell him what
1:46:01 he's won you can make
1:46:06 money the misses isn't letting me live
1:46:09 that one down
1:46:10 [Laughter]
1:46:12 ever I took a little bit of money off
1:46:15 the table but it's all most of it's
1:46:17 invested in my current business right
1:46:19 now
1:46:22 but had a couple of good
1:46:23 years the millions are gone
1:46:28 however but it was wild times I mean you
1:46:31 know going from you know being a a
1:46:35 struggling actor in New York to starting
1:46:37 an agency and in five years growing it
1:46:40 you know we built websites for British
1:46:43 Airways for Coke for Nike for we built
1:46:47 Serius XM radios first streaming service
1:46:51 like streaming audio didn't exist we we
1:46:53 built their first streaming tool um it
1:46:57 was [ __ ] wild I mean it was like
1:47:01 nobody knew what they were doing it's
1:47:03 just like this AI stuff it's just like
1:47:06 this we won the coke business because we
1:47:10 took screenshots of the website that
1:47:13 their it group had built and we stuck it
1:47:15 in a PowerPoint
1:47:18 slide and we said here's Coke
1:47:21 and here's cocacola.com and we had the
1:47:24 headline where are the polar
1:47:26 bears and they stopped the meeting and
1:47:28 left the room there's like 15 Executives
1:47:31 from Coke there's like eight of us in
1:47:33 there trying to pitch this business
1:47:36 Alabama Shakespeare Festival sounds like
1:47:38 an SNL sketch sketch that's [ __ ]
1:47:43 brilliant so these C exex like stop our
1:47:47 presentation hang on stop right there
1:47:52 they they stormed out of the room we're
1:47:54 like oh [ __ ] what did we do [ __ ] are we
1:47:57 in
1:47:58 trouble and like 15 minutes later they
1:48:00 all walked back in and they're like when
1:48:02 can you
1:48:04 start they had never seen their website
1:48:07 we we showed them we showed them we
1:48:10 showed them you know that they're the
1:48:15 world's biggest brand and they
1:48:18 accidentally ignored it in what was an
1:48:21 emerging
1:48:23 medium and so that's that's what's going
1:48:26 to happen with this that's already
1:48:27 happening with this stuff right Ron with
1:48:29 his his newspaper buddies there we don't
1:48:32 need no stinking AI we got
1:48:35 writers you're a funny Jolly Anglo-Saxon
1:48:38 man well I thank you I
1:48:43 think he was the funniest Anglo-Saxon
1:48:45 I've ever seen yeah yeah see yeah
1:48:49 paluca um
1:48:52 who's that there's a comedian that that
1:48:54 does a whole thing on the last
1:48:58 time it was it was something about like
1:49:00 the last time white people forget how he
1:49:03 phrased
1:49:04 it I don't know like still had Moxy or
1:49:07 something like that it was like back in
1:49:08 that era in the movies they talk like
1:49:11 same yeah you
1:49:12 [Laughter]
1:49:15 paluca oh my
1:49:17 God yeah those were crazy times but I
1:49:21 mean that's it's like one of the reasons
1:49:22 I'm so excited about this stuff
1:49:26 is like I've seen the movie I've been in
1:49:29 the movie like I know this
1:49:33 movie I know
1:49:38 it and it's like it's just that thing
1:49:41 everyone thinks you're weird they think
1:49:42 you're a [ __ ] weirdo be a weirdo
1:49:44 right now be a
1:49:46 weirdo thank you for over
1:49:49 hooked for the little magic spell be the
1:49:52 weirdo right now get curious about this
1:49:53 [ __ ] start learning this stuff or you'll
1:49:56 end up like
1:49:57 [Laughter]
1:50:03 him Robert Rosy thank you very
1:50:06 much I love that chat GPT gold chat I
1:50:09 know right isn't that great thank you
1:50:11 very much um I am dyslexic and have ADD
1:50:16 so I am slower than lunatick Joker and
1:50:19 everyone
1:50:22 BS equals [ __ ] Ms equals more [ __ ]
1:50:25 PhD equals pile higher and deeper that's
1:50:27 good that's solid LOL are you hiring no
1:50:30 I wish I wish we had a shitty third
1:50:33 quarter like I've talked to a lot of
1:50:34 people that had a shitty third quarter
1:50:37 but um it looks like the fourth quarter
1:50:38 is going to be good but God it's just
1:50:41 been it's been a struggle by the way
1:50:43 this is cool look at this this is the AI
1:50:45 leaning lab somebody did that I didn't
1:50:48 do that one but that one's really good
1:50:51 with the Leaning Tower of Pizza you're
1:50:53 like an Uncle Buck for us thank you
1:50:55 there thank you there yeah Uncle Buck oh
1:50:58 Uncle Buck yeah
1:51:04 mhm all right let's see photo real EST
1:51:07 okay that was good saw that is anything
1:51:09 going on down here in Show and Tell oh
1:51:12 yeah lots of stuff oh that one's neat oh
1:51:17 wow I don't know who this is Rudy
1:51:23 pigo someone oh oh this is definitely uh
1:51:27 I think this is do but like interesting
1:51:30 kind of
1:51:31 fashion fashion looking shots those are
1:51:35 cool that's at the AI salon so if if
1:51:38 you're if you're interested if if you're
1:51:41 like if you want to be irregular right
1:51:44 if you want to be like one of these
1:51:45 people that are like I think there's
1:51:47 something with this AI [ __ ] I got to pay
1:51:49 attention to and then you you're paying
1:51:51 attention to it and you're secretly
1:51:53 getting excited about it and then as you
1:51:54 talk to people they're like you're a
1:51:56 [ __ ] weirdo if you want to find other
1:51:59 people that are like you are um I
1:52:03 started a group for
1:52:05 that it's kind it's kind of a it's kind
1:52:09 of a support group it's like AI
1:52:13 Anonymous the the
1:52:15 salon. so we meet every other week um we
1:52:19 meet physically here in Denver and then
1:52:21 we have uh we meet remotely as well and
1:52:26 then we have a Discord server we have a
1:52:29 LinkedIn
1:52:30 group so just join join start
1:52:32 participating ask questions share what
1:52:34 you're learning um we just started
1:52:37 guilds we've got a um Art Guild a
1:52:41 writing Guild a beginners
1:52:44 Guild a business Guild like AI in
1:52:48 business and then an entrepreneurs Guild
1:52:50 for startups and then Pate who's on here
1:52:52 a lot is going to start up a tech Guild
1:52:56 so like the math and ship behind the
1:52:59 scenes and then we're probably going to
1:53:01 do a nonprofit Guild and I think maybe a
1:53:03 video one maybe a music
1:53:05 one but there're these sub communities
1:53:08 within the community and it's cool and
1:53:10 there's like these showand tell
1:53:14 things
1:53:19 oh
1:53:23 hang
1:53:25 on I can't quite see
1:53:33 it I don't know what it says these are
1:53:36 those images you might be able to see it
1:53:37 from a distance these are those images
1:53:39 that have a word embedded in
1:53:41 them so squint your eyes you might be
1:53:43 able to see a word embedded in these I
1:53:46 can't quite see what they
1:53:49 are
1:53:54 that's super cool all right add equals
1:53:57 intense focus and easily distracted at
1:53:59 the same time
1:54:03 yeah Pioneers are rarely appreciated
1:54:06 yeah I know but it's you know for me
1:54:12 though like I it's so funny like like
1:54:15 that's a really interesting comment
1:54:17 Pioneers are not appreciated but
1:54:20 Pioneers
1:54:24 also here's what it is for me like I
1:54:26 don't feel like I have a
1:54:30 choice like I feel
1:54:32 like it's not even that it's it's not
1:54:35 like I'm being forced into it it's more
1:54:37 like what I recognize is that we're at a
1:54:40 singular point in
1:54:42 history November 30th 2022 was the
1:54:46 beginning the
1:54:48 beginning of a few future where how we
1:54:52 work and what we do looks nothing like
1:54:54 what it is today like that I know I know
1:54:58 that like that's not a question to
1:55:02 me and how do I know that
1:55:05 because two and a half weeks into
1:55:07 playing with chat
1:55:09 GPT here this was this was my Epiphany
1:55:11 moment this is something I'm going to be
1:55:13 oops be talking about in my talk
1:55:15 tomorrow is um what are your
1:55:18 qualifications we'll Zoom that
1:55:23 up I was sitting on the couch playing
1:55:25 with chat
1:55:27 gbt and I was like oh for years people
1:55:30 have been saying oh you should learn
1:55:31 python python would be pretty good for
1:55:33 you know the stuff you talk about blah
1:55:35 blah blah I was like oh yeah I heard
1:55:37 chat GPT can do python so so so while my
1:55:41 w wife's watching TV I just type into my
1:55:43 laptop make me a python script that does
1:55:46 something and it made a python I was
1:55:48 like oh that's impressive and then I
1:55:51 realized I don't know how to run
1:55:55 Python and so I said to chat gvt oh I
1:55:59 forgot I don't know how to run Python
1:56:02 and it said not a problem just open your
1:56:05 terminal and type in this command line
1:56:06 and install this library and I was like
1:56:08 oh God that's awful no I don't want to
1:56:10 do command line I said isn't there a
1:56:11 website that can I can just copy and
1:56:13 paste the code and and see if it works
1:56:15 and I said sure here's five so I picked
1:56:18 the first one which happened to be rep
1:56:20 which is the which is a really good
1:56:22 python site or whatever it is you can
1:56:24 run python
1:56:25 there
1:56:28 and I pasted it and I hit the play
1:56:30 button and it broke it didn't work it
1:56:32 said here's an error and and so I copied
1:56:35 the error and I pasted that into chat
1:56:37 GPT and it said oh well that means you
1:56:39 don't have this Library installed I said
1:56:41 okay I don't know how to how to how do
1:56:43 you install that library and it told me
1:56:46 in repet in the lower left hand corner
1:56:48 click on this button
1:56:50 and then in the upper right hand
1:56:53 corner do a search for this library and
1:56:55 then click install and then hit
1:56:58 run and so I clicked typed install run
1:57:03 and it ran within 90 minutes I had
1:57:06 written a fully functional python
1:57:08 application that took three text inputs
1:57:11 like a subject and couple of questions
1:57:13 about it and generated five pieces of
1:57:15 social media content by by sending um
1:57:20 calls to open AI
1:57:22 API in 90 minutes I've never written a
1:57:25 line of python in my
1:57:28 life in that
1:57:33 moment what became clear to me is that
1:57:36 the world is not like it was it's just
1:57:38 that no one knows it yet the world is
1:57:41 not like it
1:57:43 was If instead of spending three weeks
1:57:45 or three months or however the [ __ ] long
1:57:47 it's going to take me to learn enough
1:57:49 python to even understand the syntax of
1:57:52 how to write it that in 90 minutes I
1:57:54 could create a functional application
1:57:57 that was running on a computer that
1:57:59 actually did something for my
1:58:04 business nothing will ever be the same
1:58:07 like that I know that I know what does
1:58:10 it look like how does it impact it how
1:58:12 far does the impact go I don't that I
1:58:14 don't have a [ __ ] clue on but it's
1:58:17 crystal clear to me that the world is
1:58:19 not what it
1:58:21 was just no one knows it yet and that's
1:58:24 what I went through with the worldwide
1:58:25 web I
1:58:26 knew that communication was going to be
1:58:30 instant geography was going to
1:58:33 collapse you know um information was
1:58:36 going to be vast everywhere and
1:58:39 accessible and you could kind of play
1:58:41 out what the future looked like if if
1:58:43 you made assumptions that those were the
1:58:45 things you could kind of look at certain
1:58:47 industries like thank you very much
1:58:49 Sherry Banks
1:58:50 like the fax industry people that made
1:58:53 fax machines and you could be like yeah
1:58:56 that ain't going to be around for long
1:58:58 and then you know like Ron talking about
1:59:00 his newspaper Buddies the the fax
1:59:02 manufacturers like wow we've got the
1:59:04 what what we've done with the facts is
1:59:06 now we've included a drink holder okay
1:59:09 yeah they're going to buy it for the
1:59:10 drink holder yeah that's
1:59:12 it it's still
1:59:15 relevant we're focusing on specialty
1:59:18 applications for faxes
1:59:24 so that shit's
1:59:28 coming which Industries how fast how
1:59:31 deep I don't know but here's what I do
1:59:33 know if you're a Ron in a company you're
1:59:36 going to be a hell of a lot better off
1:59:39 than if you're one of the blo hards in
1:59:41 the
1:59:42 boardroom going we don't need this AI
1:59:48 [ __ ]
1:59:52 you know what those guys are going to do
1:59:55 he this this is going to this we're
1:59:57 going to see a lot of this they're going
1:59:59 to [ __ ]
2:00:01 retire because I'm 58 right I have I
2:00:05 have a weird amount of energy when I get
2:00:08 motivated about something
2:00:11 but a lot of existing companies are not
2:00:15 going to have
2:00:17 the desire
2:00:20 to completely reinvent themselves and so
2:00:22 I think what you're going to see is
2:00:23 you're going to see you know yeah let's
2:00:26 let Ron run the thing and they they'll
2:00:28 just [ __ ]
2:00:32 walk you know or they'll just you know
2:00:35 hold on with a death grip to the old
2:00:37 business model and just watch
2:00:39 everyone Zoom
2:00:43 by you know crazy it's crazy I tell you
2:00:47 it's crazy it's very interesting to know
2:00:50 when so many others don't know isn't it
2:00:52 Source Camp I mean I honest to God it
2:00:54 feels it this is like the Matrix in just
2:00:58 like like it like an almost palpable way
2:01:03 that the red pill is you know you play
2:01:08 with generative AI until you have your
2:01:10 Kevin McAllister moment right this
2:01:15 moment and then if you dig a little bit
2:01:18 deeper and go
2:01:20 can it do what I
2:01:22 do oh it
2:01:25 can can it do what my buddy does oh it
2:01:29 can what's going to happen here oh my
2:01:32 and then you're once you're in you can't
2:01:35 unsee it you can't unlearn it it ain't
2:01:39 going to be
2:01:41 uninvented you
2:01:46 know Sherry Banks yeah my wife gives me
2:01:49 loving looks but doesn't understand the
2:01:51 implications of
2:01:53 AI my my wife you know like like luckily
2:01:56 I've got an understanding wife like you
2:01:58 know we we we met as actors
2:02:00 so so all this business [ __ ] is like you
2:02:04 know go do what you need to do
2:02:08 but you know she she knows when I get
2:02:10 like this it's like all right we let him
2:02:12 go off
2:02:13 and
2:02:16 twirl cuz I'm going to figure something
2:02:18 out like do I know what it is yet I
2:02:21 don't I
2:02:24 don't but I am I going to am I going to
2:02:27 stop cuz I don't know [ __ ]
2:02:32 no
2:02:34 no do you dream talking to pie I
2:02:38 actually don't talk to Pie as much as I
2:02:40 like I played with it a lot early
2:02:42 on
2:02:45 um I I oscillate between what I'm
2:02:49 calling lateral play which is basically
2:02:51 like if here's the [ __ ] I'm good at
2:02:53 lateral play is like playing with the
2:02:54 [ __ ] out to the sides of that and
2:02:57 getting good at that so I'm oscillating
2:02:59 right now between lateral play and doing
2:03:01 [ __ ] for my business and lateral play
2:03:03 and doing [ __ ] for my business and so
2:03:05 I'm kind of I'm kind of going through
2:03:07 these waves
2:03:09 of when I do [ __ ] for my business I kind
2:03:12 of lose track of what's happening in AI
2:03:14 world and when I'm doing AI world I kind
2:03:16 of lose track what's going on in the
2:03:17 business so it's like like I'm it's it's
2:03:20 a bit schizophrenic right now but it's
2:03:21 kind of cool I struggle with do I share
2:03:24 this or hold it close to my vest so
2:03:26 Craig Bergstrom um I am this channel I
2:03:32 am this channel exists for me to just
2:03:35 get this [ __ ] out and tell as many
2:03:37 [ __ ] people as I can I am a firm
2:03:39 believer that when when you're this
2:03:42 early in an industry so if you if you
2:03:44 think about whatever AI is
2:03:47 today and I and when I'm I'm talking
2:03:49 about AI I'm not talking about the AI
2:03:51 that's been around for 20 years that's
2:03:53 embedded in all of our [ __ ] I'm talking
2:03:55 about the AI that starts November 30th
2:03:58 of last year the generative AI stuff
2:04:00 where the rest of us get access to these
2:04:03 tools that industry is like as as narrow
2:04:07 as a hair right now and it's going to do
2:04:10 this right it's going to expand out like
2:04:14 this because it's so narrow anyone
2:04:18 anyone
2:04:20 who does anything in the industry right
2:04:23 now is going to have massive amounts of
2:04:26 opportunity when the thing does
2:04:29 this if you keep it close to your vest
2:04:33 no one will know that you're one of the
2:04:35 people that knows what the [ __ ] is going
2:04:36 on so it's it's it's my
2:04:41 deep passionate opinion that you should
2:04:44 [ __ ] share with
2:04:47 everyone even if you're even even if
2:04:49 your company has a [ __ ] AI policy
2:04:52 there's a no chat GPT policy we block it
2:04:54 at the server
2:04:55 level in the lunchroom talk about what
2:04:59 you learned over the weekend writing a
2:05:01 story book for your kids and and now you
2:05:04 can make images in this thing too and so
2:05:06 not only did you write the story book
2:05:08 but you made the images right in line
2:05:10 with that and they'll be like what yeah
2:05:13 yeah it was amazing imagine what we
2:05:15 could do at work if if we had that tool
2:05:18 yeah we're not allowed I know oh you
2:05:20 want to know another cool thing so like
2:05:22 be that [ __ ] weirdo because what's
2:05:24 going to happen is when the when when
2:05:27 whatever this thing is starts to do
2:05:29 this people are going to start to go who
2:05:32 knows anything about this and if you're
2:05:35 one of those
2:05:37 people that'll be
2:05:40 good and what you might also find is
2:05:43 that as you learn about this stuff maybe
2:05:46 you're working in a company where you
2:05:48 keep telling people and they they're
2:05:49 like no
2:05:50 no and then you're running in another
2:05:52 circle of friends and they work for a
2:05:54 company that's really aggressively
2:05:56 adopting Ai and you're like huh that
2:06:00 company seems to be thriving mine seems
2:06:02 to be struggling maybe there's a pattern
2:06:04 here maybe jump over there right so so
2:06:08 I'm a firm believer [ __ ] tell
2:06:11 everyone start a [ __ ] Tik Tok Channel
2:06:14 I
2:06:15 did am I qualified no I am so not
2:06:18 qualified that that the supporters of
2:06:20 the channel make me signs that say what
2:06:22 are your qualifications with a bunch of
2:06:24 angry people staring
2:06:28 on because no one's qualified right now
2:06:31 that's the big secret there are no
2:06:33 experts zero zero I'm not an expert I'm
2:06:36 learning this [ __ ] on the Fly and
2:06:39 sharing what I'm learning in real time
2:06:41 and trying to get anyone who will pay
2:06:43 attention to hang out in here well I
2:06:45 don't care if you hang out in
2:06:47 here to just go play with this [ __ ] go
2:06:50 play with in fact there's a there's a
2:06:51 bunch of new people in here let me let
2:06:52 me go over and just
2:06:56 say start if you haven't played with
2:06:59 chat
2:07:00 gbt go there right
2:07:04 now don't just while I ramble you're
2:07:07 like man this this guy's this is kind of
2:07:10 interesting the geriatrics are on the
2:07:12 Tik Tok trying to teach each other how
2:07:14 to use AI yep whatever judge me troll me
2:07:17 I don't give a [ __ ] go there
2:07:20 chat. open.com and start playing if you
2:07:23 haven't played with it yet it's
2:07:25 free and then if you're like but I don't
2:07:27 know what to do it's okay go tell it to
2:07:30 write you a country song about your
2:07:33 ex the one that broke your heart or your
2:07:36 dog tell it to write you a business plan
2:07:39 for the flower business you always
2:07:41 wanted to get off the ground but haven't
2:07:43 had the guts to
2:07:44 do have it write you a project plan have
2:07:47 it make you a meal plan for the family
2:07:49 that only includes arvar meat as the
2:07:51 primary protein it
2:07:54 will and then these other ones you know
2:07:57 you can go play with them too but just
2:07:58 go spend 30 minutes with chat
2:08:02 chpt just while I'm talking and then if
2:08:07 you do it right you'll have your first
2:08:09 Kevin McAllister moment which
2:08:13 is I had no
2:08:17 idea that's an amazing [ __ ] moment
2:08:20 when you think that this thing oh it's
2:08:22 just like fancy Google it's not like
2:08:24 fancy Google it is staggeringly more
2:08:27 profound than
2:08:28 that staggeringly more profound than
2:08:31 that how do you know I don't know it's
2:08:33 it's impossible to describe it so you
2:08:36 have to go this is this is like one of
2:08:38 those did did you ever take uh I did the
2:08:41 landmark forum for a while which is one
2:08:43 of those you know self-help kind of is
2:08:46 it a cult or is it a self-help group
2:08:50 and and it's it's just like one of those
2:08:52 things
2:08:53 where people want to know what is it
2:08:55 what is it and it's hard to describe it
2:09:00 like I I figured out a way to articulate
2:09:02 it but it's really hard to describe and
2:09:06 and and so you kind of have to do it and
2:09:09 that's what this stuff is right
2:09:11 now because it looks like fancy Google
2:09:13 it's not fancy Google and now that
2:09:15 they've added DOL image generation into
2:09:17 this thing if you pay the 20 bucks for
2:09:20 that all these images that that Ted is
2:09:23 making these really cool like there's a
2:09:26 wash washer one with an AI shirt in the
2:09:29 dryer that's pretty cool um he's making
2:09:33 in in chat GPT he's just talking to chat
2:09:35 GPT asking it to make him pictures and
2:09:37 it's making him pictures it that thing
2:09:40 that thing the thing you think is fancy
2:09:42 Google is way deeper than fancy
2:09:44 Google you can go into this thing in
2:09:47 there called Advanced Data analysis
2:09:49 where you can upload a CSV file of data
2:09:52 and have it do the analysis and create
2:09:55 charts and graphs and identify Trends
2:09:57 and clean up the data and understand the
2:09:59 data if you know nothing about data
2:10:02 analysis like I
2:10:04 don't it can do it well you can do it
2:10:08 using it it's remarkable so go do that
2:10:12 stuff go play go play go play go what so
2:10:17 I'd like to inquire what are your
2:10:20 qualifications sir sir sir I find you
2:10:25 tiresome and bothersome sir good day to
2:10:32 you how we doing time wise I got 15
2:10:35 minutes till I got to start thinking
2:10:37 about wrapping this [ __ ] up you know
2:10:40 people you come in here and you just
2:10:42 like hey Kyle what's chat jpt hey Kyle
2:10:46 how do I do this hey Kyle hey Kyle Kyle
2:10:49 Kyle and then you're like three hours
2:10:51 later you're like but Kyle you can't go
2:10:53 I got more
2:10:55 questions well I got to sleep two
2:10:58 people what you think this is a cake
2:11:01 walk just cuz I walk on
2:11:04 cake oh look AI
2:11:07 Anonymous someone made that image that's
2:11:10 a great one hi my name is Kyle I've been
2:11:13 AI literate since November 30th 2022 oh
2:11:17 hey Kyle yeah yeah yeah yeah last time I
2:11:21 used AI was well I I right before I
2:11:24 started talking I was I was uh I was on
2:11:27 chat GPT yeah we all are we're no one's
2:11:30 listening to you than yeah that's great
2:11:32 that's great that's great in fact I'm
2:11:33 transcribing you and turning that into a
2:11:35 poem right now oh awesome that's
2:11:37 fantastic yeah
2:11:39 yeah yeah so I tried to stop and then
2:11:43 uh didn't work out so good if you know
2:11:46 what I mean then they added dolly three
2:11:48 to the chat GPT is now it's multimodal
2:11:52 and I just can't I'm a little moist
2:11:57 um
2:11:59 is can anyone stop can anyone stop it if
2:12:02 they're going to stop it right the
2:12:04 government's going to shut it down right
2:12:05 they're going to turn it
2:12:07 off I I heard they have a button do they
2:12:09 have a button to stop it no okay all
2:12:12 right it it what it does
2:12:16 video I you can watch a it makes video
2:12:21 oh my oh no really
2:12:23 oh I like video should I should I try
2:12:28 no thoughts on
2:12:30 autogen um autogen is pretty
2:12:36 cool so a lot of the autonomous agent
2:12:39 things Auto
2:12:44 [Music]
2:12:47 Jan
2:13:00 this is the one that's the software
2:13:01 studio right is that
2:13:10 right man that's a big
2:13:14 [Music]
2:13:17 PDF
2:13:21 it might not be that one um so I think
2:13:23 all these autonomous agent things are
2:13:25 very very very very interesting they're
2:13:28 very new um so so basically what what
2:13:31 these things
2:13:33 are
2:13:36 is you're going to have these things
2:13:38 called agents so agents are like chat
2:13:42 chpt except they can prompt
2:13:46 themselves and then they can take
2:13:48 actions
2:13:51 autonomously which is which is pretty
2:13:53 cool but it's also like a little [ __ ]
2:13:55 scary but so imagine imagine being able
2:14:02 to have an agent that understands your
2:14:05 preferences and books travel for you and
2:14:08 then you have another agent that
2:14:10 understands your preferences and
2:14:12 coordinates with your friends agents to
2:14:14 book social Gatherings and then you have
2:14:16 another agent that understands your
2:14:19 preferences that finds activities for
2:14:21 you to do and then all three of those
2:14:23 agents could talk to one another and you
2:14:25 also have a work agent that understands
2:14:27 when you're busy at work so you could
2:14:29 basically just say to your agents
2:14:32 maximize my vacation time for the next
2:14:34 two months and then they would all
2:14:36 coordinate with one another take actions
2:14:38 book [ __ ] ask you hey Sally's available
2:14:41 go to Cancun on this weekend and it
2:14:44 doesn't interfere with work and it's
2:14:45 fine in your budget do you want me to
2:14:48 ask Sally if she wants to go and you're
2:14:50 like no I don't like Sally is Bob
2:14:52 available no Bob's not available but Bob
2:14:54 could do it with Jim three days after
2:14:57 that so if you went in the middle of the
2:14:58 week the three of you could go and then
2:15:00 you go yeah that's good and then it'll
2:15:01 just go do
2:15:03 [ __ ] there's I don't know if this was
2:15:06 the paper the autog genen
2:15:08 one but I I think it might have been
2:15:11 this one or there's a there's a related
2:15:12 project where there's now a it it looks
2:15:15 like uh Sim life from like from the 9s
2:15:18 but it's all these little agents that
2:15:20 perform different functions of a
2:15:22 software company like quality control
2:15:26 and engineering and interface and ux and
2:15:29 project management and so you give it a
2:15:31 task I want a asteroids style arcade
2:15:37 game that has that that you shoot
2:15:40 cactuses with a six
2:15:42 shooter and then all of these
2:15:45 independent agents for this virtual
2:15:47 software company start collaborating and
2:15:50 they go make your
2:15:51 software like that's out there right
2:15:54 now does it work really good is it
2:15:57 perfect no is it going to make you janky
2:15:59 ass software that doesn't quite work yes
2:16:02 cuz that's the stage we're at we're
2:16:04 early we're way early we're way earlier
2:16:06 than anyone
2:16:08 knows
2:16:10 um
2:16:12 but play it out 18 months play it out 18
2:16:16 months in 18
2:16:19 months that will exist as a thing that
2:16:22 you can just say I want a video game
2:16:25 that does
2:16:26 blank I want it to be AAA I want it to
2:16:29 be this I want it to have driving
2:16:30 mechanics I want it to have gambling
2:16:32 mechanics I want it to be social I want
2:16:35 it to only be available for my friends
2:16:40 go 15 minutes later you get a little
2:16:43 note want to
2:16:45 play there's your custom game
2:16:48 or there's your custom vacation plan or
2:16:50 there's your custom book that you want
2:16:54 to publish so that you can say you're a
2:16:56 published author so you can get more
2:16:58 speaking
2:17:01 gigs so I think I think the auto gens of
2:17:05 the world and the auto gpts and the baby
2:17:08 agis and all all of the tools that that
2:17:11 do that Lang
2:17:13 chain is in the hood are all fascinating
2:17:17 Technologies I'm not digging I'm not
2:17:19 messing around with them too much right
2:17:20 now just because things are changing so
2:17:22 fast there I'm going to let them sort of
2:17:24 bubble up to the next layer of interface
2:17:26 and simplicity but if I were at all
2:17:28 geeky I'd be digging into this [ __ ]
2:17:31 heavy because it's it's absolutely the
2:17:33 next thing in fact you you want to see
2:17:35 let's let's go
2:17:37 play I can show you a little bit of what
2:17:40 autonomous
2:17:41 agents uh look
2:17:45 like we're going to go to chat JP I
2:17:48 don't know if you've heard of chat jpt
2:17:50 but you can make money with chat
2:17:52 jpt did you know that how do you think I
2:17:56 can afford a gold button
2:17:58 how because I know
2:18:01 Parker all right we're going to go to
2:18:03 Advanced Data
2:18:05 analysis and what that is is it's a mode
2:18:08 within chat GPT that can
2:18:11 do come on say it Advanced Data analysis
2:18:15 was that an AI event today in the
2:18:17 predent presenter said a tool was still
2:18:20 a little janky and I led that's
2:18:24 awesome that's awesome
2:18:28 Chris oh these tools right now they
2:18:31 janky pieces of [ __ ] so let me show you
2:18:33 a little bit of autonomous agent so what
2:18:36 what you do here is you can attach a
2:18:39 file so I'm going to go grab a zip file
2:18:42 that has a CSV in it um let me see
2:18:47 archive
2:18:55 [Music]
2:18:56 I don't know which one this one is I
2:18:58 literally don't know what's in this ZIP
2:19:00 file but here's what I'm going to
2:19:03 [Music]
2:19:04 say figure out what's in the zip file
2:19:08 and
2:19:11 identify the structure of the data and
2:19:14 what might be interesting about it based
2:19:16 on that
2:19:20 so that's not so much a prompt as kind
2:19:23 of a an imperative like like it's just
2:19:26 kind of like like hey I'd like to know
2:19:29 what's in there and so it
2:19:33 says it it did this work and it said you
2:19:35 can show the work well what was the work
2:19:37 it did well
2:19:39 it the way Advanced Data analysis Works
2:19:42 in chat GPT is it creates what's called
2:19:44 a container which is sort of a
2:19:46 self-contained little um like think of
2:19:51 it like a a mini computer that can that
2:19:53 can run and execute um code in your chat
2:19:57 session it's like a little bucket in
2:19:59 your chat session that can run code and
2:20:01 it's been given access to a bunch of
2:20:03 python libraries and so it uses the
2:20:06 large language
2:20:07 model to interpret what you asked and
2:20:11 then it gives itself a set of
2:20:12 instructions and it writes and executes
2:20:15 its own python code so it imported those
2:20:18 two libraries it wrote this code to
2:20:20 unzip the file and it unzipped the file
2:20:23 and then it says okay it contains
2:20:24 unemployment data great and it says to
2:20:27 better understand it let's look inside
2:20:29 the CSV file and so then it Imports
2:20:32 pandas library and it goes and looks at
2:20:35 the data and like I what all I typed was
2:20:39 what's going on in here look it did all
2:20:41 this it's prompting itself so so right
2:20:45 now this is kind of this self-contained
2:20:47 little container within a chat session
2:20:50 um I can I can say
2:20:53 um make a pretty chart
2:20:57 that visualizes one of the more
2:21:00 significant Trends in the
2:21:04 data and I want it to be aesthetically
2:21:09 beautiful and so it's going to go off
2:21:11 and do this so now
2:21:14 imagine this is an agent that's not only
2:21:18 sort of doing this stuff within your
2:21:19 chat session but it's going out to the
2:21:21 internet and executing stuff out there
2:21:24 it's got access to your travel sites and
2:21:28 you've given it permission to go to your
2:21:30 travel sites and it can go out and find
2:21:32 the cheapest prices across 10 different
2:21:34 sites and so it can become its own
2:21:36 Expedia and then it can interact with
2:21:40 another agent that's interacting with
2:21:41 some other part of your life and you
2:21:42 have all these different agents and
2:21:43 they're all taking action on your behalf
2:21:46 and all you're doing is you're kind of
2:21:48 like Oz behind the curtain right you're
2:21:51 just saying here's what I want
2:21:53 done go do
2:21:56 that and and the go do that part is
2:21:59 going to get more and more and more and
2:22:01 more
2:22:04 sophisticated and I mean I I'll go back
2:22:06 to Ron and his and his newspaper buddies
2:22:09 there what a newspapers look like in the
2:22:12 future am I going to go to the New York
2:22:15 Times probably not but what I'm probably
2:22:19 going to do is I'm probably going to use
2:22:21 something like Google bard who cut a
2:22:24 $100 million deal with the New York
2:22:27 Times to get their data so that I can
2:22:30 say to Bard hey tell me what's going on
2:22:32 in the news in this particular
2:22:35 area and limit your search to trusted
2:22:38 sources and so it will limit it to New
2:22:41 York Times and whatever other sources
2:22:43 Google cut deals
2:22:46 with and and so how you interact with
2:22:49 trusted news sources in the future is
2:22:51 probably going to be hyper
2:22:54 personalized um and it may not look at
2:22:57 all like the current advertising
2:23:01 model and
2:23:04 so what do those companies do but that's
2:23:07 that's kind of the the future of
2:23:08 autonomous
2:23:10 agents and I mean even you know even the
2:23:13 um even the do [ __ ] right we'll put it
2:23:16 in do mode
2:23:19 here and go
2:23:26 um make me an image
2:23:31 of an audience watching the clock and
2:23:35 the clock is about to strike 10:30
2:23:43 p.m. so Dolly 3 now it says it's
2:23:46 creating images what it's actually doing
2:23:48 is it's off writing its own prompts it's
2:23:50 writing four distinct
2:23:52 prompts it's creating one as a photo one
2:23:55 as an illustration one is a 3D rendering
2:23:58 I don't I don't I don't know what it's
2:23:59 doing it doesn't matter it it will be
2:24:02 completely irrelevant to any of us what
2:24:05 the [ __ ] it's doing now who it may be
2:24:08 relevant to are the people that
2:24:10 programmed
2:24:16 it but like that's an autonomous agent
2:24:19 right that
2:24:21 thing if I wanted to
2:24:24 find something like that in stock photos
2:24:28 I probably couldn't find it so I'd
2:24:29 probably have to go find an image of a
2:24:31 clock an illustration of an audience
2:24:35 hopefully it's stylistically similar pop
2:24:37 them into Photoshop layer them trim them
2:24:39 cut them
2:24:42 or
2:24:43 perfect that's exactly what I needed
2:24:46 that's good enough back to that I was
2:24:47 talking earlier about triangles of the
2:24:49 triangle of content where at the very
2:24:51 tippy top you need handbuilt bespoke
2:24:53 content but most of the content that you
2:24:56 need is the bottom half of the pyramid
2:24:58 or the bottom 2third of the pyramid it
2:25:00 it just needs to be good
2:25:02 enough so this is good enough and we're
2:25:05 going to have these agents that are
2:25:06 going to go out and they're going to
2:25:07 book travel plans that are good enough
2:25:08 they're not going to be
2:25:10 perfect and then over time they'll get
2:25:12 more and more perfect and they'll just
2:25:13 become what we expect but you know
2:25:18 so there are little glimpses of
2:25:19 autonomous agents in the tools we have
2:25:21 access to today
2:25:24 but hi my name is Kyle I've been into AI
2:25:28 for like two years hi Kyle
2:25:31 hey so you know when I started I was
2:25:35 just making pictures and I thought they
2:25:36 were cute you know and then I got into
2:25:39 dream booth and dream Booth was like oh
2:25:42 I could upload pictures of myself and
2:25:44 I'm like I'm kind of chubby I don't know
2:25:46 if you knew that it's you probably can't
2:25:47 see it but I like I hide it well so
2:25:50 anyway I would like make pictures of
2:25:53 myself and then I put them in dream
2:25:54 booth and sometimes it made me look like
2:25:56 a superhero and I was like oh I'm pretty
2:25:58 goodlooking and I kind of liked it it
2:26:00 helped my self-esteem and then other
2:26:02 times it would just make me look like a
2:26:04 400 lb tubby and I was like oh my
2:26:08 God and then Chachi came out oh yeah
2:26:13 yeah we know that one yeah yeah so and
2:26:15 then I was I would use TBT and it would
2:26:18 just make all these words and I was like
2:26:20 I I make
2:26:22 words so and I cried a lot and then I
2:26:26 thought oh it's very it's super exciting
2:26:28 and then I got sad again and then I got
2:26:32 excited again and I got really sad again
2:26:34 and I got excited again and that's where
2:26:37 I am now is I'm sad and excited and uh I
2:26:41 cry a lot and um I I'm oh I'm I'm a
2:26:47 little
2:26:49 fump
2:26:50 yeah I love this image I know it's so
2:26:53 good isn't
2:26:56 it I think we'll keep it like this for a
2:26:58 while let people actually this pretty
2:27:00 good people are scroll scrolling through
2:27:02 Tik Tok rather than seeing Rosie odonald
2:27:05 they're like AI Anonymous what the [ __ ]
2:27:07 is
2:27:10 that that's good it's such a good
2:27:14 image turtles and llamas
2:27:19 oh my goodness people you all make me
2:27:22 laugh you all make me laugh it's so good
2:27:25 it's so
2:27:27 good all right let me get to a few more
2:27:30 questions I'm going to shut this thing
2:27:32 down you know we got to get out and
2:27:34 smoke our cigarettes after our little
2:27:36 cup of coffee in the
2:27:39 meeting I can't believe Larry forgot the
2:27:42 donuts again it's like come on Larry you
2:27:45 got one job well
2:27:47 two don't use Ai and bring the
2:27:54 donuts autogen available for regular
2:27:58 people
2:28:00 um I mean what's what's interesting
2:28:02 about the Irregulars in here is that
2:28:04 some of them are pretty Technical and
2:28:05 geeky like herod's pretty technical
2:28:08 pate's super technical there's there's a
2:28:10 couple of people in here that have are
2:28:11 good on the on the uh on the technology
2:28:16 side I would call myself technology
2:28:19 literate but you
2:28:22 know I know just enough to piss off the
2:28:26 engineers cuz they can't get one over on
2:28:28 me but I have unreasonable expectations
2:28:31 about what they can uh they can do
2:28:34 Microsoft autogen is there interesting I
2:28:36 haven't played with it yet
2:28:39 so uh uh uh check on Discord please Rudy
2:28:44 I think I saw yours Rudy didn't I Rudy
2:28:46 have new one
2:28:48 there oh Rudy oh oh AI learning lab okay
2:28:51 this is one of those this now I can see
2:28:53 it oh wait hang
2:28:55 on so Rudy did
2:28:58 so squint see how it says AI learning
2:29:01 lab but if you go up close to the image
2:29:03 it's just like people
2:29:05 with stuff isn't that
2:29:10 trippy Rudy what's that what's that
2:29:13 tool I forget what what the tool is cuz
2:29:16 I went and played with it and then I
2:29:17 couldn't remember what it
2:29:20 was
2:29:22 um what's the tool again again
2:29:29 again God the the uh keyboard on my iPad
2:29:33 is it's just too small so I mistype
2:29:37 because I'm not used to that letter
2:29:39 spacing Microsoft autogen is the teams
2:29:43 of AI agents working together like
2:29:45 employees of a company yeah that's super
2:29:46 cool
2:29:52 cool those images are
2:29:56 trippy oh that's control net but someone
2:29:59 made a didn't someone make a
2:30:02 tool didn't someone make a I guess they
2:30:05 made a standalone app so yeah so you're
2:30:07 doing those things in control net that's
2:30:09 pretty
2:30:12 slick AI Anonymous how can I help you
2:30:16 yeah ah listen I think I might have an
2:30:19 issue with AI uh what's that I'm not
2:30:22 sleeping anymore I wake up thinking
2:30:24 about AI I go to bed thinking about AI I
2:30:27 watch this older fella on Tik Tok every
2:30:30 single night for like hours oh yeah yeah
2:30:33 we've had some of your type around here
2:30:36 welcome welcome yeah yeah is it the it's
2:30:40 the AI learning lab yeah how did you
2:30:42 know yeah we're all there yeah yeah yeah
2:30:47 so this isn't so much to improve your
2:30:51 situation as to as to accept
2:30:53 it so welcome welcome so you all are
2:30:57 doing this too you're not sleeping
2:30:59 nobody Sleeps No one's sleeping no no no
2:31:03 we don't well when he talks we fall
2:31:05 asleep to that that's it's the only time
2:31:07 we can really sleep is when is when
2:31:10 Kyle's talking yeah he he's got he's got
2:31:13 a way about him he
2:31:15 seems he seems normal and then uh and
2:31:19 then he just starts talking and you know
2:31:21 we're right out we're just right out and
2:31:23 then we'll occasionally wake up when he
2:31:24 yells about something the government or
2:31:26 this or that you know he's he's cute
2:31:28 that way where he's got the little comp
2:31:30 conspiracy theories he just thinks
2:31:32 everyone's out to get him or something
2:31:33 you know I don't we think he's normal
2:31:35 it's we think he's normal but yeah yeah
2:31:37 no we we we think he's nice we think
2:31:40 he's nice and it's um but we can't
2:31:43 control it and we've just come to accept
2:31:45 it and so yeah yes okay well and that
2:31:48 special and that
2:31:51 special in that
2:31:55 special all right and
2:31:57 [Laughter]
2:32:00 scene I don't listen I got no control
2:32:03 over this I hope you know that right
2:32:05 this is just this is what this is I came
2:32:09 here to learn and he was just making
2:32:10 jokes it was it was like he was like on
2:32:12 a Saturday Night Live skit but it was
2:32:14 only Friday night it's like okay well
2:32:16 could could you at least wait till
2:32:17 Saturday to do your stupid
2:32:21 jokes I
2:32:26 could M Gunna [ __ ] weirdos I mean
2:32:34 Irregulars can we make money what should
2:32:37 I do now I'm poor I need money you can
2:32:39 make money
2:32:42 with so listen felicific what's your
2:32:45 name felicific felic phic philic
2:32:50 philic your name seems like it needs one
2:32:53 less or one more
2:32:54 syllable
2:32:59 phif thank you
2:33:01 Winston I like llamas llamas is tasty
2:33:06 um here's the thing about making money
2:33:08 with chat GPT and you're not going to
2:33:10 like this answer fil
2:33:13 philic Phil philis you're not going to
2:33:16 like this you
2:33:18 might but you're
2:33:22 not generative AI tools are tools
2:33:27 they're not a strategy they're not a
2:33:29 get-rich quick
2:33:31 scheme will I show Rudy's image again
2:33:34 who's
2:33:36 Rudy I will show it oh Rudy's image wait
2:33:40 hang
2:33:41 on yes hold please the AI learning Lab
2:33:45 One
2:33:48 screenshot
2:33:49 away you could also join the
2:33:54 um let me try to get it so it's not
2:33:57 quite so
2:33:59 reflect you could also join the AI salon
2:34:03 so if you go to the
2:34:08 salon. there you go
2:34:11 um yeah that's a really cool image if
2:34:13 you go to the
2:34:15 salon. it's a link tree and and the
2:34:17 third link in that link tree is the link
2:34:20 to the to the Discord the AI Salon
2:34:22 Discord so just join the Discord and you
2:34:23 can go get the image
2:34:25 wholesale can you address the naali
2:34:29 question what was the naali
2:34:31 question was that the one about
2:34:35 um what was I just talking about how to
2:34:38 make money what was the noall
2:34:40 question or no wait no
2:34:43 laoi I've said it before Kyle is three
2:34:45 moves away from
2:34:48 turning this into an AI plus Turtle
2:34:51 themed
2:34:52 cult two moves away two moves away Pate
2:34:55 this is not this is this is we are
2:34:58 quickly we are quickly descending into T
2:35:00 cult
2:35:07 territory I don't know what it is about
2:35:09 Kylie's just
2:35:15 magnetic oh I'll do anything he
2:35:18 says don't do that don't do do not do
2:35:25 not do not take advice from an outof
2:35:27 workor actor okay okay people
2:35:39 okay Nilla I can't get a graph response
2:35:43 like you
2:35:44 do you mean in in uh code
2:35:49 interpreter hang
2:35:56 on
2:36:01 so why is that not
2:36:06 there you mean you mean like this kind
2:36:08 of thing so if
2:36:10 you're so here's what you need to do
2:36:15 is you need need to be in data analysis
2:36:19 mode and then you need to upload a CSV
2:36:22 that's got enough data to plot in it I
2:36:25 think so what I
2:36:28 do NWI is sorry you mentally I've got
2:36:33 the W before the L so now I have to
2:36:35 think about how to say your
2:36:38 name I
2:36:40 can't yeah whatever the name is um so I
2:36:45 go to this site
2:36:46 to get data called kaggle K A
2:36:50 ggle
2:36:53 e.com and then you you get data sets and
2:36:58 so you
2:37:00 can and the only reason I'm doing this
2:37:02 is like I'm not a data scientist like I
2:37:04 don't know where yeah thank you mods
2:37:06 yeah this has been really great thank
2:37:07 you appreciate that
2:37:10 um so you go to data sets and then you
2:37:12 just search for stuff like most stream
2:37:15 songs in Spotify how big is this 48k
2:37:19 this should work so if I go in here and
2:37:22 I download
2:37:25 this archive
2:37:28 14 and I go back over to chat JPay
2:37:33 T make sure that I'm in data analysis
2:37:36 mode and then you get your little plus
2:37:38 button and upload the
2:37:42 thing there's 14 and say
2:37:47 this ZIP file contains data for
2:37:51 streaming performance on Spotify can you
2:37:55 please analyze the
2:37:57 CSV and tell me what's interesting about
2:37:59 the data I mean part of it might
2:38:03 be if you're just telling it to plot
2:38:06 data that's in there you might need to
2:38:08 do a step before that and have it
2:38:10 identify Trends or somehow
2:38:13 understand the
2:38:16 the data a little bit more um before you
2:38:20 ask it to plot something I don't know I
2:38:22 don't quite know what you're trying to
2:38:26 solve or were you talking about um the
2:38:31 do stuff those
2:38:33 images because if you don't have DOI
2:38:35 that's another thing is hang
2:38:39 on I might be making an assumption cuno
2:38:42 I think you might be right he might be
2:38:43 asking about Dolly okay so let me let me
2:38:45 show you one other thing
2:38:47 here so when you go to chat
2:38:50 gp4 I've got five
2:38:54 modes of chat GPT you may have
2:38:59 less um so I've got default browse with
2:39:03 Bing Advanced Data analysis this is the
2:39:05 one that will write python code and make
2:39:08 those kind of charts then I got plugins
2:39:10 some plugins can make images but it's
2:39:12 very hit or miss and then there's do and
2:39:15 this is the thing that's making the
2:39:17 really pretty images so if you don't
2:39:19 have DOI it won't make images but this
2:39:23 is coming to anyone who's a chat GPT
2:39:25 plus member all of these should be
2:39:27 coming like in the next week or two
2:39:29 here's the deal open AI has their
2:39:32 developer conference on November 6th so
2:39:35 we
2:39:36 are October it's Friday the 13th people
2:39:40 it's
2:39:42 scary um we're 3 weeks out from the open
2:39:46 AI developer conference so what what it
2:39:49 what it appears to me is that they're
2:39:51 getting all of these new features out
2:39:53 there cuz they're probably going to
2:39:54 announce some additional [ __ ] on top of
2:39:58 this my gut is telling me they're
2:40:01 releasing all this stuff as a preemptive
2:40:04 move against Google's
2:40:06 Gemini which is also supposed to be
2:40:08 multimodal and better than gp4 so I
2:40:12 think they're sort of pushing this stuff
2:40:13 out a bit early and then they're
2:40:15 probably going toce some really cool
2:40:17 [ __ ] at their developer
2:40:18 conference um is my guess so I think
2:40:22 within the next 3 weeks we should all
2:40:23 have access to this which AI is a good
2:40:25 source of legal advice relevant to the
2:40:27 law so zindel there's there's a site
2:40:31 called do not pay.com I think it's doc
2:40:34 let me just confirm that do not pay yeah
2:40:38 doc so so this was started by a
2:40:42 19-year-old like seven years ago and
2:40:45 it's basically a virtual lawyer um I
2:40:47 wouldn't say this is a good source for
2:40:49 legal advice but it's got like really
2:40:52 specific things like if you want to
2:40:53 fight a traffic ticket it'll fight that
2:40:56 for you
2:40:59 um
2:41:02 there's I would go to um perplexity
2:41:07 doai does searches with sourcing in it
2:41:13 and then also let's go look this might
2:41:15 be a place to go look for um plugins
2:41:19 hang
2:41:21 on so we're going to go back to chat
2:41:26 GPT and we're going to go to
2:41:29 plugins and then now in plugins so the
2:41:32 way plugins work is plugins are very hit
2:41:34 or miss so I don't I don't talk about
2:41:36 them a ton because they they're just
2:41:38 they're so janky they're just they're
2:41:41 just [ __ ] janky okay so the way they
2:41:44 work is you got a you got a list of
2:41:45 plugins in in any given chat session you
2:41:48 can enable up to three of them but you
2:41:51 can install as many as you want but
2:41:52 you're only activating you know up to
2:41:55 three in a chat session if you go to the
2:41:57 plug-in
2:41:58 store I'm going to search for legal L G
2:42:05 L and let's see what it comes up with
2:42:08 how many pages are there there are are
2:42:11 there only four of
2:42:14 these answers any inquiries related to
2:42:17 Chinese
2:42:19 law fiscal note enables access to select
2:42:24 Market Market leading real-time data
2:42:27 sets for legal political and regulatory
2:42:29 information so fiscal note is a is a
2:42:32 kind of law mid page case law interact
2:42:35 with us case law searches summarize
2:42:37 answers so let's do mid page case law
2:42:40 and fiscal
2:42:42 note oh I've got to have an account
2:42:44 there some of these things you have to
2:42:45 have an account on the sites that's
2:42:47 that's why these things are Hit or Miss
2:42:48 is cuz just sometimes they're just
2:42:52 crap um [ __ ] let me go back there dang
2:42:56 it plugin
2:42:58 [Music]
2:42:59 store
2:43:02 legal fiscal note okay that one
2:43:05 installed and then let me look up
2:43:09 law California law lawyer search Lawyer
2:43:14 PR there's two page pages of this US
2:43:19 federal law get
2:43:21 access to federal law all right let's
2:43:26 do interact with oh that was the one you
2:43:29 had to sign up mid page case law let's
2:43:31 do US federal law okay that's
2:43:34 installed okay so we're going to close
2:43:37 this
2:43:38 window so now we've got that fiscal
2:43:41 thing fiscal note and US federal law are
2:43:44 plugins that are enabled for this chat
2:43:47 session and then I'm going to
2:43:50 say
2:43:54 um I'm trying to find law pertaining
2:44:01 to limits in how many llamas I can raise
2:44:06 within the city
2:44:08 limits of a mediumsized city in the
2:44:11 United
2:44:14 States I need to see what has been
2:44:17 publicly released about Gemini ooh has
2:44:19 there has has stuff been publicly
2:44:22 released the only thing I've heard Pate
2:44:23 is that it's coming and that and that
2:44:26 it's going to be I heard both five and
2:44:28 20 times better than chat GPT I'm I
2:44:32 would be happy with as good as okay so
2:44:35 now I asked this so so there it's it's
2:44:37 using the US federal law
2:44:39 plugin as a lodg language model I'm not
2:44:42 a lawyer okay fine but did you find me
2:44:44 anything
2:44:46 hi this is Madison wait this is Mason
2:44:49 the developer I'm trying to improve the
2:44:51 plug-in and expand the types of
2:44:53 questions it can answer wait what to
2:44:56 find a find specific regulations that
2:44:58 apply to situation you need to check
2:45:00 with the municipal codes in the
2:45:02 city
2:45:04 huh can you find me a
2:45:10 specific case law that pertains to
2:45:15 keeping animals in
2:45:19 captivity damn
2:45:21 it I mispronounced captivity and it
2:45:23 wrote ctivity
2:45:26 he
2:45:43 captivity this is a useless plugin I
2:45:46 told you that sometimes the plugins are
2:45:48 useless oh wait for instance under Title
2:45:50 18 chapter 3 something 42 I guess
2:45:53 paragraph 42 or section 42 I don't know
2:45:56 anything about law I'm [ __ ] useless
2:46:00 doing
2:46:01 this uh uh uh uh another example title
2:46:06 16 chapter 1 section 404 c-3 prohibits
2:46:10 hunting or the killing of any wild
2:46:12 animal or bird within the limits of city
2:46:14 parks wait what
2:46:16 you can't kill an animals in city parks
2:46:18 that makes no sense to
2:46:19 me it's a joke people it's a joke it's
2:46:22 comedy it's what we call comedy it's
2:46:24 like comedy without the
2:46:26 laughs um these laws indicate Federal
2:46:29 interest preserving
2:46:31 Wildlife all right so it sort of did
2:46:34 that now it didn't hyperlink them so the
2:46:37 other thing that I would do is I would
2:46:39 go to there's an AI for that and I would
2:46:42 go to futurepedia doio here let me show
2:46:45 you you
2:46:47 those these top two here I'd go to those
2:46:51 two and those are full of AI tools and
2:46:54 just look up legal and law there might
2:46:56 be specific applications people have
2:47:01 made so he had a CSV file okay I I I I
2:47:06 don't know I don't know why his thing
2:47:08 wasn't sometimes
2:47:11 um uh what you call it code interpreter
2:47:15 or the Advanced Data analysis sometimes
2:47:16 it just sucks sometimes it just doesn't
2:47:19 do stuff good you just got to keep
2:47:20 trying again so let me try this again
2:47:24 um can you
2:47:28 make a circle
2:47:34 chart no I don't want a circle
2:47:39 chart what is the kind of chart I could
2:47:42 make that shows an XY AIS and has
2:47:46 Circles of different sizes where the
2:47:49 size of the circle represents one thing
2:47:52 and its position on the XY AIS
2:47:54 represents another thing that would be
2:47:56 relevant for this
2:47:59 data so I'm going to ask it to figure
2:48:01 out for me like like I know what I want
2:48:05 the pretty picture to be I'm going to
2:48:06 see if it can figure out here we
2:48:09 go x-axis danceability of the
2:48:12 track
2:48:14 wait
2:48:16 y AIS energy of the track bubble size
2:48:18 number of streams or release year number
2:48:22 of times it's appeared let's go with the
2:48:23 first
2:48:25 one um I like the first idea and can you
2:48:29 make the
2:48:31 circles beautiful
2:48:33 colors where the color of the circle
2:48:37 represents a fourth piece of
2:48:44 data
2:48:47 absolutely we can add a color Dimension
2:48:49 to the
2:48:50 bubble what's that
2:48:53 say the hell does that
2:48:56 say oh that's weird it puts some Json in
2:48:59 in the in the response here that's weird
2:49:02 color of the bubble is the mode of the
2:49:03 track oh the color of the bubble is
2:49:05 minor or major key that's kind of
2:49:10 cool it appears there was an issue with
2:49:12 the way the color palette was defined
2:49:16 so it's going to redo
2:49:20 it it seems we're still encountering an
2:49:22 issue with the color palette so this is
2:49:24 another example of it being an
2:49:26 autonomous agent where it's trying to it
2:49:28 it's making mistakes and trying to fix
2:49:30 its own
2:49:33 mistakes it appears there's a MIT
2:49:35 mismatch between the unique categories
2:49:37 of mode to resolve this we can use a
2:49:41 larger pallet or let Seaborn AC
2:49:44 automatically handle the coloring based
2:49:45 on the unique values in the mode
2:49:53 column so what's it doing
2:49:55 here it's writing code it's doing it's
2:49:59 doing what it's
2:50:03 doing did it did It win did we win or
2:50:06 are we
2:50:06 losing I think we're losing whatever
2:50:09 Seaborn
2:50:10 is relational pi. Pi 65 points ax
2:50:16 scatter so it looks like it's trying to
2:50:18 do
2:50:27 something boom not exactly a circle
2:50:30 chart but it's the X's
2:50:33 are I mean it did
2:50:36 something the size of the X's
2:50:40 right there's energy danceability is
2:50:43 there any statistic iCal
2:50:46 clustering sort of I
2:50:50 guess high energy and High dance ability
2:50:54 yield high
2:50:56 plays there you go we we didn't need a
2:50:59 [ __ ] chart for that but we got one
2:51:01 and it didn't exactly make circles on
2:51:03 the chart it made X's but they're the
2:51:05 X's are of different sizes and they look
2:51:07 like one's blue and one's green so maybe
2:51:09 they're different colors but anyway it
2:51:10 did that so I don't quite know why
2:51:12 you're not getting it to do stuff but if
2:51:14 you see how many times this failed like
2:51:17 it had an issue then it tried again then
2:51:19 it had an issue then it tried again then
2:51:20 it had an issue then it tried again then
2:51:22 it finally did something
2:51:25 so so that's what's happening all right
2:51:28 people listen
2:51:30 listen let's do this we got a few people
2:51:33 in here feel free to take screenshots
2:51:35 these
2:51:36 are things that I think are worth your
2:51:38 time to play with here's the top four of
2:51:40 the large language models here's the
2:51:43 Kids Table these are the next three down
2:51:45 that I think are worth your time
2:51:47 screenshot
2:51:50 away here are image
2:51:52 tools that I think are worth your
2:51:55 time take a screenshot
2:51:58 there then we got video
2:52:03 tools then we got people to follow and
2:52:07 some
2:52:10 directories then we got the AI
2:52:13 salon
2:52:15 that's the uh Community where you can
2:52:18 join this is just a cool list of tools
2:52:20 that tonus put together that's my
2:52:23 company and that is uh the Youtube
2:52:27 channel where we put archives of these
2:52:28 lives cuz if you just can't get enough
2:52:31 of me there's
2:52:33 like there's like 130 videos up here
2:52:35 I've been doing this for four and a half
2:52:37 months every [ __ ] night people every
2:52:40 night so listen it's getting late I got
2:52:42 to go it's probably 11: yeah it's 11: I
2:52:46 got to get out of here people if
2:52:47 anyone's on the East Coast still
2:52:49 watching this what are you doing with
2:52:50 your lives get it together get yourself
2:52:52 to AI Anonymous
2:53:01 Pronto let's let's end it let's end it
2:53:04 on this lovely
2:53:11 image oh what's my day job I am the CEO
2:53:14 thank you Winston for the Llama llamas
2:53:17 is tasty uh I am the founder and CEO of
2:53:20 storyvine is my day
2:53:23 job and I'm speaking if anyone's here in
2:53:27 Boulder I'm speaking at a at a
2:53:29 conference tomorrow called AI Peaks
2:53:33 which is Colorado's first AI
2:53:35 Summit and I am talking about how 20123
2:53:39 is the new
2:53:40 1995 and what's coming with AI and
2:53:43 what's here and what you should be doing
2:53:46 and what you should not be doing is
2:53:48 sitting on the sidelines get yourself in
2:53:50 a group like this and say I'm Kyle and I
2:53:52 just can't stop thinking about AI what
2:53:54 do I do with my
2:53:56 [Laughter]
2:54:02 life it's crazy I tell you it's crazy I
2:54:05 tell
2:54:09 you oh okay it looks like it it looks
2:54:13 like um the link on
2:54:18 the whatever you call it the link tree
2:54:21 for the salon the the Discord link might
2:54:24 be expired I'll check that out I'll fix
2:54:26 that um I wonder if there's a way I can
2:54:29 fix it now if I can get anyone there now
2:54:32 no you know what if if you're interested
2:54:33 in the salon and you couldn't get into
2:54:35 the Discord just DM me um or join the
2:54:38 Meetup and and you know we I'll I'll
2:54:41 update the uh I'll up the Discord I'll
2:54:45 update the Discord link there record
2:54:47 your presentation and please share oh
2:54:50 the one tomorrow for the uh for the AI
2:54:52 Peaks I hope they're recording it it it
2:54:54 seems like they've got their [ __ ]
2:54:55 decently together so um but I've got an
2:54:59 example of it I'll share I'll share an
2:55:02 example of it on the salon I've uh I I
2:55:04 did a a version of this like six months
2:55:07 ago for um creative mornings and so it's
2:55:10 kind of a a modified version of that
2:55:12 presentation so I'll share that
2:55:15 all right everybody listen listen it's
2:55:17 late here I'm going to go now I'm going
2:55:19 to go so thank you all let me see here
2:55:21 hair growth geek was good to see you P
2:55:24 Pate Pate P Pate good to see you hair
2:55:28 growth geek let's see L let's talk
2:55:31 talent that photo is hilarious to me I
2:55:33 know it's great isn't it um cuno
2:55:35 Amelio's wife Ted thank you for all you
2:55:37 do Sherry Banks thank you so much you've
2:55:39 been popping those gifts up all night I
2:55:42 really do appreciate it cam cat came
2:55:43 from Cleveland
2:55:45 fantastic seeing you again who else we
2:55:48 got who else we got in here forever
2:55:50 hooked um cuno as always thank you you
2:55:54 always contribute really cool things
2:55:56 Robert Rossy thank you peace Share Bear
2:55:59 forever hooked Milo's wife of course
2:56:02 late as always but but here late but
2:56:06 here all right
2:56:08 everyone peace out it's Friday night you
2:56:12 can go on your date night now it's too
2:56:14 late for that
2:56:16 exactly but at least I gave you the
2:56:18 option peace out
2:56:21 everybody