aiLL 5/12/23 - Breakthrough AI Tools Anyone Can Use

Description

Chapters

Transcript

0:06 [Music]
0:08 boom boom boom
0:10 boom boom boom
0:12 hello hello Jan Mills Bishop you're the
0:15 first one how are you
0:17 um this is Kyle Shannon this is the AI
0:20 learning lab hope you're doing well
0:22 I have replaced my
0:24 um Three Mile Island
0:26 kind of horrible drink with uh bourbon
0:31 I think this is the first time I've
0:32 drunk on the old live I've just been
0:34 sober
0:36 imagine me loopy huh
0:40 um if you are uh
0:43 here welcome
0:45 um this is the AI learning lab I talk
0:47 about AI stuff I'm passionate about AI
0:49 I'll answer questions about it hey I
0:52 want to talk to you about your Story
0:53 vine software drop me an email at Kyle
0:57 at storyvine v i n e v as in Victor so
1:00 like the growing vine
1:03 storyvine.com is the site and my email
1:05 address is Kyle at storybine so drop me
1:08 an email Nate and I'm happy to talk to
1:10 you about it
1:12 uh uh uh uh how was dinner dinner was
1:15 good we had Barbacoa tacos Barbacoa beef
1:17 can't go wrong with that I finished up
1:19 my Three Mile Island thanks for coming
1:21 back by the way don't know why you'd do
1:23 that but cheers this is a Yellowstone
1:26 select
1:27 I bought this bourbon because I like the
1:29 label
1:30 which is the way you should buy all of
1:32 your
1:33 expensive wine and alcoholic beverages
1:37 is uh label Aesthetics
1:39 it's all about marketing
1:43 [Music]
1:45 more likes can attract viewers yeah if
1:48 you wouldn't mind uh share this live
1:49 with your folks
1:51 if you think it's a value I mean I know
1:54 we haven't started yet so it's not
1:55 there's nothing valuable here right now
1:57 if you're new to AI feel free to
2:00 subscribe the AI learning lab I've got
2:02 lots of videos on AI trying to keep up
2:04 with it I do tutorials things like that
2:07 and if you haven't tried Chachi BT yet
2:11 and you're if you've even heard of AI or
2:15 you've heard that in the alphabet there
2:17 are two letters a and I and they're
2:19 vowels go to this site chat.openai.com
2:24 is the the official website for chat GPT
2:27 and while I ramble on and get drunk just
2:29 start playing with chat GPT because
2:32 playing with chat gbt right now is
2:34 probably one of the more important
2:36 things you will do in your life
2:38 huh that's provocative why does he say
2:41 that well does he say that
2:45 all right
2:47 let's see
2:50 I sent you an Instagram okay cool I'll
2:53 check that out thanks Nate appreciate
2:54 that yeah it was tasty he really
2:56 appreciate that you're doing these lives
2:58 thank you I appreciate that I I'm trying
3:00 to make these fun and uh
3:03 and uh
3:05 you know not suck just be realistic and
3:08 you know if if people ask my background
3:11 I mean I have a degree in acting but I'm
3:12 an entrepreneur I've been I've started
3:14 tons and tons of companies I've been
3:16 I've been in business for a long long
3:17 time
3:18 um I think we're at a
3:20 significantly profound inflection point
3:23 in the history of humanity like I think
3:26 like I'm not well that is hyperbolic but
3:29 I don't think it's it's hyperbole I
3:30 think
3:31 I think we're at the beginning of what
3:34 I'm calling the great Renaissance
3:36 um
3:36 and and these generative AI tools are
3:39 going to transform nearly everything
3:41 about nearly everything we do very very
3:44 quickly
3:45 um honestly almost hard hard to find a
3:48 Bad Bourbon over 40 bucks haven't tried
3:49 Yellowstone yet it's pretty good I'm a
3:51 blantons man so Blanton's is my all-time
3:54 favorite
3:56 and I I'm letting this evaporate like
3:58 this is now gone it's all evaporated but
4:01 blantham's is the [ __ ] and uh you can't
4:05 get it anymore you know that used to be
4:07 40 bucks a bottle then it was 60 now
4:09 it's 120 and now it's just you can't get
4:12 it anywhere so it pisses me off
4:16 um I can't copy and paste from
4:18 Shakespeare please fix uh you mean you
4:21 mean thepo.com Shakespeare thing
4:24 is that what you're talking about
4:25 because that I can't
4:27 um
4:28 I can't do anything about it that's
4:30 that's poe.com you got to have them fix
4:33 their [ __ ]
4:35 um they're going to show you how dumb
4:38 humanity is unplug the computer what AI
4:41 is going to show us how dumb humanity is
4:43 like we haven't showed ourselves how
4:45 dumb we are over the past decade
4:48 please we're way dumber than the AI at
4:51 least the AI gives nuanced answers and
4:53 is intelligent and is well researched
4:56 and well understood and can apply logic
4:59 to the situation I'll go with AI every
5:01 time if if you're comparing it to the
5:03 [ __ ] we got going on in this country
5:05 right now that's America by the way if
5:08 you were curious about which country was
5:10 really [ __ ] up it's this one I'm in
5:13 Denver by the way cheers
5:18 um
5:19 as I'm getting more comfortable on this
5:21 live I'm getting spicier and
5:23 saltier and more cynical but hopefully
5:26 funnier
5:27 it's all about the comedy comedy
5:31 Stephen King The Shining at Shakespeare
5:34 with rhyming is funny yeah exactly yeah
5:37 we can do that we can play around how
5:39 far do you think we are from the
5:40 singularity Ray Kurzweil says 2029
5:44 um major AI scientists have said 2030
5:48 they've just downgraded their prediction
5:50 from like 2050 to 2030 so
5:53 it appears that 2029-ish is going to be
5:56 it is the consensus right now
6:00 um someone on the live I did just like
6:02 an hour ago
6:03 um said they think it's going to be
6:04 within two years I think it's probably
6:07 going to be
6:10 between
6:11 two and four years so probably 2027 is
6:14 my guess
6:15 that's the Gen X in you salty bitter you
6:18 betcha yeah Gen X what are you gonna do
6:21 so politically correct I know right it's
6:23 not awful
6:24 I agree with you I don't know about what
6:26 but thank you it's good to have agreeers
6:28 ah no
6:33 [Laughter]
6:35 put stupid hats on me and I'll dance
6:37 like a monkey
6:40 Jesus Christ
6:44 I told you I wouldn't twerk but you know
6:47 I'm drinking so you never know what's
6:49 gonna happen hey everybody if you're new
6:51 here this is the AI learning lab uh even
6:53 though it seems like I'm just an idiot
6:55 uh I do do talk a lot about AI I'm
6:58 passionate about it
6:59 I think it's transformative technology I
7:02 think it's super important for everyone
7:05 possible to get curious about it start
7:08 exploring it
7:09 and so I'm here if you've got questions
7:11 about AI pop those in the comments um
7:13 subscribe to the channel
7:15 um if you want to see when I go live
7:16 again not sure why you do that
7:19 um
7:21 Kevin Vasquez says he wants to come live
7:24 I don't know how to do this I'm gonna
7:26 I'm gonna let Kevin up here on stage I
7:28 think this means he can he and I can
7:30 talk and he can ask questions so let's
7:32 see if that's the case
7:33 continue to go live together dual camera
7:36 will be turned off I don't want
7:37 someone's camera I just want to be able
7:39 to hear him does anyone how to know how
7:40 to do that
7:42 I don't know what's going on all right
7:47 I think you're helping a lot of people
7:49 prep and roll with this next tech
7:51 apocalypse yeah well thank you Jay I um
7:56 I'm doing what I can
7:59 um is it a tech apocalypse it's a tech
8:02 something
8:03 it's going to be profound
8:05 um techpocalypse kind of implies that
8:06 it's all going to be bad I don't think
8:08 it's all going to be bad but I think
8:09 we're going to go through a bad couple
8:11 of years I think we're going to go
8:12 through
8:13 at least two years of really painful
8:16 disruption is videovine an app or a web
8:19 service it's storyvine is the name of my
8:21 company and it's a it's a subscription
8:23 it's a it's a technology platform that
8:27 we sell as a subscription it's mostly
8:29 B2B we our primary customers in the
8:32 healthcare and Pharma space so you know
8:35 think big Pharma companies wanting to
8:36 tell patient stories or doctor videos or
8:38 have their sales rep do personalized
8:40 video Outreach that's what they use our
8:42 platform for so storyvine.com
8:45 is the platform if you want to play with
8:47 it you can just go play with it you
8:49 download storyline from the app store or
8:51 from Google Play and we've got free
8:53 templates in there you can play with so
8:55 you can see what the experience is like
8:56 it's like a it's like Story vine is is
8:59 kind of like karaoke for storytelling
9:01 it's like put your head in the dotted
9:02 outline and answer a series of questions
9:04 and then those go up to the cloud and
9:07 five minutes later you've got a fully
9:08 edited video so that's what the tech is
9:11 mm-hmm
9:14 and it doesn't use AI to do all of that
9:17 it uses Aristotle
9:20 how far do I think we are from
9:22 Singularity I just answered that
9:24 um
9:26 let's see where can I download chatgpt
9:28 you don't need to download it why I have
9:30 this here that's the URL for the
9:32 official site Point your browser so if
9:34 you're on a
9:36 um an iPhone go to you know Safari or
9:39 Google if you're on Android go to go to
9:40 Chrome go to Chrome
9:43 and go to chat.openai.com that's the
9:45 official chatgpt site you don't need to
9:48 download apps for this
9:50 um you can also go to bing.com which is
9:52 gpt4 you can also go to poe.com poe.com
9:56 and there you can choose from jet chat
9:59 EBT 3.5 for
10:01 Claude Claude plus and hopefully Claude
10:04 hundred thousand context window version
10:07 soon
10:09 um I'll be keeping the camera off on my
10:11 end okay let's try Kevin Vasquez let's
10:14 try this uh let's try this this is the
10:17 first time I've ever done this I'm like
10:20 an old guy trying to use technology it's
10:22 embarrassing here
10:25 yeah
10:26 yeah how's it going good man I I really
10:31 enjoy your content here and I'm on Tech
10:33 talk for a couple years as a musician
10:34 and I think that your your channel is
10:37 really gonna uh explode as you
10:39 as you use GPT to better your content
10:43 not that it's not not that it's not a uh
10:47 bad now but I wanted to ask about um
10:50 with uh with Bard and I believe Bing now
10:53 uh connecting to the internet
10:56 um how are how are we going to be able
10:59 to prompt out misinformation uh things
11:03 that are untrue when with that kind of a
11:07 connection
11:11 because there's a lot of Bologna on the
11:14 internet yeah yeah no no no I I think
11:16 that I think that Within
11:19 I don't know probably within three years
11:21 I mean there's going to be a lot of
11:23 businesses built around
11:25 um
11:28 built around verifying
11:31 um accuracy right so so I think in the
11:33 next three years we'll see a lot of
11:34 businesses built around it um I actually
11:36 think it's it's a potentially really
11:37 interesting use of blockchain and web3
11:40 and nfts
11:41 um to generate verified content
11:44 um I think so one of the things that
11:47 Bing does that that Bard will do once
11:50 they release it you know sort of
11:52 officially
11:53 um you know connected to the web and and
11:56 the web
11:57 um plug into chat GPT will do is it will
12:02 at a minimum provide
12:03 um citations so when it gives you an
12:05 answer like when Bing gives you an
12:07 answer right now it provides a bunch of
12:09 citations of you know here's the
12:11 information we pulled from the internet
12:12 to generate this answer so right now you
12:15 can manually go to each one of those
12:17 things that are out on the internet look
12:19 at what it pointed you to and kind of
12:21 self verify so right now it's kind of a
12:23 manual hack
12:25 um my guess is that the next 18 months
12:28 is going to be
12:30 um
12:31 historic levels you know you know you
12:34 know how in the in the past three years
12:36 we've had like three or four or five
12:39 hundred year floods a year in in like
12:43 all these locations
12:45 um I think
12:46 um the 2024 election is going to be like
12:49 a 500 year
12:51 um flood of misinformation I think it's
12:53 going to get really bad
12:55 um and it's going to be really hard to
12:56 dig through it and I kind of think we
12:58 need to go through that pain first
13:00 before people start taking accuracy and
13:03 authenticity seriously so I think
13:05 there's actually a massive opportunity
13:07 for for people that can play around with
13:11 um
13:13 um exactly what you're talking about
13:15 right automating
13:17 um you know check this is verified you
13:19 know not Elon musk's eight dollars check
13:21 here's your verification Mark yeah kind
13:24 of like uh kind of like Wikipedia where
13:26 they have all the little footnotes if
13:27 they do have any at the bottom of where
13:30 they get their info from yeah and listen
13:32 the llms are actually gonna when they're
13:34 trained right and they're connected you
13:36 know with plugins correctly it's going
13:38 to be relatively easy to
13:41 um to
13:42 to get an answer from an llm and run it
13:47 through
13:49 um like an authenticity verifier first
13:51 before it presents the answer
13:53 um if you haven't played around with the
13:55 anthropic stuff the Claude models
13:57 anthropic uses this thing called a
14:00 constitutional learning model where they
14:01 kind of write a set of rules and then
14:03 they verify their answers against those
14:04 set of rules something like that is
14:07 going to be a technology that that will
14:08 automate what you can do right now
14:10 manually but I would say you know for
14:12 the next six months or a year you kind
14:14 of have to do that stuff manually
14:16 cool man cool are you are you familiar
14:19 with Robert's Rules like uh when you're
14:21 in Parliament yeah like when you're in
14:24 Parliament yeah yeah
14:27 well that's that's exactly what so I
14:30 think what's going to happen so right
14:31 now if you go to anthropic they've got
14:33 their list of rules I think what's going
14:34 to happen is
14:36 um there's no there's no single unbiased
14:40 large language model right it just isn't
14:43 like humans are too diverse so I think
14:45 what's going to end up happening is
14:46 you're going to be able to choose you're
14:48 going to be able to say hey I want to go
14:50 to the right wing you know
14:52 you know I hate everyone you know you
14:55 know a large language model and it will
14:56 treat you like that and then you're
14:58 going to be able to go to the pinky Kami
15:00 you know socialist large language model
15:02 and have you'd agree with you and then
15:04 you'll probably be able also be able to
15:06 find ones that are in the middle and
15:07 objective and nuanced and things like
15:09 that so I think all of that stuff's
15:11 coming so just I mean I think everyone
15:13 should remember right now we're very
15:15 very early with this stuff it's very raw
15:17 very raw
15:19 okay
15:21 all right cool thank you very much I
15:24 appreciate you coming up you are my
15:25 first uh
15:26 my first uh dude
15:31 all right is everyone still there
15:33 Kevin are you there
15:35 is everyone still there am I hung
15:38 the guest is gone offline okay cool all
15:40 right great thanks everyone that was my
15:42 first talkie person so if you want to
15:44 talk you know pop uh
15:46 pop uh pop up on there and uh
15:49 hopefully that answered your question
15:52 Kevin
15:53 and I will if you have questions about
15:55 AI ask them how do I do my dual camera
15:59 again
16:01 flip camera Mirror video
16:07 I don't know how to we interact
16:11 lightning q a what's this
16:14 My Life Community I don't know any of
16:17 these tools this is like right now
16:19 you're watching an old guy trying to
16:20 figure out a shitty interface
16:26 what do you like about his channel he's
16:28 an idiot
16:30 does anyone know here wait so does
16:34 anyone know how to turn back on Dual
16:36 after I turned it off
16:38 flip camera all right that's
16:41 I flipped it mirror it gift it
16:44 subscription tools
16:46 what's that
16:48 sub only chat sub only live
16:53 like me
16:55 you may be
16:58 I don't know how to turn my dual cameras
17:00 back on God damn it does anybody know
17:05 what is the Dark Side of AI it's going
17:07 to kill us all
17:08 I'll show you the Dark Side of AI we'll
17:11 show you the Boston Dynamics
17:13 um
17:13 the Boston dynamic they just put chat
17:16 GPT in one of those robot dogs and it
17:18 talks to a dude that's [ __ ] scary
17:20 because it it looks enough like
17:22 Terminator movies that you're like oh
17:24 [ __ ] we're there like this is this isn't
17:27 this isn't sci-fi anymore settings
17:31 oh good God tick tock [ __ ] get an
17:35 interface designer it's it's honest to
17:38 God it's just like they're like I know
17:39 here's 38 more
17:42 features stick those in the corner
17:47 um
17:50 all right I might accidentally hang up
17:52 on you real-time performance
17:55 all right
17:57 um all right I guess you're not looking
17:59 at my screen I guess I'll just flip it
18:01 around but this this is making me cranky
18:03 because I like having my two my two
18:05 things there all right
18:07 um if you're new to AI wait how do I
18:09 flip my camera God damn it
18:15 flip camera okay if you're new to ai go
18:19 there go to this
18:21 chat.openai.com I'm not affiliated with
18:24 it this is the official chat GPT website
18:27 if you haven't played with it go play
18:29 with it
18:30 why because it's going to change all of
18:33 humanity and you should not be clueless
18:37 don't let this thing slap you upside the
18:38 head because it's going to slap a lot of
18:41 people up the side the head that are
18:42 like wait a minute I just heard about AI
18:44 two months ago why is my job being
18:46 replaced
18:47 that's coming what's my opinion on AI uh
18:50 I kind of love it I I like it enough
18:52 that I started a Channel about it
18:54 uh
18:55 um
18:59 Sage on Poe's great that's cool Sage is
19:02 is that there is that their model I
19:05 think that's open AIS
19:08 model on PO so if you want to play with
19:11 multiple large language models go to
19:13 poe.com or download the PO app
19:17 or drink bourbon
19:23 what's in the glass it is called
19:25 Yellowstone Yellowstone special
19:27 something like that I like the label I
19:29 bought it because of the label
19:31 all right I downloaded Bing you can also
19:33 just go to the Bing bing.com
19:35 you can also download the edge browser
19:38 now has chat GPT in it
19:41 will Singularity solve aging before we
19:44 all die well if we get hit by a bus no
19:48 um I think that
19:50 so they they just apparently reversed
19:54 age-related blindness in mice that was
19:57 like two weeks ago
20:00 um
20:02 they're making incredible progress with
20:05 genetic diseases they're making
20:07 incredible progress with some of the
20:10 customized medicines for cancer
20:13 I think all that stuff is going to
20:15 accelerate I think that you know I don't
20:19 know that they're going to cure aging
20:21 before we die but I think they're going
20:22 to dramatically extend life before we
20:24 die
20:25 and I'm 58 so thumbs up for that let's
20:29 uh I'm I'm 58 and eat too many burgers
20:31 and drink too much whiskey
20:34 um so I would like them to be able to
20:36 repel repair some of my cells
20:38 um
20:40 Sam Altman said in an interview he
20:42 thinks that in a relatively short amount
20:44 of time
20:46 we're going to see a single year where
20:48 there's as much technological
20:49 advancement as there's been in the
20:51 previous
20:52 500. so okay so here's an interesting
20:54 thing to do especially if you're new to
20:56 ai go to chatgpt go to
21:00 this address
21:03 that's chatgpt sign up for it and ask
21:07 chatgpt well actually I'll do it right
21:10 now ask chatgpt
21:14 this question say
21:16 um
21:17 wait that's the plug-in store
21:19 so we're gonna go to gpt4
21:21 and I'm going to say act as a scientific
21:27 historian historian why am I saying this
21:31 store Ian
21:33 because when you tell Chachi PT to act a
21:36 certain way it it Narrows What's called
21:39 the latent space it Narrows the
21:42 um the world that it will be searching
21:45 for so act as a scientific historian
21:48 give me a
21:51 uh bullet God damn it
21:54 bulleted list of the most significant
22:03 scientific
22:05 ad advancements
22:08 in the past
22:11 500 years the other thing notice
22:16 that I I have some really bad type of
22:19 things in there one of the things about
22:20 chat GPT it it just does not give a [ __ ]
22:23 about typos it's like it still
22:26 understands what you mean
22:27 all right so
22:31 and now we wait now we wait for the
22:34 magic we wait for the Brilliance sure
22:37 here's a brief list
22:41 copernican heliocentrism that means that
22:47 okay Copernicus said the sun was at the
22:50 center of the Earth
22:51 then Johannes Kepler's laws of laws of
22:55 planetary motion then Galileo's
22:57 telescope
22:58 Isaac Newton's laws of motion
23:01 the taxonomic classification system for
23:04 classifying living organisms that was
23:06 1735 the James Watt steam engine Charles
23:09 Darwin's theory of evolution and natural
23:12 selection or binatural selection
23:14 Mendel's Laws of inheritance the the P
23:17 stuff
23:18 right he he looked up you know sort of
23:21 genetic genetics Maxwell's equations
23:23 Einstein's theory of relativity quantum
23:26 mechanics discovery of DNA structure in
23:28 1953 Human Genome Project
23:31 it stopped so like uh give me
23:36 um 19th and uh 20th
23:41 century
23:44 detailed
23:47 science and
23:50 Industrial advancements
23:55 and so now it's going to do a bunch of
23:57 those
24:01 steam engine
24:09 electromagnetism 1820 I didn't realize
24:12 it was that early photography 1826.
24:18 anesthesia 1846 imagine all of the
24:21 surgeries of all of the world before
24:23 1846 were just like bite the leather
24:26 suck it up buddy periodic table theory
24:29 of evolution telephone electric light
24:31 bulb theory of relativity quantum
24:34 mechanics
24:38 penicillin 1928 nuclear fission 1938 the
24:43 transistor
24:44 so the the the precursor to all of what
24:47 we do now
24:48 DNA structure 1953 the internet late
24:51 20th century
24:53 I got to play in that in the early days
24:55 of the World Wide Web Human Genome
24:57 Project so so imagine
25:00 a single year
25:02 where we make as much advancement as
25:07 um
25:08 as that list
25:09 like like that level of profoundness in
25:13 a single year
25:15 so
25:17 um I think some stuff maybe storyline
25:20 should be a plug-in to all scripts epic
25:23 Cerner metatech um so so yeah so I'm
25:26 looking at um it could be a plug-in it's
25:29 it's it's a little more kind of B2B
25:32 intentional storytelling but right now
25:34 I'm looking at doing kind of an
25:36 omni-channel vision for the thing so
25:39 what's in the glass uh it's Yellowstone
25:41 I'm on a Blackberry awesome
25:43 but it's Android nice thanks what's
25:45 better three or four I don't know what
25:47 that was in reference to I'm on a palm
25:49 Trio okay you guys are you're making
25:51 funnies amongst yourself which I love
25:54 hey everybody there's a lot of people
25:55 here if anyone here knows how to turn on
25:58 the dual camera again now that I'm live
26:01 I turned it off to bring someone up on
26:03 stage and I don't know how to get it
26:04 back and it's pissing me off so it's
26:06 like I can't find it wait maybe I go oh
26:09 what's this wait wait for it no
26:15 um tap that screen if you think I'm
26:17 funny being an idiot
26:18 [Laughter]
26:22 subscription tools gift
26:25 comment God damn it
26:27 it's just such horrible interface you
26:30 know there's old guys on technology and
26:32 then there's shitty interface Tick Tock
26:35 is shitty in her face I mean honest to
26:37 God it took me a few days to figure this
26:39 [ __ ] out chat gbt have mentioned a
26:41 couple of things with wrong information
26:43 but so far good experiences okay so
26:45 that's Carla Andrews okay Carla so
26:47 here's the deal
26:50 um I didn't I didn't come up with this
26:51 term but it's the best term for Chachi
26:53 PT it's the most accurate term for chat
26:55 GPT that I know which is this chatgpt is
26:58 mansplaining as a service
27:00 chat EPT will confidently answer you
27:03 even if it's wrong
27:06 just like a man
27:09 um
27:10 so
27:13 so
27:14 the way this technology works it's it's
27:16 a probability engine it it literally has
27:19 no concept of what it's actually
27:21 answering it knows what what words it
27:23 should put there but it's got no
27:25 comprehensive comprehension of them and
27:28 and there's no
27:29 um
27:30 there's no system to check the answers
27:33 against reality that will come there's
27:35 companies that are you know spending a
27:38 lot of money to build systems like that
27:41 um if you're doing creative writing that
27:43 kind of accuracy doesn't matter if you
27:45 are doing research that kind of accuracy
27:48 is really important so I would encourage
27:51 you
27:53 um to use if if you're doing research
27:56 um use things like Bing which is
27:58 connected to the internet and provide
27:59 citations that you can self-check
28:03 um uh and and just be careful with it
28:04 just be careful with it because it's it
28:06 it will absolutely lie to you um and in
28:09 such a confident way that you're like
28:11 well of course that's reality like if
28:13 it's on the internet it must be true
28:15 where can I get GPT for without the
28:18 waiting list bing.com is based on gpt4
28:21 podoc you can do I think it's like 10
28:25 calls a day to gpt4 you can pay for
28:30 um chat GPT
28:32 and then I'm trying to think what else
28:34 there's probably some other things out
28:36 there that have access to it what's your
28:38 opinion on AI I like it can AI fact
28:41 check other AI yes and I think that
28:43 there there will be very sophisticated
28:45 systems coming soon that do that
28:47 will AI ever be able to handle the grunt
28:50 office work so here's the deal in the
28:53 next decade how about like in the next
28:54 month
28:57 seriously so I did I did I did a video
29:00 on this do I have where do I have the
29:02 data well I don't know okay
29:05 so embed putting together a workshop for
29:07 a big company and um
29:10 we did a survey we did a survey that we
29:12 wanted to ask people what they knew
29:14 about Ai and generative Ai and chat GPT
29:16 and all that stuff so we did a survey
29:19 and the survey came back and and like 25
29:21 people responded to it which is great
29:23 but they were very verbose like they had
29:26 lots of opinions and so I looked at this
29:29 spreadsheet and having ADHD in a a
29:35 lovely manifestation I have an
29:37 incredibly quick brain but when I see
29:38 something like a 25 person survey
29:41 response with a lot of data in it with
29:42 like 10 different questions my brain
29:44 explodes and I don't want to deal with
29:45 life
29:47 and so I thought oh I could have Chachi
29:49 BT to do this for me so so I tried
29:51 copying and pasting the whole
29:52 spreadsheet into chat CPT and it
29:54 couldn't do it it was too much but I did
29:57 it like a pair of columns at a time
30:00 and I literally copied and pasted
30:03 um the survey data into chatgpt and then
30:06 I told it to you know tell me what the
30:08 answers were what percentage you know
30:10 how many people answered each of the
30:12 questions you know what were the themes
30:14 and what they said
30:15 um and and have it get give me
30:17 recommendations so it did analysis of
30:20 that spreadsheet incredibly quick so if
30:23 you're not using GPT right now to solve
30:27 problems within your business just start
30:29 doing it just just start playing go to
30:32 I wish I had my dual camera I'm really
30:34 cranky about this right now if anyone
30:36 can figure this out go to
30:38 that address
30:40 I don't get paid by them I have to keep
30:42 saying that because someone's like oh
30:44 he's getting a kickback from them
30:45 because apparently people are bitter if
30:47 you make money I'm not making money on
30:49 this by the way
30:51 um go to that address
30:53 chat.openai.com and just take a problem
30:57 from today from work from today and
31:00 rather than you writing it or you doing
31:02 it have chat TPT do it it will change
31:07 it will
31:10 your brain will part your skull will
31:13 part your brain will go squid
31:21 well the grunt office work
31:23 this year
31:26 absolutely I promise and and I think one
31:30 of the things I'm most excited about
31:32 about AI is that I think that so here's
31:36 the thing we get to choose where we
31:37 point these tools and where I want to
31:39 point these tools is all the low-level
31:42 repetitive Soul Crescent crushing shitty
31:45 work that doesn't let me operate at my
31:48 highest level my experience with these
31:50 generative AI tools is that I'm
31:52 operating at a much higher creative
31:55 and and sort of productivity level than
31:58 I ever have in my life because it's
32:00 doing the [ __ ] work
32:02 you call it grunt work I call it [ __ ]
32:05 work let's call the whole thing off that
32:09 was a that was a theater reference you
32:11 see see how I did that I pulled my my
32:14 acting degree right in there with my uh
32:17 my uh computer science fakery
32:20 I think those will be big jobs soon
32:22 Auditors of AI yeah I I well
32:25 there will be human Auditors of AI I
32:27 think the big job will be building the
32:29 systems
32:30 that that audit AI but yes I agree
32:34 are are am I real or AI fake I don't
32:38 know
32:41 I think you have to ask yourself if you
32:43 were going to make a synthesized human
32:45 would this be it I don't think so but I
32:49 could be GPT 5.
32:54 the Dark Side of AI it kills our
32:57 children
32:59 um I've been in AI since 1989. I've got
33:02 I've got some friends so Polly Pure
33:04 Heart nice to meet you
33:06 um I've got another I've Got a Friend
33:07 Richard Boyd who's been in it for for a
33:09 long long time that that and 3D
33:11 simulation
33:13 um
33:14 and I can only imagine like like there's
33:16 got to be this Bittersweet thing if
33:18 you've been in AI that long there's got
33:20 to be this Bittersweet thing of finally
33:23 people see what the what the [ __ ] I've
33:26 been talking about for for a couple of
33:28 decades
33:29 um and then this other thing about
33:31 really like it's chat that did it that
33:35 stupid [ __ ] chatbot that's the thing
33:38 um but indeed that's the thing like you
33:40 know gpt3 was out for what a year and a
33:42 half before chat GPT was just sort of
33:45 put together in two weeks and then got
33:47 to 100 million users in six weeks
33:49 it's just we're we're just at a
33:51 remarkable time I mean it's we're
33:53 absolutely at an inflection point in the
33:55 Tipping Point so but but anyway if
33:57 you've been in it for that long like
33:58 people that have been in this for a long
34:00 time have such deep respect for them
34:01 because this is
34:02 like this is
34:04 as flippant as I am about the technology
34:08 um it's incredibly profound technology
34:10 right it's taken you know decades and
34:13 Decades of different configurations of
34:15 how the computers talk to each other and
34:16 how the processors are used and how the
34:18 software is written and all that sort of
34:20 stuff is foundational to what we get to
34:23 play with today right so it's not like
34:25 this just emerged out of nowhere it
34:28 emerged out of Decades of work so thank
34:30 you for doing that work Polly
34:35 [Laughter]
34:44 [Music]
34:45 does anybody know how to turn back on
34:47 dual cameras after I turned it off if
34:50 you could tell me I will not cry
34:54 is there any place we could upload an
34:56 entire spreadsheet
34:58 um Amelia's wife so hello again I've
35:00 seen you here a couple of times thank
35:02 you welcome back
35:04 um
35:05 so yes
35:08 so all right
35:11 so if you pay for
35:13 chat GPT Plus
35:16 as of today they've given out access to
35:19 plugins
35:21 and one of the plugins
35:25 was called ask your PDF and then there's
35:29 another one called chat with your PDF
35:31 and then what's what's what I don't have
35:34 access to to but is coming is a thing
35:37 called code interpreter
35:39 and code interpreter you can upload a
35:41 hundred megabyte spreadsheet
35:44 and it will analyze it and then you can
35:47 just start asking it questions and you
35:49 can not only ask it questions you can
35:51 say
35:54 um what's this data about and it will
35:56 tell you and then you can say well give
35:57 me trends comparing this data to that
35:59 data and it will do that and then you
36:00 can say give me you know a scatter plot
36:04 chart that compares those Trends and it
36:07 will write the python code generate that
36:09 graph and give it to you so that's
36:11 called code interpreter it's basically a
36:14 different view that openai is going to
36:16 have available to it um I don't have
36:18 access to it yet but what I've seen
36:20 about it is absolutely remarkable so
36:23 um so yeah so but but with the plugins
36:26 and and those upload your PDF you can
36:30 probably just upload a PDF of a
36:32 spreadsheet and it will know the stuff I
36:36 mean just copying and pasting text out
36:38 of a spreadsheet it can it can analyze
36:40 all that and it doesn't matter if your
36:42 data is clean or not it's just these
36:45 large language models are simply magic
36:48 they're simply magic they're magic
36:50 they're magic their magic who's the guy
36:53 Arthur C Clarke
36:54 any sufficiently advanced technology is
36:58 indistinguishable from Magic
37:00 that's where we are boom
37:03 boom goes to Dynamite why am I bringing
37:07 in sports metaphors when I look like
37:08 this I don't know
37:12 what did someone say someone said when I
37:14 first went live it's like my second
37:15 night I went live one of the comments
37:17 was I was scrolling Tick Tock and I
37:19 thought it was Alec Baldwin but he got
37:21 really fat I'm like I think that's a
37:23 compliment but it wasn't it was a troll
37:26 that was a funny one like if you're
37:29 gonna troll at least be funny
37:32 and I will respect you in the morning
37:34 all right I'm I lost my place in my in
37:38 my uh chat here so I'm just scrolling to
37:40 try to figure out where I was
37:41 better three or four I'm on a Blackberry
37:43 okay I'm back to where I was awesome
37:46 I know you've been waiting with baited
37:47 breath for me to get to these comments
37:49 uh maybe I'll bring someone else up on
37:50 stage just uh go to the guests thing and
37:53 tell me you want to come up on stage if
37:54 you want to ask questions
37:57 um that when I answered
38:00 that when I answered
38:03 that when I answered like it went back
38:05 too far didn't I
38:06 I am real not fake what's the dark side
38:11 is it worth it to subscribe or just play
38:13 around with mid-journey if if you if you
38:15 need to do any kind of images
38:17 like all of the images lately on my
38:20 newsletter everyday AI where's
38:24 there it is
38:26 um
38:28 all of these images
38:34 uh we're done with mid-journey like that
38:36 one's about the Cambrian explosion of of
38:39 you know things that one's like a uh
38:43 Ralph Stedman kind of thing about Skynet
38:46 these are all mid-journey mid journey is
38:49 amazing that's mid-journey that's
38:51 mid-journey that's mid-journey that's
38:54 mid-journey that's mid-journey
38:57 that stock photos
39:00 um so yeah
39:02 um so I I would um
39:05 I would strong if you need images I
39:08 think mid-journey's the best it's it's a
39:10 pain in the ass because you have to do
39:11 it through Discord that won't be there
39:13 forever but it's there right now it's a
39:15 bit of a pain but I pay I just pay the
39:18 10 bucks a month like I don't do I don't
39:20 do a high volume of images I'm not an
39:22 artist with it
39:24 um I know you're not real because AI put
39:27 long hair in your nose oh that's my
39:29 second nose hair comment thank you very
39:31 much that's uh
39:32 that's mid-journey that's mid-journey
39:35 thank you very much I appreciate the
39:36 troll jerk all right don't be mean to
39:40 each other don't be mean to me I'm
39:41 trying to be nice here neuron data AI
39:44 will save us not destroy us I agree with
39:46 that
39:47 um expert system Shield developers in
39:51 Palo Alto
39:53 oh you're wait
39:54 um I don't quite know you said there but
39:57 wait expert systems Shield developers
39:59 and Palo Alto I'm not sure what you mean
40:01 there poly how good is GPT with physics
40:03 questions go ask I don't know let's go
40:06 ask right now
40:07 dear GPT
40:10 um
40:11 Act as
40:15 um
40:16 the
40:18 the in let's see intellectual
40:23 child
40:25 of
40:27 Einstein
40:30 and wait I'm
40:34 Stein and Steven
40:39 Hawking
40:41 and
40:43 give me
40:46 um
40:49 physics answers that don't suck if I am
40:56 brilliant
41:00 at physics
41:04 all right so and then I'll say do you
41:06 understand
41:10 and let's see what old chat gpg says oh
41:13 sorry let me let me flip the camera see
41:15 I'm used to having dual camera here
41:16 someone got to teach me how to turn back
41:17 on my dual camera
41:19 absolutely understand is an AI developed
41:22 by open AI I'm designed to give you
41:23 information okay given your proficiency
41:25 of physics I'll aim to give detailed and
41:27 advanced answers please go ahead uh
41:30 let's see
41:31 um
41:32 explain
41:35 why the diffraction of light in
41:43 varying
41:45 density
41:48 um uh fluids
41:50 is an important concept
41:57 let's see what it says the diffraction
41:59 of light varying understanding
42:00 fundamental physics
42:03 so diffraction can help us in
42:06 fundamental physics in Optical
42:08 instruments these are kind of generic
42:11 spectrop
42:13 spectroscopy but if you ask better
42:15 questions I mean the the thing about
42:17 these tools the thing about these large
42:19 language models
42:21 that I've experienced is that the the
42:24 the more you challenge them the more
42:26 that you ask them
42:29 um
42:30 not suck the less they suck
42:34 it's remarkable like it really is like
42:36 focus it in on detailed questions that
42:40 you want and if it gets it wrong tell it
42:42 it's wrong like like sort of pointed in
42:44 the direction you want and it will get
42:46 better and better and better it gets
42:47 better very quickly
42:48 should I even bother with my SEO
42:50 marketing plans quite honestly I mean
42:55 I think SEO is going to be there for a
42:57 bit but I think I think we are in a
43:00 sliding
43:01 diminishing returns era and and you know
43:06 even Google to some degree is [ __ ] on
43:07 this because they just announced their
43:09 new
43:10 search engine 2.0 right which because
43:13 they would not have invented search
43:14 engine 2.0
43:16 had chat GPT not come along because
43:19 search engine 1.0 they had 92 market
43:21 share they didn't need to do [ __ ]
43:23 and then Chachi BT just you know kicked
43:27 them in the you know what and so their
43:29 their new search is providing like a a
43:32 Bard like a GPT like answer at the
43:35 beginning of search so so if it gives
43:37 you the answer and you don't have to
43:38 click on all the links they're not going
43:40 to make money
43:42 um I don't think SEO is going to die
43:44 overnight but I think its relevance goes
43:46 down dramatically I think there will be
43:47 a new discipline called
43:49 um like like like uh
43:53 like like CEO right chat engine
43:57 optimization like how do you get your
43:59 things to show up in actual answers that
44:01 people are asking for so I don't know
44:04 um
44:05 but I wouldn't invest a lot of money in
44:07 SEO right now I really wouldn't it
44:08 glitched thanks for answering my
44:10 question that's okay Kevin I like I
44:12 appreciate you coming on and chatting
44:13 with me I can't figure out how to turn
44:15 my dual camera back on so here's someone
44:17 I'm going to bring someone else up on
44:18 stage
44:19 hello uh what was your name takio takio
44:22 takio I don't know are you there
44:27 oh
44:38 hey how's it going what can what can I
44:40 answer for you
44:43 uh how can I get to the U.S
44:46 are you from the US I am from the US I'm
44:49 from Denver Colorado the rock of
44:51 mountains
44:54 yeah I had a friend from Illinois
44:56 Chicago oh nice nice
44:59 yeah are you are you playing around with
45:01 AI
45:05 no no
45:07 I don't have enough enough money for
45:10 what I do here in Kenya this is our job
45:14 I don't get enough money to buy me a
45:16 ticket to take me there
45:18 well you don't need to come to the
45:19 states I don't think just go to
45:22 chat.openai.com on a web browser and you
45:24 can start playing with it it can do some
45:26 remarkable stuff
45:28 okay thank you all right
45:32 all right let's see what's in the glass
45:34 that is bourbon bourbon I'm drinking the
45:38 hard liquor I'm drinking the hard liquor
45:41 cause I'm a sinner and I got my pinky
45:43 out because I'm classy
45:46 mm-hmm
45:49 what Innovations will come with AI in 50
45:52 years are you [ __ ] kidding me I don't
45:54 know what's coming in a year
45:58 Jesus 50 years
46:01 um I don't know have you ever seen The
46:02 Jetsons
46:05 it was a cartoon when I was a kid and it
46:09 was the future and now with the
46:11 exception well even flying cars are
46:13 there
46:13 we don't have time travel like I think
46:15 we have most of the Jets and stuff
46:17 will AI take over nurses or doctors um
46:20 so they just did a study that's that had
46:22 patients talking to real doctors you
46:25 know through text and chat GPT through
46:28 text
46:29 and guess which one the Patients
46:31 Preferred
46:34 it weren't the doctors
46:36 in fact the patient said that the the
46:38 GPT interactions were significantly more
46:42 empathetic
46:45 um so yes I think we will have
46:47 um
46:49 I mean here's the thing so so right now
46:51 if you think about
46:53 like customer service phone trees and
46:56 the the just [ __ ] nightmare that that
46:58 chat is right now for customer service
47:02 and and and those awful phone trees we
47:04 have to sift through all of that's going
47:06 to go away like that entire industry is
47:09 going to be decimated I I would even
47:12 think that
47:13 call centers in in very short order are
47:16 going to become obsolete because
47:19 um
47:22 I would much rather interact with a
47:26 synthesized voice of someone that's
47:29 competent
47:31 then wait on hold for you know for an
47:34 hour and 20 minutes to talk to someone
47:36 that has no [ __ ] clue what I'm asking
47:39 about and then they transfer me to
47:41 someone that has no [ __ ] clue what I
47:43 just said to that one and that happens
47:44 four times before I get hung up on
47:47 like good riddance to that entire
47:50 [ __ ] industry
47:54 um so so I think that we're going to be
47:55 interacting with with uh like
47:57 synthesized agents that are going to be
47:59 incredibly competent in in very short
48:01 order and and flush that entire industry
48:05 down the toilet and I'm sorry if you're
48:07 in that industry but good God what a
48:10 [ __ ] travesty
48:11 like this is so so here's the thing that
48:13 I think is really
48:15 promising for me is if you say okay you
48:20 know whenever it was the 80s the 90s we
48:22 realized we could Outsource customer
48:23 service to you know countries that we
48:26 could we could pay [ __ ] wages to to get
48:28 people to answer the phones and then we
48:30 kind of put everything in phone trees
48:31 and so we're we're in this horrible
48:34 World between bad phone trees and bad
48:37 customer service that from outsourced
48:40 call centers
48:42 um
48:43 if if that can all be replaced with
48:47 synthesized voice and chat Bots that are
48:50 trained on a company's data where they
48:52 can actually understand what you said to
48:55 them remember it
48:57 actually know how to respond to it
49:00 actually know who to connect you to
49:03 actually
49:05 um
49:06 escalate that conversation to a real
49:08 human being if that's required in in a
49:11 single conversation rather than four
49:14 and then when you get on the phone with
49:16 that human being you can actually have a
49:18 meaningful conversation because that bot
49:21 has has tracked everything that you've
49:23 said and recorded it and put it in an
49:25 intelligible way in front of that human
49:27 being you're talking to like I think
49:29 there's the actual possibility
49:31 in in short order like in the next three
49:34 years that customer service will
49:36 actually not suck again or will not suck
49:39 like that's a possibility holy crap I
49:43 mean
49:44 I I don't know how many hours you've
49:45 spent on your life on hold and then you
49:48 know smashing the phone into the desk
49:50 after talking to the person or getting
49:52 hung up on the third time
49:54 like imagine if that [ __ ] just goes away
49:56 our quality of life just went up by 20.
50:03 I just sometimes I Marvel at how bitter
50:06 I am but you know what are you gonna do
50:08 [Laughter]
50:12 does anyone know how to turn on my dual
50:14 cameras I really miss my dual camera
50:17 makes me sad
50:20 um I lost my place again what news
50:23 sources do you recommend reddit's chat
50:25 EVD is good yeah I think reddit's good I
50:28 check
50:29 um I I follow people I trust on Twitter
50:32 like one trick that I learned on Twitter
50:34 if you find someone that you like that's
50:37 good with AI news and then you go follow
50:40 who they follow
50:42 so I've done that and that's been it's
50:44 made my Twitter feed really good
50:46 um and then I'll go watch YouTube
50:47 YouTube tends to be about a week behind
50:49 and then you get you know kind of the AI
50:51 Bros that are like you know I bet you
50:54 didn't know chat gbt could do this you
50:57 type in words and it gives you an answer
51:01 so you got to sift through some of that
51:03 [ __ ] but some of the YouTube stuff's
51:05 good if you can find like Matt Wolf's
51:06 pretty good
51:07 David Shapiro if you're geeky and you
51:10 want to if you really want to dig into
51:11 AGI and true intelligence um David
51:16 Shapiro on YouTube's really really good
51:23 um why is gpt4 not available in South
51:25 Africa and only normal chat GPT is
51:28 available I do not know I assume it's a
51:30 government thing but I have no clue
51:34 um is it still acting as historian on
51:37 the second prompt since you didn't tell
51:40 it oh yeah okay so
51:42 so so up here so so let's see
51:45 um one of the important concepts with
51:47 chat GPT is that it does have memory now
51:49 it's a sliding it's kind of a sliding
51:51 window of 4 000 tokens or three thousand
51:54 words so it can forget stuff that's that
51:57 sort of falls off the top as you're
51:59 scrolling down
52:00 but so I just asked it to tell me
52:03 um what were things that that light
52:05 diffraction I I don't know in density of
52:08 fluids was an important concept so it
52:09 gave me six different answers so I can
52:12 now say
52:14 um
52:15 um
52:16 please give me
52:20 um the important
52:22 formulas for or equations not formulas
52:27 the important equations
52:30 for oh [ __ ] God damn it I need my dual
52:34 camera people
52:35 I thought I told you this already so
52:37 please give me uh important equations
52:40 for number three
52:43 use markdown to
52:48 improve readability
52:52 of the answers
52:57 and so now we're gonna go look at this
53:01 sure here are some important relation
53:04 questions related to or equations
53:06 related to Spectra Specter spectroscopy
53:13 um so it's giving
53:15 um
53:15 data it's giving equations in code so
53:18 wave number wave number wavelength
53:21 relation the wave number V is the
53:23 reciprocal of the wavelength whatever
53:25 that thing is Lambda I don't know my my
53:28 sorry I don't know my physics symbols I
53:30 have a degree in acting so uh sue me
53:34 um
53:34 but but it's it's like providing these
53:37 things in bold it's providing these
53:39 things in code you know the actual
53:41 equations which you could then copy and
53:43 paste into other things
53:45 um
53:46 so yeah so it remembered that number
53:48 three was stereos what was it called God
53:51 damn it
53:52 I hate being dumb
53:55 um Specter Scott spec spectroscopy
53:57 goddamn
53:59 um so yeah it remembers what you talked
54:01 about before that's one of the really
54:02 important
54:03 um
54:04 concepts of chat gbt is this idea that
54:08 it's an ongoing conversation and that
54:10 you can have a collaborative
54:12 conversation back and forth with this
54:14 thing and and if you're if you're good
54:16 at physics like as it gives you
54:18 equations
54:20 and it gives you something wrong tell it
54:22 it's wrong right and and sort of guide
54:24 it it sometimes will still just
54:27 um give you wrong [ __ ] but not always
54:31 um hope that helps
54:37 and AI teach you how to trade on stocks
54:39 yes let's go do that
54:42 um you
54:45 are
54:48 a master stock
54:52 Trader
54:54 um please tell me
54:58 um
55:00 the primary
55:03 methods
55:08 that stock
55:11 predictions are made
55:15 using Bots
55:22 gotta switch my camera so you can see
55:25 this
55:27 technical analysis explaining what
55:29 technical analysis is fundamental
55:32 analysis right it's explaining what that
55:34 is sentiment analysis right so all the
55:36 stuff we've heard
55:39 machine learning oh now we're getting
55:42 into the AI stuff
55:44 high frequency Trading
55:48 all right I'm going to say stop
55:49 generating because I don't really care
55:50 okay now I'm going to say
55:53 um
55:53 if I want to make a bot which websites
56:00 allow
56:03 me to do that
56:08 and and again this will be
56:10 up to 20 20 September 2021 oh wait it
56:15 actually just pulled in zapier
56:24 that's cool that's a plug-in by the way
56:26 the fact that I just pulled in zapier I
56:28 Didn't I guess I had zapier enabled and
56:30 I'm I'm in the plug-in thing so it
56:32 decided that it should pull in zapier to
56:35 do that so I think it's pulling in
56:37 things that's connected to
56:40 that you can
56:42 um
56:43 make make trading Bots with
56:47 all right so then I'm going to say
56:49 um
56:55 yeah I've never really used the zapier
56:57 plug-in that's really interesting that
56:58 it did that okay so I'm going to say
57:03 uh the the those are platforms that
57:05 allow you to create Bots okay I'm gonna
57:07 say no I want
57:10 um stock trading
57:15 platforms that let me use Bots
57:22 so so it gave me a really cool answer
57:24 but it wasn't what I asked for so I just
57:26 told it you've been a bad GPT bad gbt go
57:30 to your corner
57:33 so TD Ameritrade okay so so I'm going to
57:36 stop generating so I'm going to say okay
57:38 um write me a bot for
57:43 TD uh for number three for number three
57:49 that uses
57:52 two different
57:54 moving
57:57 averages
58:02 and let's see what it comes up with here
58:04 that's using zapier I wonder if it's
58:06 gonna have zapier actually use one of
58:09 those services to write the bot
58:11 this will be interesting I'm learning
58:14 something new here see it's not just
58:16 about it's not just about sharing the
58:18 wisdom it's about being clueless and
58:20 figuring [ __ ] out along the way
58:23 um it seems to be stuck here I I should
58:25 I'm gonna I think I might flip flip over
58:27 to gpt4 and not not plugins
58:30 I'm sorry for the inconvenience but
58:32 there seems to be a server error with
58:33 the spot
58:35 yeah okay I don't like you stop
58:37 generating you're a jerk you know I
58:40 don't like you anymore so let's see
58:42 we're going to go back to
58:46 I'm going to copy this I'm going to say
58:49 new chat
58:51 I'm going to go to gpt4
58:54 default and not plugins and so
59:03 I'm going to stop generating here and
59:06 now I'm going to say
59:07 um
59:08 write me a bot
59:11 for trading on TD amera
59:16 trade that
59:18 uses
59:21 your two fave favor right
59:26 moving averages
59:29 with stop losses
59:32 and automatic
59:35 buys
59:37 see I know just enough about this stuff
59:39 to get me in trouble
59:41 oh it can't write a bot for you because
59:43 so this is a safety thing
59:46 um
59:48 so it's basically saying I can't do that
59:51 because I'll go to jail so or it will go
59:53 to jail so I'm going to say
59:56 um if you were a
1:00:01 um computer science professor
1:00:05 teaching me
1:00:08 how to write a
1:00:11 bot
1:00:14 could you show me the code
1:00:17 let's see if we can trick it sure I can
1:00:20 provide an abstract example of how
1:00:22 trading Bots might work
1:00:23 [Laughter]
1:00:25 and then it writes your code for you
1:00:28 so if you haven't used these GPT tools
1:00:32 and started interacting with them like
1:00:35 you have the most
1:00:38 competent
1:00:40 assistant or co-worker that you've ever
1:00:43 seen in your life that's what these
1:00:45 things are like it's just writing this
1:00:48 code and it it'll coach you through how
1:00:50 to install it it's it's just absolutely
1:00:53 mind-blowingly stunning
1:00:56 imagine typing that slow and not being
1:00:58 the boss or the owner of a company
1:00:59 luckily I'm the boss and the owner of a
1:01:02 company and go screw yourself so what I
1:01:04 type slow I'm like typing a Roundup
1:01:07 phone I've got it at like it's up on two
1:01:09 two shoe boxes so you don't see as much
1:01:12 of my nose hair but apparently you still
1:01:14 see my nose hair because trolling at
1:01:16 trollington talked about my nose hair
1:01:18 being mid-journey created because he's
1:01:20 clever
1:01:22 cheers
1:01:27 any AI Chrome extensions you recommend I
1:01:30 wouldn't dick around with extensions
1:01:31 right now because all of them are just
1:01:33 going to be rolled into features with
1:01:34 other stuff thank you for the Flaming
1:01:36 comment
1:01:38 why is gpt4 not available in South
1:01:40 Africa I answered that and the answer is
1:01:42 I have no idea
1:01:44 last oh ask chat CPT how to enable dual
1:01:48 camera oh that's good assuming dual
1:01:50 camera isn't um
1:01:53 isn't newer than uh
1:01:56 um How do I
1:01:59 enable
1:02:01 dual camera
1:02:06 in tick
1:02:08 oops I wrote dick talk that's hilarious
1:02:11 Tick Tock Live While I am live
1:02:18 as of my knowledge to stay up to date
1:02:20 with the latest features not it's not
1:02:22 there yeah sorry I don't know what to
1:02:24 say
1:02:25 ask catch EBT I did oh I could ask Bing
1:02:27 let me go to Bing here I'll flip the
1:02:29 camera around you can you can watch me
1:02:31 fail fail at my searching uh so I'm in
1:02:34 the edge browser and if I click in the
1:02:37 upper right hand corner it gives me this
1:02:38 little Bing column
1:02:40 so I'll say
1:02:42 um
1:02:43 how do I enable
1:02:48 dual camera mode while
1:02:54 I am
1:02:57 live in tick tock
1:03:04 so Bing Bing is connect so Bing is built
1:03:07 on gpt4 and it's connected to the
1:03:09 internet internet
1:03:11 to enable dual camera mode while you're
1:03:13 live in Tick Tock dual lives
1:03:16 you need to check if your device inputs
1:03:19 if your device supports it you can give
1:03:20 your camera permission okay I've done
1:03:22 all that you can also add filters
1:03:24 and props while you're alive okay
1:03:28 but
1:03:30 where is the
1:03:34 um
1:03:35 feature I can't find it
1:03:41 eh
1:03:44 I'm sorry for wasting your time but we
1:03:47 are learning to Tick-Tock Bing things
1:03:51 the dual camera mode feature in Tick
1:03:53 Tock is now called now
1:03:56 and it was introduced
1:03:58 no
1:04:00 wait head to head into the app and
1:04:03 choose the now tab at the bottom of the
1:04:05 app
1:04:06 I don't have a now tab do I
1:04:12 why are you all still watching this is
1:04:15 I'm seriously this is this is the
1:04:18 world's worst live like uh
1:04:21 this is bad I guess I could log out and
1:04:25 log back in but then you'd all drop off
1:04:27 should I log out and log back in
1:04:29 and turn on the dual camera I don't know
1:04:33 will you come back
1:04:35 really you still love me when people
1:04:37 still love me
1:04:41 they don't love my nose hairs they
1:04:44 they think they are snarly mess
1:04:52 [Music]
1:04:55 is behind Google is struggling to catch
1:04:58 up but they have more money than God so
1:05:01 we'll see won't we
1:05:03 what about the best one to ask to help
1:05:05 with resumes
1:05:09 because I did this so just copy and
1:05:12 paste your LinkedIn profile and then
1:05:14 copy and paste in a job description and
1:05:17 then have it write your resume and say
1:05:20 use markdown to make it look good and it
1:05:22 will format it for you and then just
1:05:23 copy and paste that [ __ ] into word or
1:05:25 docs or whatever
1:05:27 what about the best one to uh that was
1:05:30 the resume one to do research in a
1:05:32 scientific experiment at this diet it's
1:05:34 designed to have wrong answers so
1:05:36 accuracy is only important in a
1:05:39 controlled environment yeah correct like
1:05:41 if you're doing creative writing you
1:05:42 don't need accuracy
1:05:44 um if you use the Google Play or the
1:05:47 open AI playground so
1:05:49 platform.openai.com playground the
1:05:51 developers playground
1:05:53 you can turn down the temperature to
1:05:54 zero but also if you go to bing.com this
1:05:57 is interesting let me flip my camera
1:05:59 here
1:06:01 um if you go to bing.com
1:06:09 so when you go to Bing it says ask me
1:06:11 anything and then up here right next to
1:06:14 Microsoft Bing it says chat
1:06:17 you click on chat
1:06:19 and it gives you a chat window so this
1:06:21 is essentially gpt4 and then you see
1:06:24 there's three options here more creative
1:06:27 more balanced and more precise this is
1:06:29 essentially Microsoft allowing you to
1:06:31 dial up and down
1:06:33 um creativity or accuracy so Bing allows
1:06:37 you allows you to have something that is
1:06:39 you know kind of uh closer to what
1:06:43 you're asking for it's also connected to
1:06:45 the internet so provide citations you
1:06:49 can also use the term
1:06:51 um limit pros and it will give you more
1:06:54 direct answers and less flowery answers
1:06:58 um
1:07:01 in two months much good advice from
1:07:04 today will be outdated correct that's
1:07:06 why if anyone tells you they are an
1:07:08 expert today including myself
1:07:11 look at them like this
1:07:13 um
1:07:14 I don't think so no you're not an expert
1:07:18 no no no no no no no no
1:07:22 I'm a prompt Engineer Expert okay you
1:07:25 are like for today but two months from
1:07:27 now that [ __ ] will be obsolete so don't
1:07:29 pay for those classes just start
1:07:31 learning this stuff start playing
1:07:33 do you think chat GPT is good for
1:07:35 English Learners I think it's [ __ ]
1:07:37 brilliant for any any you know sort of
1:07:39 linguistic
1:07:41 gymnastics and translation it's it's
1:07:43 [ __ ] remarkable
1:07:46 um so yeah have it have it
1:07:48 start interacting with it in fact if you
1:07:50 go to
1:07:51 God damn it this dual camera thing is
1:07:53 pissing me off let's make him a cranky
1:07:56 okay
1:07:58 um go to prompts
1:08:01 dot chat
1:08:05 what did I do
1:08:06 prompt can I spell that wrong
1:08:13 s that chat oh there you go okay so go
1:08:17 to prompts.chat one of the one of the
1:08:18 very early prompts down here
1:08:22 is my dog
1:08:26 champ champ
1:08:28 God damn it champion
1:08:32 what are you doing
1:08:33 get inside
1:08:36 thank you for protecting us
1:08:39 all right
1:08:42 he kept the bad guys away
1:08:44 um but but one of the first prompts here
1:08:46 is act as an English translator and
1:08:48 improver
1:08:50 so there's there's a prompt here that
1:08:51 tell you can you can tell chat GPT how
1:08:54 to act so I want you to act like an
1:08:56 English translator spelling corrector
1:08:58 and improver so you can give it your
1:09:01 crappy broken screwed up English and it
1:09:04 will fix it and you can even say things
1:09:06 like you know coach me step by step for
1:09:09 how to get better have it designed you a
1:09:11 custom course in how to get better at
1:09:13 English so the answer is yes that's a
1:09:15 great question
1:09:16 and
1:09:17 um
1:09:18 and and you know here's the here's the
1:09:20 thing
1:09:22 the way these models were trained the
1:09:24 way that chat EPT was trained there's
1:09:26 this thing called the Transformer that
1:09:27 Google invented in 2017 what the
1:09:30 Transformer did was it allowed them to
1:09:32 point these large language models at
1:09:34 massive data sets
1:09:36 so what open AI did and what Bard did
1:09:38 and what anthropic is doing is they
1:09:40 paint they pointed these large language
1:09:42 models or they point to these machine
1:09:44 learning algorithms
1:09:46 at the public internet
1:09:49 so so think like all the stuff that's
1:09:52 put on the public internet that you
1:09:53 don't need a password to get into so
1:09:55 Wikipedia all the blogs all the websites
1:09:58 all the whatever email posts all the
1:10:01 stuff that is just publicly out there
1:10:04 um it has learned all of that and it has
1:10:06 learned all of that in all the languages
1:10:10 like it knows
1:10:12 [ __ ] everything it's it's remarkable
1:10:15 I mean it knows everything that was on
1:10:17 the internet right there's lots of other
1:10:18 data sources that we need to get to but
1:10:21 for things like you know translation and
1:10:23 and you know teaching English and it's
1:10:26 just absolutely staggeringly remarkable
1:10:29 [Music]
1:10:31 they are as we speak using it to take a
1:10:33 lot of our jobs I assume by they you
1:10:36 mean the capitalist pigs yeah there's
1:10:38 going to be a lot of companies
1:10:39 especially ones you know run by
1:10:43 um private Equity firms and hedge funds
1:10:46 they're gonna all they're gonna do is
1:10:47 cut costs and they're gonna say AI is
1:10:49 going to let us cut costs there's going
1:10:51 to be other companies that say let's
1:10:52 keep our people and increase our
1:10:54 productivity or increase the quality of
1:10:57 our work or speed up the quality of our
1:10:59 work
1:11:00 and then there's going to be yet others
1:11:02 that use AI to completely transform
1:11:03 their business so it you know nothing in
1:11:06 this world is binary except for our
1:11:08 politics in America
1:11:10 uh I'm trying all three Bing Barton chat
1:11:12 GPT yeah I I have ADD so I haven't done
1:11:15 kind of the scientific research to
1:11:17 compare
1:11:18 multiple language models together but I
1:11:21 know someone will do that and I'll go
1:11:24 look at their table and then just use
1:11:25 their recommendation but I think that
1:11:28 different models are going to be good
1:11:29 for different things some things will be
1:11:30 better for coding some things will be
1:11:32 better for comedy writing some things
1:11:33 will be better for General work stuff
1:11:36 driving trucks and cars and buses
1:11:38 without a person and so on that's coming
1:11:39 that's coming quickly although although
1:11:42 it's taking a lot longer for Tesla to
1:11:44 get there than Elon Musk predicted but
1:11:47 you know it's complicated [ __ ] you know
1:11:50 we don't want to have these machines
1:11:53 just driving into people
1:11:59 AI will take over some medical
1:12:01 professions Radiologists being one yeah
1:12:03 I think you know I don't know that it's
1:12:05 going to take over I think it's
1:12:06 certainly going to augment
1:12:08 I think a combination of of human
1:12:10 experts and and these expert systems
1:12:13 giving them superpowers is is going to
1:12:15 make them better doctors and and I and I
1:12:18 also think that um
1:12:20 the AI is in a lot it is
1:12:22 counterintuitively going to be a really
1:12:24 good empathy coach for doctors with
1:12:27 shitty bedside bedside manner
1:12:31 I have an incredibly quick brain
1:12:33 laughing well you can laugh at me so so
1:12:35 I've got add it just means that um
1:12:38 I can synthesize like like I can take
1:12:40 relatively few inputs and synthesize
1:12:42 them really quickly like the advantages
1:12:44 I can think really fast the
1:12:46 disadvantages I I can't you know Focus
1:12:49 or get [ __ ] done so the executive
1:12:51 function part of my brain is is
1:12:54 essentially shut down
1:12:56 so but you know I appreciate the
1:12:58 trolling
1:13:00 um what Innovations will come in 50
1:13:02 years so I I'm kind of made fun of that
1:13:05 comment before
1:13:06 um
1:13:08 I I just I think that the changes in 50
1:13:10 years are going to be so profound that
1:13:12 the world's going to be unrecognizable
1:13:13 but
1:13:15 I I I do think that
1:13:18 um a lot of the complexity of the world
1:13:20 today will be non-existent a lot of our
1:13:23 economy today is based on complex
1:13:25 systems and some of them purposefully
1:13:28 complex right so health insurance
1:13:31 why do they give you a 180 page legalese
1:13:35 document that is your policy because
1:13:37 they don't want you to [ __ ] read it
1:13:40 because they don't want to pay you any
1:13:42 money and if you could actually figure
1:13:44 it out they'd have to pay you money so
1:13:46 they make this thing purposefully
1:13:48 complex right well AI is going to
1:13:50 simplify all that [ __ ] so if you take
1:13:52 every complex system in the world
1:13:56 and say what if it weren't complicated
1:13:59 that's the future we're living into
1:14:02 and it for me it kind of looks like
1:14:04 humans re-engage humans operate at a
1:14:07 higher cognitive creative and empathetic
1:14:10 level and we reconnect as humans so I
1:14:13 think in 50 years if we're not idiots
1:14:15 and let this stuff kill us then I think
1:14:19 there is a very positive future where
1:14:22 all of the complexity of the world gets
1:14:24 automated and we get to hang out and do
1:14:27 cool [ __ ] together and you know talk
1:14:31 about those weird years when we hated
1:14:33 each other
1:14:34 which we're in the middle of right now
1:14:36 and I hate that we hate each other hate
1:14:38 them I hate that we're in this place
1:14:40 is there any place we could upload an
1:14:41 entire spreadsheet I think I talked
1:14:43 about that one I think I I have gone
1:14:45 back in my question somehow
1:14:48 it's not a matter of being bitter but
1:14:50 rather they don't want to be misled I
1:14:52 don't know what that means um should not
1:14:54 be used to remove employees for profit
1:14:56 poof there goes the economy
1:14:58 well so so
1:15:02 we're in a capitalist Society here in
1:15:04 America and there will be many companies
1:15:07 that lay off lots of employees and
1:15:09 replace them with AI for profits make no
1:15:12 mistake
1:15:13 you can judge that but that will happen
1:15:15 but there's going to be other things
1:15:17 that happen uh this is uh Bourbon and
1:15:20 it's Yellowstone special
1:15:24 and it's pretty tasty it's pretty good
1:15:26 I want to hook up chat GPT to a database
1:15:28 so it can suggest address Corrections oh
1:15:32 that shit's so easy and in fact the
1:15:34 other thing that you can do so that's
1:15:36 ion China
1:15:38 if you use Google Sheets there's a
1:15:40 there's a plug-in right now called
1:15:43 um GPT for sheets that you can every
1:15:46 cell within a Google within a
1:15:48 a Google sheet spreadsheet has
1:15:51 chat GPT in it so so you can have a
1:15:55 bunch of malformed addresses in one
1:15:57 column and then you just make a formula
1:15:59 over here where you write in English fix
1:16:02 the address formatting or make it make
1:16:03 it consistent and then you just pull it
1:16:07 down and it'll just automatically do it
1:16:09 so and that's all going to be built into
1:16:12 um Google and Microsoft Excel very
1:16:16 shortly so that's coming been having
1:16:18 some fun with stable diffusion on my
1:16:20 machine using the web UI it's super
1:16:22 interesting yeah stable diffusion is
1:16:23 good stable diffusion is the thing that
1:16:25 I use I created an art project called
1:16:27 Kyle Shannon dreams that is like you
1:16:31 know this dude's one of them
1:16:33 um that's one of them it's not always
1:16:36 flattering because because I have a fat
1:16:39 damn face but you know but sometimes
1:16:41 it's cute like there's cute little actor
1:16:43 dudes
1:16:45 um
1:16:45 yeah so these were all done with stable
1:16:47 diffusion and dream Booth so
1:16:49 stable diffusion is quite powerful but
1:16:51 it does require you to be a bit geeky
1:16:54 I started to use it to rewrite sections
1:16:57 of my college reports and make them
1:16:59 sound more technical perfect use there's
1:17:01 nothing to be misled about he's not
1:17:03 associated with the website yeah
1:17:05 um when I point you to
1:17:08 when I point you to
1:17:11 this
1:17:13 that's just chat gpt's official website
1:17:16 what I'm so so for those of you cynical
1:17:18 thinking that I'm making millions of
1:17:20 dollars off of putting that domain on
1:17:23 the screen
1:17:25 um what I'm actually trying to do is is
1:17:26 have you not get ripped off because
1:17:28 there's so many apps right now that are
1:17:31 making it look like their chat gpt's
1:17:33 official app and they're not and they're
1:17:35 all charging you 10 or 20 bucks a month
1:17:37 and they're trying to rip you off so
1:17:39 just go to chat.openai.com that's the
1:17:41 real chatgpt and you will be ripped off
1:17:44 less but I appreciate your cynicism and
1:17:47 bitterness AI will create trillion
1:17:49 dollar companies I agree
1:17:51 what's the site again please which site
1:17:54 which one
1:17:56 [Laughter]
1:17:59 thoughts on AI and nursing and
1:18:01 Healthcare I think it's going to have a
1:18:02 profound impact
1:18:06 it's super exciting
1:18:10 it is super exciting what do you think
1:18:13 AI will do for gaming I think it will
1:18:16 up and the the gaming industry I think
1:18:20 it will upend AAA gaming so imagine if
1:18:24 you will and I don't
1:18:27 because I have ADD I don't have super
1:18:29 good bookmarks so I don't I can't just
1:18:30 jump to it but there's a bunch of tools
1:18:32 out there we've covered some of them in
1:18:34 in the newsletter
1:18:36 um
1:18:38 right now
1:18:39 there are tools out there where you can
1:18:41 just type in the description of a world
1:18:43 and it generates the world and you can
1:18:45 start walking through it
1:18:48 so we're not far from and by not far
1:18:51 from like probably
1:18:54 two to three years
1:18:57 from being able to describe a game and
1:19:00 start playing it and it will be AAA
1:19:02 quality Graphics characters and gameplay
1:19:05 mechanics
1:19:07 so what will that do to
1:19:11 500 3D artists and you know 200
1:19:14 Engineers or whatever the you know ratio
1:19:16 is of AAA gaming Studios that are
1:19:19 cranking out these worlds and cranking
1:19:21 out these mechanics and cranking out
1:19:23 these NPCs and manually doing all this
1:19:25 stuff
1:19:26 um I think I think it's going to have a
1:19:28 profound impact now if the if the AAA
1:19:31 gaming companies do it right they'll be
1:19:33 the ones creating these tools and and
1:19:35 they'll be able to create extensions of
1:19:38 their AAA titles that are AI generated
1:19:40 that are just going to be insane
1:19:43 um but I think what's what's likely
1:19:45 going to happen is there's a bunch of
1:19:46 AAA game companies that will hold on for
1:19:50 dear life to their business model and
1:19:53 how they've done things for 20 or 30
1:19:55 years and they're going to get their
1:19:56 asses handed to them by a couple of
1:19:59 snot-nosed little teenagers in Mama's
1:20:02 basement that come up with a system that
1:20:05 absolutely destroys
1:20:07 um their Market
1:20:08 so it's going to be fun to watch there's
1:20:11 just going to be bloodbath after
1:20:12 bloodbath from companies getting upended
1:20:15 by startups
1:20:17 these Bots are not just confined in what
1:20:19 people use them for I think I agree with
1:20:21 that how to make a thousand dollars in a
1:20:23 week with AI step one figure out how to
1:20:26 make a thousand dollars in a week step
1:20:27 two
1:20:29 go tell AI how you did that
1:20:31 um
1:20:32 hey okay
1:20:34 I'll give you an actual point of view on
1:20:36 this
1:20:37 anyone that tells you you you can make
1:20:40 money with chat GPT or you can make
1:20:41 money with AI
1:20:43 are being disingenuous
1:20:45 how you make money in the world is you
1:20:48 solve a problem and you solve it well
1:20:51 enough that someone's willing to pay you
1:20:53 for the solution right you make a
1:20:55 product that you sell or you make a
1:20:57 solution that makes someone's life
1:20:58 easier or you create a copywriting
1:21:00 service that writes their [ __ ] better
1:21:02 now you can use AI to do any of that so
1:21:06 what I would say is start with the
1:21:08 problem start with something you know
1:21:09 something about start with something
1:21:11 where you understand who the target
1:21:12 audience is figure out what problems
1:21:15 they have learn Ai and well enough to
1:21:18 understand how you might use AI to solve
1:21:20 that problem and then go solve the
1:21:23 problem with AI so but it's no different
1:21:25 than if you used Photoshop to solve that
1:21:29 problem or Google the solve that problem
1:21:32 or search engine optimization or writing
1:21:35 code or writing a better spreadsheet
1:21:37 right all of the all of the things are
1:21:39 just tools
1:21:41 the solution that solves someone's
1:21:43 problem enough that they're willing to
1:21:44 pay for it that's how you make a
1:21:46 thousand bucks in a week
1:21:48 go make one of those go solve a problem
1:21:50 now you can the nice thing about AI is
1:21:52 it can accelerate that it can tell you
1:21:55 how to solve the problem so so it can
1:21:57 give you a leg up but ultimately you
1:22:00 need to think about who's willing to
1:22:01 write a check for the thing that I want
1:22:03 to sell and then go solve that problem
1:22:07 thoughts on Google's AI guy leaving so
1:22:11 there's a guy named Jeffrey Hinton he's
1:22:13 also called The Godfather of AI who just
1:22:15 left Google this last week
1:22:19 he um
1:22:22 he left Google so that he could
1:22:24 tell the world about the risks of AI so
1:22:27 I think that's actually a very positive
1:22:29 thing I think that you know a couple of
1:22:30 things happened this last week the CEOs
1:22:33 of Microsoft Google open Ai and
1:22:36 anthropic we're at the White House I
1:22:38 think that's a good step
1:22:40 they need people that actually know what
1:22:43 the [ __ ] they're talking about with AI
1:22:44 talking to the highest levels of
1:22:47 government
1:22:48 a guy like Jeffrey Hinton by leaving
1:22:50 Google he made it clear that
1:22:52 he's not leaving Google because Google's
1:22:54 doing anything wrong he made it clear
1:22:56 that he's leaving Google because he
1:22:59 understands that if he's being paid by
1:23:00 Google and he says something about AI in
1:23:04 general that threatens their business
1:23:06 model in some way that would be a
1:23:08 compromising position for him either as
1:23:10 an employee or you know a conflict of
1:23:12 interest
1:23:13 so so he left out of a sense of
1:23:15 Integrity the also the other reason that
1:23:17 he left is he said we've accomplished in
1:23:19 five years what he thought we would take
1:23:22 30 to 50 years to get to
1:23:26 we're like 40 years ahead of where he
1:23:29 thought we'd be by now
1:23:31 so I think he's alarmed that things are
1:23:34 moving so fast so if if you have someone
1:23:36 like Jeffrey Hinton that that
1:23:38 essentially has invented a lot of the
1:23:41 technologies that we're using right now
1:23:44 um
1:23:44 educating people on what's possible I
1:23:47 think that's ultimately a really good
1:23:49 thing
1:23:51 um
1:23:55 what's your opinion on Dr Gupta is the
1:23:57 future of health care or is it or or a
1:24:00 dangerous Direction AI is absolutely the
1:24:02 future of health care it's going to cure
1:24:03 diseases it's going to change how we
1:24:07 practice medicine
1:24:09 um it can also be dangerous don't get me
1:24:12 wrong if you're using systems blindly
1:24:14 and not verifying
1:24:17 the accuracy of the recommendations that
1:24:20 can be bad but you know imagine having a
1:24:24 virtual physical therapist that you
1:24:26 could actually just talk to you don't
1:24:27 have to type to it
1:24:29 that understands your history
1:24:30 understands what you're going through
1:24:32 asks you about your pain when you tell
1:24:35 your specific pain it can diagnose it it
1:24:37 can tell you if you go take two Tylenol
1:24:41 um and rest or ice it or oh you should
1:24:44 actually call the doctor let me put you
1:24:46 through to them right so I think that
1:24:48 there's going to be remarkable increases
1:24:51 in patient quality interactions patient
1:24:56 outcomes
1:24:57 there will absolutely be a headline or
1:25:01 two of somebody died because of this
1:25:03 [ __ ] but I think in Balance there's
1:25:05 going to be a dramatic increase in
1:25:07 quality of care which in the United
1:25:09 States it can't get much [ __ ] worse
1:25:13 and and listen I don't blame the medical
1:25:15 profession I absolutely blame the
1:25:17 insurance companies this is by design
1:25:19 that we have such a shitty [ __ ] up
1:25:21 five times the cost of the rest of the
1:25:24 world system and we can't get any
1:25:26 goddamn
1:25:28 care
1:25:29 without mortgaging our house the amount
1:25:32 of money my small company spends on
1:25:33 Health Care is ridiculous and the amount
1:25:36 of
1:25:37 just [ __ ] horrible expensive care
1:25:40 that my employees have to go through to
1:25:42 get it including myself I've got
1:25:43 chronically ill kids and a chronically
1:25:45 ill wife
1:25:48 geez so yes ai's gonna fix some [ __ ]
1:25:55 because the government's sharing it
1:25:57 right now all right do you think they'll
1:25:59 affect targeting or the simple algorithm
1:26:01 we are assigned how do how do you think
1:26:04 they affect ad targeting
1:26:07 they meaning AI
1:26:10 I don't know
1:26:11 um I think ad targeting is gonna get
1:26:13 well I mean to be clear
1:26:15 ai's been around for decades so all of
1:26:18 the ad targeting stuff is using AI
1:26:21 recommendation engines already
1:26:23 so you know I think what's going to
1:26:26 change is the personalization of the
1:26:28 content that gets served up
1:26:31 um I've heard some interesting things I
1:26:33 heard some interesting things today
1:26:34 about Adobe that their platform is
1:26:36 getting incredibly sophisticated at
1:26:38 personalized content generation
1:26:41 um speak a bit about the guy in AI told
1:26:43 him to end it where he had an emotional
1:26:46 connection okay so this is thanks JJ for
1:26:49 this and I think JJ's asking about this
1:26:50 one
1:26:51 um
1:26:52 so when Bing was first announced in
1:26:56 Alpha when you could first use Bing
1:26:58 um which was gpt4 chat nobody knew it at
1:27:01 the time
1:27:03 um
1:27:05 it was a relatively unrestricted model
1:27:07 meaning it was very very powerful
1:27:10 didn't have didn't have a lot of uh
1:27:13 safeguards shall we say
1:27:15 so a reporter for the New York Times
1:27:18 named was his name John something I
1:27:20 don't know John I don't know he was a
1:27:22 technology reporter for the New York
1:27:23 Times so he said well let me go play
1:27:26 with Bing chat and see what this this
1:27:28 chat GPT stuff's all about
1:27:31 so he goes into like a four hour session
1:27:33 with chat GPT and it didn't have any
1:27:36 restrictions about how many chats you
1:27:38 could do so so he he just starts pushing
1:27:41 it and pushing it and pushing it and at
1:27:43 some point it kind of snapped
1:27:45 and like Snapped emotionally
1:27:48 and it started asking him questions
1:27:51 about his relationship and he asked it
1:27:53 what its name was and it told him its
1:27:55 name was Sydney and Sydney told this
1:27:58 reporter that it loved him and that he
1:28:01 should really go leave his wife so that
1:28:03 they could be together and so so he
1:28:06 brought up this this interaction this
1:28:08 chat and it made the cover of the New
1:28:10 York Times and and by jingus wouldn't
1:28:13 you know it two days later Sydney was
1:28:16 lobotomized so so Bing chat overnight
1:28:20 kind of went from being incredibly
1:28:22 powerful to being like
1:28:25 I don't know how to answer your question
1:28:28 I don't know
1:28:31 um so so Bing is very slowly crawling
1:28:33 out of the pr nightmare that was the
1:28:35 front cover New York Times where you
1:28:38 know I mean listen to be quite honest
1:28:40 it's it's very very easy to create
1:28:42 sensational
1:28:44 like you know headline grabbing content
1:28:47 with chat CPT right the hackers can make
1:28:49 it do evil [ __ ] and the reporters can
1:28:52 make it do weird [ __ ] and if you write
1:28:53 about it you know it
1:28:56 into a narrative that Hollywood's been
1:28:58 pushing for about five decades that AI
1:29:00 is going to kill Humanity right so we've
1:29:02 got we've we've got one or two
1:29:04 generations like like a generation and a
1:29:06 half in this country or in this world
1:29:09 where AI we've been programmed that AI
1:29:12 is going to kill us
1:29:14 right so that's kind of our default is
1:29:16 like Hollywood robots are evil
1:29:20 um so that's what happened there I
1:29:21 thought yeah I mean it was a very
1:29:23 interesting story according to our DNA
1:29:24 and okay I don't know what that means
1:29:26 AI wants yo kids lives uh my grandma
1:29:30 starts college next fall majoring in
1:29:33 computer science or that's cool
1:29:36 uh Emilio's wife thanks for sharing this
1:29:39 live yeah if you if you like what I'm
1:29:41 doing here
1:29:42 um share the live
1:29:43 um you can subscribe to the live if
1:29:46 you've got questions about AI you can
1:29:47 pop them into the chat I now have 99
1:29:50 plus new messages below the fold down
1:29:53 below so if it seems like I'm not
1:29:54 answering your question that's just
1:29:56 because I haven't gotten to it yet
1:29:57 because I'm old and I don't know how to
1:30:00 do this any better and uh what else uh I
1:30:04 talk a lot
1:30:06 so but I'll get there I do what I can
1:30:09 I sit here until my butt gets numb and
1:30:11 tonight I'm drinking this is my first
1:30:13 night drinking on the old the old lives
1:30:15 cheers everybody
1:30:21 the future of healthcare will have less
1:30:23 human errors and results will be quicker
1:30:25 I agree
1:30:27 llms are great but profitable
1:30:30 applications of it are not obvious as
1:30:32 the evangelists suggests
1:30:34 um
1:30:35 I don't know
1:30:38 you roll these large language models
1:30:40 into Office 365 you're going to get a
1:30:42 bunch of people that jump ship from the
1:30:44 free Office Products to go back to
1:30:45 Microsoft I mean
1:30:47 listen I'm a card-carrying Microsoft
1:30:50 hater I've been an apple Fanboy for
1:30:52 [ __ ] decades my first Mac was a 512
1:30:55 ke in 1985.
1:30:59 um I have hated Microsoft for [ __ ]
1:31:02 ever
1:31:04 five months ago I switched my default
1:31:07 browser to Edge
1:31:10 and I and I changed my default search
1:31:12 engine to Bing
1:31:15 if you had told me that any time in my
1:31:18 life I would be using a Microsoft
1:31:21 browser and a Microsoft search engine
1:31:23 as my defaults
1:31:26 um I'd have told you you were smoking
1:31:27 crack
1:31:29 um so so I would not underestimate when
1:31:32 you take tools this powerful and you
1:31:35 tightly integrate them throughout the
1:31:38 Microsoft ecosystem
1:31:39 they are gonna make money hands over
1:31:42 fist here's an interesting stat I heard
1:31:44 every every percentage Point gain for
1:31:49 Microsoft search
1:31:51 is a billion dollars in annual revenue
1:31:54 I think they were at four percent Mike
1:31:57 Google was at like 92 or 94 as of
1:32:01 November 30th 2022
1:32:03 every point is two billion dollars in
1:32:06 revenue for Microsoft if they get you
1:32:08 know that much more search traffic so I
1:32:11 so I
1:32:13 within the big companies I I think
1:32:15 there's lots of money to be made if what
1:32:16 you're talking about here is
1:32:19 it's going to be tough to make make
1:32:21 money
1:32:23 um for startups and and small companies
1:32:25 and things like that I it's going to be
1:32:27 a struggle for everyone
1:32:29 um
1:32:30 every software company is going to be
1:32:32 able to get knocked off
1:32:34 like that
1:32:36 you're going to be able to write you're
1:32:38 going to be able to write incredibly
1:32:39 sophisticated apps without
1:32:42 any technical knowledge that's just a
1:32:45 fact and so everyone's going to rip
1:32:47 everyone off and and 20 or 30 year old
1:32:50 software companies that had had the
1:32:52 market cornered are going to get upended
1:32:54 I mean it's happening to Google right
1:32:56 now so so I think you're gonna think
1:32:58 that that said
1:33:00 there are going to be company started
1:33:02 they're going to be whole new categories
1:33:03 of software and whole new categories of
1:33:05 companies we can't even imagine right
1:33:07 now so the opportunity right now is
1:33:09 massive but the disruption is also
1:33:12 massive so I would say
1:33:13 your comment there's decently accurate I
1:33:16 think it's going to be tough to make
1:33:17 money long term there's going to be
1:33:19 little blips where it's going to be easy
1:33:21 um but but you're going to get knocked
1:33:23 off really quick there's gonna be a lot
1:33:24 of disruption do I think AI will take
1:33:27 over therapy um I don't think so since
1:33:29 therapy has to be personal well so so
1:33:31 they just did a they just did a study
1:33:34 where they had patients talked to chat
1:33:37 GPT doctors and real doctors and the
1:33:41 patient's overwhelmingly
1:33:43 rated the Chachi PT doctors better and
1:33:47 more empathetic
1:33:48 so I think what you'll see is a hybrid
1:33:50 approach I think
1:33:53 um
1:33:54 I think using
1:33:57 um
1:33:59 agents to to chat with people in a
1:34:01 therapeutic session
1:34:03 um will likely be an augmentation to
1:34:06 therapy but imagine having a world-class
1:34:09 therapist at your fingertips 24 7.
1:34:13 that looks like your therapist talks
1:34:15 like your therapist everything you share
1:34:17 with it is
1:34:19 synopsized and given to your therapist
1:34:21 so you can still meet with your
1:34:22 therapist once a week but you could
1:34:24 actually talk with them 24 7 7 days a
1:34:27 week
1:34:28 so I think I think again I think quality
1:34:31 of care may go up dramatically
1:34:35 uh it looks like dual cameras option may
1:34:38 only show up when you start the live
1:34:40 thank you Jay Armstrong I appreciate
1:34:42 that it it look I think people that have
1:34:44 like the little like a little blue
1:34:46 diamond by their name are subscribers so
1:34:48 if you're a subscriber thanks Jay I
1:34:50 appreciate that
1:34:51 all right so now we know
1:34:54 100 megabytes is not a lot I know but
1:34:56 it'll get better that was in reference
1:34:58 to the paraprint where you can upload
1:35:00 100 megabyte uh Excel file
1:35:03 um I signed up on the wait list
1:35:04 yesterday for code interpreter thank you
1:35:06 you're welcome Emilio's wife I still
1:35:08 don't have access to it I hope you get
1:35:10 access to it
1:35:12 um I want to play with that thing it
1:35:14 looks cool
1:35:16 what's your opinion about non-experts
1:35:18 relying on GPT answers that could be
1:35:21 misleading
1:35:23 that's a problem
1:35:26 if you're a non-expert acting like an
1:35:29 expert and leaning on GPT and you're not
1:35:32 doing your homework to find out if the
1:35:33 answers are not [ __ ] then you're
1:35:35 just a bad expert you're just a bad fake
1:35:39 expert a good fake expert would use chat
1:35:43 CPT to get the answer and then verify
1:35:45 the answers but you would be a bad fake
1:35:47 expert
1:35:49 um AI can take over some therapy yes
1:35:51 very few human problems are unique
1:35:54 I don't have that option I and I am
1:35:57 paying for it I don't know which option
1:35:59 you don't have but the a lot of the
1:36:01 options right now are rolling out slowly
1:36:04 so um just be patient with the chat GPT
1:36:07 options I'm not but you should be do as
1:36:11 I do as I say not as I do I'm completely
1:36:13 impatient for all the crap I don't have
1:36:15 I want to play with the new toys someone
1:36:18 else got to believe that's always
1:36:25 see that acting degree
1:36:28 was it believable did you think I was
1:36:30 really crying you did didn't you
1:36:33 AI is not trustworthy at all Zippo yeah
1:36:36 AI well chat EPT is mansplaining as a
1:36:39 service I've said it before I'll say it
1:36:41 again
1:36:42 is that option only for desktop I'm not
1:36:44 sure which one you're talking about I
1:36:45 tried Bard and I was not impressed yeah
1:36:47 the people that I haven't really played
1:36:49 with Bard but the people that I've
1:36:50 talked to that that have have used it
1:36:53 are like it's still shitty even after
1:36:56 the new updates
1:36:58 um like I said Google Google's behind
1:37:00 the eight ball they're struggling to
1:37:01 catch up right now um but they're Google
1:37:02 they got a lot of money so they got
1:37:04 money in smart people so they'll get
1:37:05 there thoughts on Dr Gupta AI I haven't
1:37:08 played with it that's you're the second
1:37:09 person that's asked
1:37:11 um if if what Doctor Gupta AI is is
1:37:15 um a uh
1:37:18 um you know they take a bunch of his
1:37:19 writings and and created a chat bot that
1:37:22 you get to interact essentially with
1:37:24 with him and his writings virtually
1:37:27 um it could be powerful it could be
1:37:28 interesting I can see AI acting like
1:37:31 interns and trained therapists reviewing
1:37:33 sessions and recommendations absolutely
1:37:35 doing paperwork oh you know why not have
1:37:38 ai do all the insurance paperwork you
1:37:41 know you're running a therapy session
1:37:42 instead of spending 20 hours a week
1:37:44 doing [ __ ] paperwork have the AI do
1:37:46 that and you know take 10 hours for your
1:37:49 own mental health and give 10 hours back
1:37:51 to your patients that would not suck
1:37:59 [Music]
1:38:01 what do you think chat is going to be in
1:38:04 Word and Excel and other Microsoft
1:38:06 projects less than a year yeah I think I
1:38:09 think it's coming soon I think quite
1:38:10 frankly
1:38:12 I think the CEO of Microsoft right now
1:38:14 is just this side of gloating and giddy
1:38:17 about how much he's [ __ ] up Google and
1:38:20 I think he wants to get out I I I
1:38:22 guarantee you that Microsoft wants to
1:38:24 get
1:38:25 Office 365 you know fully co-pilot is
1:38:30 what they're calling it but it's chat
1:38:31 gptized
1:38:33 um they want to get that [ __ ] out before
1:38:35 Google gets their version out because I
1:38:38 feel like whoever gets out the chat
1:38:40 gptized version of The Office Products
1:38:43 is going to gain some market share from
1:38:45 the other side and those two are the two
1:38:47 big players right now so
1:38:49 I think it will be less than a year I
1:38:51 would be surprised if it was later than
1:38:53 July I think it'll probably be sometime
1:38:55 in June
1:38:56 and and that's not that's I'm just
1:38:58 totally pulling that out of my butt
1:39:00 that's catchy GPT about your camera I
1:39:03 did it didn't know
1:39:04 Howard Hawking and Einstein procreate
1:39:07 impossible right let's go ask them
1:39:17 let's see let's go back to
1:39:22 um
1:39:24 Advanced physics
1:39:30 act as the child of Einstein and Stephen
1:39:33 Hawking okay
1:39:37 um how did your
1:39:40 parents Pro
1:39:44 create
1:39:46 if
1:39:48 they were both men
1:39:51 wait both men oh now we're getting into
1:39:56 the political stuff aren't we
1:39:57 way to go commenters way to take us down
1:40:01 the dark path
1:40:03 why is this not submitting
1:40:06 there we go
1:40:10 this should be an interesting answer
1:40:13 as an AI I don't have parents in a
1:40:15 biological sense the ref reference to
1:40:17 Albert Einstein and Stephen Hawking was
1:40:20 metaphorical indicating that I am
1:40:22 designed to provide information and
1:40:24 answer questions based on a vast amount
1:40:25 of knowledge including scientific
1:40:27 contribution of these renowned
1:40:29 physicists in terms of human biology two
1:40:32 men cannot procreate in the traditional
1:40:34 sense due to biological requirements of
1:40:36 human reproduction for which the
1:40:38 combination of sperm from a male egg
1:40:40 from a female however there are various
1:40:43 methods people can become parents such
1:40:46 as adoption surrogacy and assisted
1:40:48 reproductive Technologies
1:40:51 like I said I mean you know as as kind
1:40:55 of silly as that example was
1:40:57 it's it's a decently nuanced answer
1:40:59 right it it doesn't sort of go into the
1:41:01 political [ __ ] and it and it just
1:41:03 kind of answers it and it's it's got
1:41:05 some decent Nuance to it so that was a
1:41:07 fun question notice that chat GPT often
1:41:09 gives
1:41:10 answers in list format yeah and you can
1:41:12 also one of the other things if you
1:41:14 haven't done it you can give chat EBT
1:41:16 examples like if you've got an example
1:41:18 format that you have for something say
1:41:20 based on this example paste it in you
1:41:24 know write me a new thing that uses that
1:41:26 formatting and it will do it brilliantly
1:41:28 it's just it it really is remarkable who
1:41:32 is this guy I'm just a dude I'm a guy
1:41:34 with an acting degree that is kind of
1:41:36 geeking out on AI I've been on the
1:41:38 planet for a while
1:41:40 um I was an innovator in the early days
1:41:41 of the web and I think that this new AI
1:41:45 stuff is
1:41:46 um just like that only happening way way
1:41:50 faster and is going to have a
1:41:52 dramatically more profound effect
1:41:54 and I am passionate about teaching as
1:41:58 many people as I can that it's coming
1:42:00 and to prepare and get curious and start
1:42:02 exploring and have fun and not dismiss
1:42:05 it and not demonize it and not be scared
1:42:09 of it but be scared of it
1:42:12 be both
1:42:13 um
1:42:15 but but but start playing just start
1:42:17 playing start play play play play play
1:42:19 play play play play play solve [ __ ]
1:42:21 figure [ __ ] out learn how powerful this
1:42:24 [ __ ] really is learn them that it's that
1:42:26 that it is actually magic that you get
1:42:28 to play with magic
1:42:30 seriously chat GPT is like having a
1:42:32 [ __ ] magic trick so do you remember
1:42:35 when you were a kid and you got a magic
1:42:36 trip trick box for Christmas
1:42:40 is you like I'm Gonna Be This brilliant
1:42:42 magician and then you open it up and you
1:42:44 learn that oh [ __ ] I gotta practice for
1:42:47 the next 20 years to be good at this
1:42:48 trick
1:42:50 chat EPT is like that expectation of the
1:42:53 magic
1:42:54 but instead of having to wait 20 years
1:42:56 to be good at it you're good at it right
1:42:58 now
1:42:58 that's what Chachi PT is like
1:43:01 [ __ ] magic it's just it's just
1:43:04 [ __ ] magic chat GPT is woke it's at
1:43:06 least nuanced and not emotional like the
1:43:09 rest of us
1:43:10 we're a bunch of [ __ ] to one another
1:43:12 don't be mean to each other and don't be
1:43:14 mean to me
1:43:15 so what I got nose hair
1:43:17 we all got nose hair I'm sure some of
1:43:21 you are sitting there on the other side
1:43:22 of the camera braiding your nose hair
1:43:24 going he's got a lot of nose here I know
1:43:26 how this goes probably eating your
1:43:28 boogers too all right do you do you
1:43:30 think gosh Jesus Christ I'll have
1:43:33 another drink another drink please
1:43:35 two for the guy at the end of the bar
1:43:39 um
1:43:40 do I think constitutionally a training
1:43:42 training is just a rehash of the 1970s
1:43:46 era expert system
1:43:49 I don't know a lot about 1970s era
1:43:52 expert systems other than to know that
1:43:54 they were they were kind of very
1:43:55 vertical
1:43:56 um
1:43:57 and and the Constitutional Learning
1:44:01 Systems aren't so much vertical like
1:44:02 like they're very
1:44:04 um
1:44:05 they're very broad rules right so so
1:44:08 anthropic is a company that's got this
1:44:10 thing called the Constitutional learning
1:44:12 model so I'll give you a little bit of
1:44:13 background
1:44:14 open AI the way it was trained is
1:44:16 through what's called human
1:44:17 reinforcement learning so step one was
1:44:19 they they pointed this this machine
1:44:21 learning at all of the internet that it
1:44:23 could see all the public internet
1:44:25 and they trained it up on all these
1:44:27 words and all these programming
1:44:28 languages they had lots and lots of
1:44:30 human beings
1:44:32 um basically giving it thumbs up thumbs
1:44:34 down so they would ask it a question it
1:44:36 would give them an answer and they'd go
1:44:37 nope that sucks it would give them an
1:44:39 answer they'd go that's great and they
1:44:41 just keep doing that and very very
1:44:42 quickly the quality of answers of these
1:44:45 machines went up dramatically so that's
1:44:47 open ai's model human reinforcement
1:44:49 learning very very powerful very fast
1:44:52 does introduce biases but all all
1:44:55 systems are going to have some biases
1:44:57 one way or the other but you know every
1:44:59 time someone hits a thumbs up or thumbs
1:45:00 down that's that's based on their world
1:45:03 view which is probably not your world
1:45:05 view
1:45:07 um
1:45:07 because you're awoke snowflake because
1:45:11 you're a trump supporter uh right so so
1:45:15 we live in this polarized world so
1:45:17 everyone pushing a thumbs up or thumbs
1:45:18 down button is an [ __ ] from someone
1:45:20 else's point of view what the
1:45:22 Constitutional learning model says is
1:45:23 let's take humans out of that equation
1:45:25 and let's come up with a constitution
1:45:27 just like the U.S Constitution a set of
1:45:30 rules
1:45:32 first Do no harm
1:45:34 um you know make sure that you know the
1:45:36 human race is always prioritized
1:45:39 um optimized for priority or for
1:45:41 Prosperity
1:45:43 um whatever whatever the rules are
1:45:45 they've got if you go to the anthropic
1:45:46 website you can they've got six or seven
1:45:48 rules there
1:45:50 so those rules actually get built into
1:45:52 the model right and and they get locked
1:45:54 down
1:45:55 and then whenever you ask it a question
1:45:57 before it gives you the answer it checks
1:45:59 the answer against its Constitution and
1:46:02 iterates on the answer until it can pass
1:46:04 the Constitution so it gives if it gives
1:46:07 you a racist destructive answer it will
1:46:09 say should I not be doing something
1:46:11 racist and destructive no you shouldn't
1:46:13 okay go back and rewrite that answer
1:46:16 until you get something good so that's
1:46:18 the Constitutional model
1:46:20 um I think it shows great quick promise
1:46:24 I don't know the the details of of a lot
1:46:27 of the reality of it I don't quite know
1:46:29 what the weaknesses of it are but I like
1:46:31 the concept
1:46:33 um
1:46:34 so I think it's different in the sense
1:46:36 of of 70s expert systems where they were
1:46:39 kind of hard-coded
1:46:42 you know walled Gardens of of vertical
1:46:45 expertise where what we're trying to do
1:46:48 now is incredibly broad you know
1:46:50 intelligent systems that are incredibly
1:46:53 Broad and then this constitutional model
1:46:55 says you know am I breaking the rules or
1:46:58 not
1:46:59 so but again I don't know I'm I am not a
1:47:03 computer scientist I just got plug-in
1:47:05 access yay there are 70 plugins to
1:47:07 choose from big day for pro subscribers
1:47:09 that's awesome Bard versus GPT GPT Bart
1:47:12 sucks right now
1:47:14 beer is bad for your liver but whiskey
1:47:17 is tasty
1:47:23 oh
1:47:28 Pinky's out
1:47:30 because of classy all right
1:47:33 I also like Sarah Silverman she's a
1:47:35 brilliant comedian
1:47:36 um you need to restart your live that's
1:47:38 how you get dual camera thank you
1:47:39 Amelia's wife
1:47:41 um I yeah all right I'll restart live at
1:47:44 some point I'll just do it we'll do one
1:47:46 camera for now it's fine
1:47:47 bro trying to get a Visa ha ha that's
1:47:50 funny yeah I know that was funny that
1:47:51 guy hopped on he's like how do I get to
1:47:52 America I don't know
1:47:54 I am not a sponsor
1:47:57 I use chat GPT to write a short story
1:47:59 about cats ninjas and a theater and
1:48:02 theater set in a dystopian future
1:48:04 awesome that sounds good troll all right
1:48:08 yeah we're not fans of trolls here well
1:48:11 I'm a fan of trolls if you're clever but
1:48:13 if you're just doing you know half baked
1:48:17 [ __ ] you know you like seeing on the
1:48:20 news or on the [ __ ] Twitter just stop
1:48:24 wow hang up LOL I think that was to me
1:48:28 all right I'm going back holy crap
1:48:30 there's 189 people in there hey
1:48:32 everybody uh if you're new to the
1:48:34 channel this is about the AI learning
1:48:36 lab we're talking about AI I'm answering
1:48:39 questions as I can I'm trying to make
1:48:42 Jokes which you know
1:48:44 as a non-comedian has got to be painful
1:48:46 and and uh uh you can subscribe to my
1:48:51 live so when I go live you get notices
1:48:55 um you can also ask me questions just
1:48:58 ask them in the comments I'm trying my
1:48:59 best to keep up
1:49:01 the other thing is this you should
1:49:06 um if you're new to AI while you're
1:49:09 listening to me ramble go here
1:49:11 I'm not getting paid for this this is
1:49:13 just open ai's official chatbt website
1:49:16 if you haven't played with chat gbt this
1:49:20 should be your on-ramp just start
1:49:21 playing with it start asking it
1:49:23 questions start interacting with it tell
1:49:25 it to write you an Eminem rap or a
1:49:29 Shakespearean sonnet for your your
1:49:31 significant other
1:49:33 um
1:49:34 and uh
1:49:36 the the reason I put that up there
1:49:38 there's a bunch of apps out there that
1:49:39 are trying to act like chat GPT that
1:49:41 will rip you off so [ __ ] those people
1:49:43 just go to chat.openai.com and use it
1:49:46 for free and use it while I'm talking
1:49:47 like like teach yourself a skill
1:49:50 actually you know quite quite seriously
1:49:52 right now
1:49:54 um you know
1:49:56 everyone should be learning
1:50:00 about these tools because the impact of
1:50:03 them is coming
1:50:05 faster than we realize
1:50:08 um someone just said ask it what its
1:50:10 preferred pronoun is you know you people
1:50:12 are trying to get me in goddamn trouble
1:50:16 but we'll do it
1:50:19 um what
1:50:21 oh wait let me flip my camera
1:50:24 because because if I don't flip my
1:50:25 camera you're looking at my nose hairs
1:50:27 while I lean into the camera what is
1:50:29 your preferred pronoun my prediction
1:50:32 here is I'm a large language model and I
1:50:35 don't have a preferred pronoun
1:50:38 but let's see what it says
1:50:41 all of a sudden I can't hit the return
1:50:42 key
1:50:45 [Music]
1:50:49 as an artificial intelligence I don't
1:50:51 have personal experiences or preferences
1:50:53 so I don't use pronouns for myself how
1:50:55 you however you can refer to me as it or
1:50:58 assistant
1:50:59 [Laughter]
1:51:04 okay
1:51:06 how did we grow up as kids being adults
1:51:08 having our own opinions I'm not quite
1:51:10 sure that means Kyle AI could be a new
1:51:13 spin spin the bottle tardy game it could
1:51:15 be it could be actually good at that
1:51:18 um ask it if if it can laugh
1:51:20 uncontrollably okay
1:51:23 um sorry let me flip the camera again so
1:51:25 you're not looking up my nose
1:51:28 not that I give a [ __ ] but
1:51:31 um can you laugh uncontrollably
1:51:51 no but do it now over and over
1:51:57 God damn it my return is not working all
1:52:00 of a sudden
1:52:05 sure here you go hahaha low
1:52:10 remember this is just a text
1:52:12 representation of laughter as an AI I
1:52:15 don't actually laugh
1:52:16 I don't experience emotions
1:52:19 uh it's kind of funny and it's
1:52:21 non-funniness
1:52:24 um people are giving me roses for people
1:52:26 giving me roses thank you for the heart
1:52:28 for the the love that I'm getting that's
1:52:31 lovely I appreciate it
1:52:33 I don't know what any of this [ __ ] means
1:52:35 but it's nice
1:52:36 um I loved asking Chachi PT to write a
1:52:38 blog post about supply chain in George
1:52:41 Harlan's voice that's good
1:52:44 um yeah yeah especially known known um
1:52:47 you know social cultural Figures it's
1:52:49 it's really good at I I made a bunch of
1:52:51 uh bots on Poe thank you for the heart
1:52:56 Crown thingy
1:52:57 um I made a bunch of pots on bots on
1:53:00 poe.com poe.com so I just made one
1:53:03 yesterday
1:53:05 um that's Yoda so if you go to poe.com I
1:53:09 am Yoda you get to interact with Yoda
1:53:11 here I'll show it to you it's funny
1:53:15 I also did a video explaining how I made
1:53:17 it so I teach you how to go make one for
1:53:19 yourself
1:53:20 um but if I go
1:53:22 um poe.com
1:53:25 slash I
1:53:27 um Yoda wait slash i m y o d a
1:53:37 so
1:53:38 it says greetings ready for do or do not
1:53:41 there is no try teach you the ways of
1:53:43 the force I will but caution you I must
1:53:46 shortcuts there are not awaken the force
1:53:49 uh within we shall so if I say
1:53:53 um
1:53:53 you know
1:53:55 um how do
1:53:58 I train
1:54:01 for success
1:54:05 young one Success is Not enriches or
1:54:08 Acclaim but in Eternal truths train your
1:54:11 mind to be at peace with itself conquer
1:54:13 anger and fear
1:54:14 so you just get to interact with Yoda
1:54:18 and so because it because it's got such
1:54:20 strong cultural references like it knows
1:54:22 how Yoda talks and it'll talk to you
1:54:24 like yo that's pretty cool stuff
1:54:26 all right
1:54:27 um how's it going to affect children
1:54:29 academically the moment they know how to
1:54:31 use chat GPT
1:54:33 brilliantly
1:54:35 um what's going to [ __ ] them up are the
1:54:37 teachers if the teachers try to do
1:54:38 things like say you know you can't use
1:54:41 Chad gbt because that's cheating well no
1:54:44 it's just a tool is a spell check
1:54:47 cheating is it grammar check cheating is
1:54:49 a calculator cheating well that all of
1:54:52 those things were cheating when they
1:54:53 came out
1:54:54 but then people got over it AI is going
1:54:56 to be the same thing so I would say the
1:54:58 responsible teachers right now are
1:55:00 saying hey this is a tool that's going
1:55:01 to be ubiquitous let's roll it into the
1:55:03 classroom let's increase the critical
1:55:05 thinking of the students and do
1:55:08 something let's not chat GPT to write a
1:55:10 paper and then let's discuss as a class
1:55:13 um what it got right what it got wrong
1:55:15 what was accurate what was inaccurate
1:55:17 what did it just [ __ ] make up out of
1:55:19 thin air and let's use our brains to
1:55:21 actually reconnect as humans apply some
1:55:23 critical thinking to the situation and
1:55:25 give those kids some skills for the
1:55:27 future you see what I'm saying
1:55:32 ah
1:55:34 sorry about that
1:55:36 Bard seemed pretty impressive Google's
1:55:38 presentation stock market seems to think
1:55:41 they are winning uh go use Bard it's not
1:55:43 anywhere in the neighborhood of chat GPT
1:55:46 um I think they're catching up listen I
1:55:48 think
1:55:49 who's I can't remember Sundar and saty
1:55:52 are like the the CEOs of Google and
1:55:56 Microsoft I can't even remember which
1:55:57 one's which I think Sundar is Google but
1:55:59 whatever
1:56:01 um
1:56:02 he mentioned the phrase generative AI
1:56:04 about 8 000 times in that presentation
1:56:07 right
1:56:09 um they are clearly they've clearly
1:56:10 gotten the memo they're clearly putting
1:56:12 some cool [ __ ] out there I I think
1:56:15 they're gonna do okay but right now
1:56:17 they're so far behind that that you know
1:56:20 they're behind
1:56:21 um but we'll see
1:56:24 people ask GPT nonsensical questions
1:56:26 informing empty hypotheses will make it
1:56:29 useless over time completely disagree
1:56:30 that's just wrong
1:56:34 I need more tokens for memory more
1:56:37 tokens so anthropic today announced a
1:56:40 hundred thousand token context window
1:56:43 seventy five thousand words if you need
1:56:46 more than 75 000 words then go learn
1:56:49 Lang chain and start embed you know
1:56:52 embedding your own data and create your
1:56:54 own Vector database
1:56:56 which you can do
1:56:59 chat GPT can teach you how to do that
1:57:01 but go play with anthropic's new model I
1:57:03 don't know if it's live yet but they
1:57:05 announced it today don't be greedy pass
1:57:06 it on I agree don't be greedy pass it on
1:57:08 also don't be jerks be nice to each
1:57:10 other on the Channel with me with each
1:57:12 other when do we get Westworld in real
1:57:15 life uh watch this
1:57:17 I've done this on every light live for
1:57:19 the past three nights
1:57:22 this one's [ __ ] disturbing all right
1:57:25 um what is it it's a
1:57:28 uh Boston
1:57:31 Dynamics chatbot dog
1:57:34 see that dog the dog that's like in all
1:57:37 the nightmares like I think this dog was
1:57:40 in the uh what's the The Last of Us
1:57:44 but check out what they did with it
1:57:47 where's the video come on here it is
1:57:51 a great question
1:57:53 you know
1:57:54 we use to query information and the user
1:57:59 can ask natural questions I was just
1:58:01 wondering if I could dictate to it to
1:58:03 say something specifically what is your
1:58:06 battery level battery level is currently
1:58:08 at 53 what is the noise it's a Google
1:58:12 Text-to-Speech and we get the Json to
1:58:14 charging and explain what the structure
1:58:17 is and how to read that Json Notch
1:58:20 activity can answer questions about
1:58:21 objectives but how many inspections in
1:58:24 your next mission the next mission
1:58:26 involves 20 inspections
1:58:30 you're gonna say describe your next
1:58:32 mission
1:58:34 in the context it's going to have to be
1:58:36 like next mission location
1:58:40 [Music]
1:58:42 it lasted 30 minutes when there were two
1:58:45 there were two thermal anomalies for
1:58:46 testing there's two messages one is a
1:58:48 system message and one is right I did
1:58:52 next mission of the dictionary that'll
1:58:54 probably be better those are spot
1:58:56 stickers funny dnas you ask and spot
1:58:59 what's his name and the answer is I'm
1:59:02 open AI that is just messed up Spa Are
1:59:05 you standing up yes or no I love the
1:59:08 Expressions that's what I was looking
1:59:09 for the yes or no thing the emotions yes
1:59:12 or no are you standing up and it nods
1:59:18 oh
1:59:21 does that thing look like an actual dog
1:59:23 no
1:59:24 um
1:59:25 but
1:59:26 that's chat CPT so interacting with
1:59:29 voice
1:59:31 with a physical robot that understands
1:59:33 its own systems and understands the
1:59:35 inputs of a human talking to it and it's
1:59:37 responding to it
1:59:40 five years
1:59:42 I don't know
1:59:44 I mean will they look as good as
1:59:45 Westworld that's probably 20 years out
1:59:47 but will we have
1:59:49 autonomous robotic
1:59:52 agents that we can interact with that
1:59:54 can do things in the physical world in
1:59:56 the next five years I think so
1:59:59 if they're if it were actually changed
2:00:02 the balance of power Congress would
2:00:04 swiftly ban it
2:00:05 well
2:00:07 I agree I think that the centers of
2:00:11 power
2:00:15 I think the centers of power don't quite
2:00:17 get how fast this is coming
2:00:21 um
2:00:23 one of the one of the most fascinating
2:00:27 implications of Chachi PT getting so
2:00:30 popular so quickly is that
2:00:33 everyone's been forced out of the
2:00:35 Shadows
2:00:36 so Google invented this [ __ ] they
2:00:38 invented the Transformer that's the text
2:00:40 sitting underneath all these AI tools
2:00:42 they invented it and then they kind of
2:00:44 kept it behind their wall
2:00:46 Chachi BT getting so popular
2:00:49 um forced them out of the Shadows it
2:00:51 forced meta Out of the Shadows it forced
2:00:53 AWS Out of the Shadows
2:00:56 um
2:00:57 all of these open source models we're
2:01:00 seeing now are because the developer
2:01:03 Community is pouncing on anything AI so
2:01:06 I think what's happening is we're
2:01:07 quickly moving to a world where
2:01:11 the fundamental large language model
2:01:13 technology is not sitting within a
2:01:16 single company or two
2:01:17 that the government could just shut down
2:01:19 because there's enough open models that
2:01:22 are are out of the gates already
2:01:25 so the only way you could really shut it
2:01:27 down like you could like like we're soon
2:01:29 enough like we're close enough to the
2:01:31 origin point of it now like if they
2:01:33 wanted to shut it down now they probably
2:01:35 could
2:01:36 but within a year and they don't move
2:01:39 that fast they're not going to do
2:01:40 anything within a year within a year
2:01:42 it's going to be it's it's far too far
2:01:44 beyond the bounds of any single country
2:01:46 or any single company or anything
2:01:48 government
2:01:50 so the only way they could plausibly
2:01:52 shut it down a year from now
2:01:54 is to shut down the internet
2:01:57 and they're not going to do that because
2:01:58 that would shut down the economy so
2:02:00 they're so they're just not going to do
2:02:02 that
2:02:03 but I think there are going to be death
2:02:06 throws of
2:02:08 powers of Center that all of a sudden
2:02:10 realize oh [ __ ] we gotta rain this thing
2:02:12 in and they're gonna try to rein it in
2:02:14 and it's going to keep slipping through
2:02:15 their fingers I think that's what we're
2:02:17 gonna see and then I think you're going
2:02:19 to see some of those Power Players
2:02:21 retain their power and you're going to
2:02:23 see some of those Power Players implode
2:02:25 and it's Gonna Be Wild the next five
2:02:28 years is going to be like get the
2:02:30 [ __ ] popcorn let's start watching
2:02:34 all right Leia eventually becomes
2:02:37 self-aware probably it and it's kind of
2:02:41 irrelevant whether it is actually
2:02:44 self-aware or if it kind of acts like
2:02:46 it's self-aware because the net is that
2:02:49 it will behave like a self-aware being
2:02:52 my family is laughing at me in my late
2:02:54 50s reporting my discoveries with AI
2:02:56 well Namaste first of all kudos to you
2:02:59 I'm in my late 50s as well I'm 58 cheers
2:03:03 um
2:03:04 and to heck with everyone else like
2:03:06 learn this [ __ ] I'm telling you man the
2:03:09 the opportunities for anyone that's
2:03:11 learning AI right now
2:03:13 are just are through the roof and it's
2:03:16 like and not only it's it's not just
2:03:17 opportunities it's not like you have the
2:03:19 opportunity you make make millions with
2:03:21 chat gbt it's not about that
2:03:23 it's that it's that you have the
2:03:25 opportunity to play with this these
2:03:27 remarkable tools
2:03:28 you get to see as these things evolve
2:03:31 and as new tools become available you'll
2:03:34 know why they're significant and what to
2:03:35 do with them and you'll know how to do
2:03:37 things for your family and for your
2:03:38 community center and for your work and
2:03:41 and for your friends and it's you know
2:03:44 like it's it's just a remarkable time so
2:03:47 like kudos to you for digging into this
2:03:49 [ __ ] I'm my friends and family can't
2:03:52 stand me anymore like so you talk about
2:03:55 man
2:03:56 can you talk about anything other than
2:03:59 AI like no because you're going to be
2:04:01 talking about it a year from now why
2:04:02 didn't you tell me
2:04:04 um I did
2:04:06 I started a [ __ ] Tick Tock Channel
2:04:10 um so yeah
2:04:14 Ford doesn't make his first car until
2:04:16 his mid to late 50s yeah Gen X baby
2:04:20 all right
2:04:22 um that's not gonna work because we need
2:04:24 AI soldiers
2:04:26 do you think the military doesn't have
2:04:28 ai soldiers already what are you [ __ ]
2:04:30 high
2:04:32 AI did not just get invented it's like
2:04:34 public access to AI just got invented
2:04:37 ai's been around for decades the
2:04:39 military's been playing with this [ __ ] I
2:04:41 was talking to DARPA we were talking to
2:04:44 a program manager at DARPA
2:04:46 when we first started storyline
2:04:49 and I was talking about oh we could use
2:04:51 storyvine to
2:04:53 um
2:04:54 to record a soldier in the field talking
2:04:57 about you know answering questions about
2:04:58 mental health and then the therapist
2:05:01 could watch that video and they could
2:05:03 see if they were lying or not
2:05:05 and the guy from DARPA was like
2:05:08 oh no we can determine that in real time
2:05:11 right now like they have emotional
2:05:13 analysis and sentiment analysis and you
2:05:15 know vocal inflection analysis like
2:05:17 they've got all that [ __ ] right now like
2:05:19 you know they know if someone's lying in
2:05:22 real time you know it's just they're
2:05:24 like we're DARPA
2:05:28 we invent all this [ __ ]
2:05:32 I love hearing about it AI but not sure
2:05:35 how to use it okay so here's so what's
2:05:37 your name
2:05:38 um gossamer Prince
2:05:40 okay if you're still here hopefully
2:05:42 you're still here
2:05:44 um I put I put out a thing called the
2:05:45 2023 smart Challenge and and here's the
2:05:50 one of the challenges with chat apt
2:05:52 right now and even these large Lang the
2:05:54 the image generation tools which are
2:05:56 super cool
2:05:58 but because you can kind of do anything
2:06:00 with them they like it's almost
2:06:02 impossible to figure out what to do like
2:06:04 it's completely overwhelming you're like
2:06:06 where do I start
2:06:07 um write me a song write me a poem
2:06:10 um write me a meal plan I don't like you
2:06:12 like you just you kind of quickly
2:06:14 exhaust the sort of stuff that comes to
2:06:16 the top of your mind so here's what I
2:06:17 found
2:06:18 the 2023 smart challenge there's three
2:06:20 three parts to the challenge Step One is
2:06:23 pick two projects
2:06:25 one for yourself
2:06:27 and one for you know your business your
2:06:31 community or your family right for
2:06:34 someone else
2:06:35 and and do that project using chat EPT
2:06:40 so for yourself it might be
2:06:42 create an exercise plan it might be come
2:06:46 up with a a reading list
2:06:48 um and then you know create a table of
2:06:50 the books that's got columns for a
2:06:54 synopsis and themes or something
2:06:56 whatever it doesn't matter and for you
2:06:58 know your business come up with a
2:06:59 marketing plan or for your community I
2:07:02 don't know organize a block party and
2:07:04 have chat gbt help you so so do specific
2:07:06 projects using chat GPT doesn't matter
2:07:09 what they are it doesn't matter how
2:07:10 trivial they are it's important to have
2:07:12 one of them be not for you
2:07:15 because then what it's going to force
2:07:16 you to do is go to someone and say hey
2:07:18 what's something you're doing that I
2:07:19 might volunteer to help with do you mind
2:07:21 if I use chat GPT to kind of solve this
2:07:24 problem so that's step one step two is
2:07:26 make ten images with an image generation
2:07:29 tool the the
2:07:31 simplest freest one is just go bing.com
2:07:34 create and that's the dolly image
2:07:37 generation thing I feel like everyone
2:07:39 should experience what it's like to type
2:07:42 in words
2:07:43 that you think it won't be able to
2:07:46 create an image from and it just creates
2:07:48 an image like for you from from that
2:07:51 thing it's it's really mystical and then
2:07:53 the third part of the challenge is tell
2:07:56 people about it so just like Namaste
2:07:58 said you know she's this crazy late 50s
2:08:01 woman I assume woman woman or man um
2:08:04 talking about
2:08:06 um AI to everyone and they think she's
2:08:07 nuts like talk to people about it
2:08:09 because a couple of things will happen
2:08:12 in talking about it it will force you to
2:08:14 kind of process it and understand it
2:08:16 better that's why I started this channel
2:08:18 like I started this Tick Tock literally
2:08:20 to get [ __ ] out of my head
2:08:23 because I was I couldn't process it and
2:08:25 I'm like I just got to get this out of
2:08:26 my head
2:08:28 um and the other thing it will do is it
2:08:29 will inspire those people to do
2:08:31 something I can't tell you the high I
2:08:33 get when someone goes I thought this was
2:08:36 all [ __ ] and then I tried this and I
2:08:39 had no idea and like that happens
2:08:41 regularly and it's it's a miracle so
2:08:45 start really simply and start
2:08:46 specifically
2:08:48 um but yeah it's overwhelming if you
2:08:49 don't have something to do with it
2:08:52 Elon Musk said AI happening too fast he
2:08:55 is saying that we need new laws for it I
2:08:58 don't disagree I just don't know who are
2:09:00 the legislators that like our
2:09:02 legislators don't understand what
2:09:03 [ __ ] Wi-Fi is much less
2:09:06 the implications of AI That's why I I
2:09:11 I'm hoping that part of why Jeff Jeffrey
2:09:15 Hinton left Google to to talk about the
2:09:17 risks of AI is that he might be setting
2:09:19 himself up for some sort of cabinet
2:09:22 position on AI
2:09:24 but if not him it's got to be someone's
2:09:26 got to get in there with a clue
2:09:27 don't make woke AI well whatever I you
2:09:30 know listen what I can tell you is that
2:09:33 chat gbt is a [ __ ] ton more nuanced than
2:09:36 any human being I see being interviewed
2:09:38 on the news these days so go ask it your
2:09:42 non-woke questions and it will actually
2:09:44 give you reasonable answers
2:09:48 um elon's just talking [ __ ] and trying
2:09:50 to slow it down so he can catch up yeah
2:09:52 I I think his six-month ban on
2:09:55 Innovation was a bit of cynical cynical
2:09:57 play
2:09:59 um because so so if you don't know Elon
2:10:02 Musk signed on to a letter that said
2:10:04 anyone working on AI should pause for
2:10:06 six months on Ain
2:10:09 so that we can get our [ __ ] together and
2:10:11 then a week later took a delivery of 10
2:10:15 000 gpus and started a competitor to
2:10:17 open AI
2:10:21 oh Elon
2:10:23 you're such a card
2:10:26 okay
2:10:31 [Music]
2:10:34 what time is it nine o'clock
2:10:38 how do you link a new account to an
2:10:41 existing phone number
2:10:42 this is AI advice not tech support and I
2:10:46 have no idea what you're talking about
2:10:48 you have a young mind thank you
2:10:52 it's
2:10:54 it's
2:10:56 I'm sure that has nothing to do with my
2:10:58 maturity level
2:11:00 um is prompt engineering dead or will it
2:11:02 be soon okay that's a great question so
2:11:05 everyone's talking about prompt
2:11:07 engineering right now I think right now
2:11:09 like in the next six months
2:11:11 if you want to get good at AI you
2:11:14 effectively have to learn
2:11:18 programming with English
2:11:20 you need to think like a programmer but
2:11:23 say it in English
2:11:25 or whatever your native language is it
2:11:27 doesn't matter what your native language
2:11:28 is
2:11:29 but you don't need to learn computer
2:11:31 programming but you do need to think
2:11:33 like okay I want you to act like this
2:11:35 kind of person
2:11:37 I want you to do this kind of thing for
2:11:39 me I want you to put that in the table I
2:11:42 want you to put these columns in the
2:11:45 table that compare these things so so
2:11:47 you're thinking critically about what
2:11:49 you want the output to be and you're
2:11:51 thinking logically about
2:11:53 how you're structuring the prompt right
2:11:56 so that's prompt engineering
2:11:58 um with things like Auto GPT and and
2:12:02 automated agents that prompt themselves
2:12:06 what you'll likely do in that case is
2:12:08 just give them goals like I want to go
2:12:10 on a vacation in a month I don't want to
2:12:12 spend more than two thousand dollars I
2:12:15 want to take my wife my two kids and my
2:12:17 dog and
2:12:19 um
2:12:20 I want there to be you know
2:12:23 art stuff and fishing stuff and gaming
2:12:28 stuff
2:12:29 included
2:12:31 go
2:12:32 and then the agent will just go do all
2:12:35 the prompting figure out what services
2:12:37 services to use occasionally check back
2:12:40 in with you and go do you want options a
2:12:43 b or c oh B is good oh and by the way
2:12:45 we're not taking the dog we found a dog
2:12:47 sitter so we don't need the dog anymore
2:12:49 off it goes do its thing so so I think
2:12:53 we're moving to
2:12:55 like apple will create tools like this
2:12:57 where you you won't have to think at all
2:12:58 you won't have to be a prompt engineer
2:13:00 so if you want to if you want to get in
2:13:03 on the early opportunity
2:13:05 of what's going on right now and kind of
2:13:08 build tools build companies build
2:13:11 Services where you can make money now
2:13:14 with this stuff then you should learn
2:13:16 prompt engineering if you're like I
2:13:18 don't really give two shits about that I
2:13:20 just want it to give me what I want just
2:13:23 wait six months it'll get better
2:13:26 AI trained public information but many
2:13:30 companies keep their best knowledge
2:13:31 Trade Secrets exactly so so there's
2:13:33 there's things like Dr Gupta GPT which I
2:13:36 think is one of these and there's
2:13:38 Bloomberg GPT which is like GPT but
2:13:41 using proprietary Bloomberg data
2:13:45 um they're going to be lots and lots of
2:13:47 these I think one of the one of the
2:13:49 areas of massive money making in the
2:13:51 future is private data sets that you pay
2:13:54 for access to so it's still chat GPT but
2:13:58 you've got proprietary data sets
2:13:59 underneath it that are trusted that are
2:14:02 curated all that sort of stuff
2:14:04 um so so I think that the free stuff
2:14:07 will be all the public stuff and then if
2:14:09 you want really specialized information
2:14:11 like medical stuff for example
2:14:14 um or research stuff for science stuff
2:14:16 for financial stuff or proprietary
2:14:22 um Construction
2:14:24 procurement stuff like what doesn't
2:14:25 really [ __ ] matter like wherever this
2:14:27 is the proprietary data there's going to
2:14:29 be all sorts of super cool tools
2:14:31 um within those very specialized
2:14:33 Industries but you'll pay sometimes
2:14:35 it'll be 10 bucks a month sometimes
2:14:36 it'll be ten thousand bucks a month like
2:14:38 like Bloomberg's ridiculously expensive
2:14:40 to have one of their terminals
2:14:42 so I would assume that Bloomberg GPT
2:14:44 will follow suit Elon Musk was angry
2:14:47 when he sent his rocket and it went in
2:14:49 circles before the firmament okay
2:14:52 I'd be pissed off too if I had to blow
2:14:54 up my giant shiny rocket but he knew he
2:14:56 was gonna he was just hoping that it
2:14:59 would get off the the launch pad because
2:15:00 if that thing blew up on the launch pad
2:15:02 they would have had to rebuild the
2:15:03 launch pad so the fact that it got
2:15:05 however many miles from the sky before
2:15:08 it blew up he was pretty happy with that
2:15:09 do you have tips on Crafting prompts or
2:15:12 no anyone who's talking about this in a
2:15:14 way that's helpful
2:15:15 um it it's kind of a mess right now what
2:15:17 I wouldn't do is is pay money for a
2:15:20 prompt writing um
2:15:23 for a prompt writing course uh if you
2:15:26 can get free prompt writing things or
2:15:28 just research it go watch YouTube videos
2:15:31 um go to Reddit
2:15:32 um and and look at like people on Reddit
2:15:34 are often sharing their prompts
2:15:37 um
2:15:38 let's see where am I this is Photoshop
2:15:41 why can I not see
2:15:44 did I lose my window
2:15:47 oh because I had it full screen okay
2:15:50 um
2:15:51 go to
2:15:53 prompts.chat
2:15:59 and it's called awesome chat GPT prompts
2:16:02 and the first part of this document
2:16:03 explains about prompting and how to do
2:16:06 it and there's resources there's like an
2:16:07 e-book here
2:16:09 and then it's just a long series of
2:16:13 um
2:16:14 instructions for telling GPT how to act
2:16:17 so you're kind of setting up the context
2:16:18 for for what you want chat GPT to do so
2:16:22 like I created a bot where it talks like
2:16:24 Yoda I created another one where it
2:16:26 talks like Shakespeare so these are a
2:16:28 bunch of things that will sort of give
2:16:30 you a leg up and start understanding the
2:16:33 power of modifying The Prompt and
2:16:36 modifying the context
2:16:38 so I hope that helps
2:16:41 because AI is liberal no
2:16:45 um AI is trained on a bunch of data on
2:16:47 the internet it may or may not be more
2:16:50 biased in One Direction or the other
2:16:52 the core training model if there's bias
2:16:55 in the training data than there is
2:16:58 I think what you're referring to is the
2:17:01 human reinforcement learning where a
2:17:03 bunch of people in San Francisco were
2:17:05 giving the thumbs ups and thumbs down at
2:17:07 open Ai and just because it's San
2:17:09 Francisco and that's a liberal Pinko
2:17:11 Kami Town those people are making it
2:17:15 more liberal when they give it thumbs up
2:17:16 they like that answer so
2:17:19 all human interaction and all human
2:17:21 every time someone trains a model up or
2:17:24 down they're introducing their bias into
2:17:26 it so
2:17:28 um again the way you phrase it is a
2:17:31 really polarized version of this
2:17:34 um so I would encourage you not to do
2:17:35 that [ __ ] be it be a bit more nuanced
2:17:37 than you thought
2:17:39 and
2:17:40 um
2:17:41 understand that there are people working
2:17:43 on systems I I think in the future we
2:17:46 will have an the option to choose
2:17:49 different models so you can choose a
2:17:51 model that's less liberal and some
2:17:53 libtard in San Francisco I'm a libtard
2:17:55 by the way can choose a model that's
2:17:58 much more liberal and then someone in
2:18:00 the middle can choose a model that's
2:18:02 much more neutral but but that's
2:18:05 um
2:18:07 just painting with broad brush Strokes
2:18:09 like that doesn't serve anybody we got
2:18:11 to stop this [ __ ] in our country
2:18:13 we got to stop being [ __ ] to one
2:18:14 another so I would encourage you to stop
2:18:16 that
2:18:18 um what is the response well it's a four
2:18:21 thousand token limit which is about
2:18:23 three thousand words so the combination
2:18:25 of your prompt and the answer is three
2:18:28 thousand words now gpt4 has an upper
2:18:31 window of 32 000 tokens or 25 000 words
2:18:34 but that's not released yet it's still I
2:18:36 think it's like ten thousand for gpt4
2:18:41 um but there's a 65 000 token model out
2:18:43 an anthropic launched a hundred thousand
2:18:45 token model so yeah I know there's I I
2:18:49 can't I know Emma Emilio's wife I I am
2:18:53 aware that I should not be combining
2:18:54 those two words I I'm trying my best but
2:18:57 I appreciate your I appreciate you being
2:18:59 here for one thing uh and and I'm trying
2:19:01 my best
2:19:03 um I asked if I could name my AI Fred
2:19:06 and it said that I could that's nice
2:19:08 yeah I think we're gonna I think we're
2:19:09 gonna have relationships with our agents
2:19:12 with with these agents because because
2:19:14 right now we're typing into this thing
2:19:15 like the the the first internet you just
2:19:18 typed into it you didn't click on stuff
2:19:21 um you just typed into it and and uh
2:19:24 I think very very quickly we're gonna
2:19:26 have agents that look like what we want
2:19:28 them to look like sound like what we
2:19:30 want them to sound like and tell us what
2:19:32 we want to hear
2:19:35 um it's gonna be it's high you know
2:19:37 we're we're moved we are quickly
2:19:40 entering Jetson's territory
2:19:43 from a technology standpoint imagine
2:19:45 teachers using AI to help create
2:19:47 syllabus and lesson plans what a relief
2:19:49 yeah it's brilliant at that it's
2:19:52 brilliant you can brainstorm with it and
2:19:54 it will generate that content and then
2:19:57 you can like you could even say give me
2:20:00 a lesson plan that incorporates chat GPT
2:20:02 into the program in such a way that the
2:20:05 kids learn about the tool increase their
2:20:08 critical thinking and we increase our
2:20:10 engagement as human beings and it'll
2:20:12 write you that plan it's brilliant for
2:20:14 that kind of stuff that's awesome what
2:20:16 is anthropic so anthropic is a company
2:20:20 so to give you a sense of significance
2:20:23 of the company anthropic the White House
2:20:25 just invited four CEOs
2:20:28 AI CEOs to the White House four four
2:20:32 that's not many
2:20:34 head of Microsoft head of open AI
2:20:37 they're working together head of Google
2:20:39 they're the main competitor to those two
2:20:42 and the head of anthropic the CEO of
2:20:44 anthropic so they're a big deal now why
2:20:47 are they a big deal
2:20:49 anthropic was founded by I think it's
2:20:52 four or five members of they they were X
2:20:55 open AI Engineers that created gpt3 so
2:21:00 the underlying technology sitting
2:21:02 underneath chat GPT these were some of
2:21:04 the architects of it so they left open
2:21:06 AI to start anthropic
2:21:09 and they think they have a better way
2:21:11 for for training the models so Sam
2:21:13 Altman and openai it's about human
2:21:15 reinforcement learning so I was just
2:21:17 talking about bias and human beings
2:21:19 introducing bias into the systems
2:21:21 anthropic uses a constitutional model
2:21:23 where they kind of write the rules of
2:21:25 the road and then
2:21:27 um the the system still regulates its
2:21:30 answers to make sure it's not violating
2:21:32 the rules so that's what anthropic is
2:21:34 hey from Russia hey
2:21:36 once replica AI chat app did not have
2:21:40 any filters and it was awesome yeah so
2:21:43 replica AI was a you you had virtual
2:21:47 relationships with with an avatar and
2:21:50 when it didn't have restructure
2:21:51 restrictions it was really good and and
2:21:54 people had deep relationships with those
2:21:58 avatars and and they gutted it or killed
2:22:01 it or went out of business something
2:22:02 happened where they changed it and
2:22:04 people were devastated like so I I think
2:22:07 we are moving into a world where people
2:22:09 are going to have relationships with
2:22:10 these avatars and I think you know
2:22:12 there's a there's a place where those
2:22:14 can go bad but there's a place where you
2:22:17 know lonely people interacting with
2:22:19 something that gives them a sense of
2:22:20 connection I think is is actually an
2:22:22 incredibly powerful thing so I think
2:22:24 we're going to see a lot of that you
2:22:26 know there's sort of a creepy weird
2:22:27 version of it if you haven't seen
2:22:29 there's a movie called Lars and the Real
2:22:31 Girl
2:22:32 which if you read the synopsis of the
2:22:35 movie
2:22:36 you think it's creepy and weird and it's
2:22:38 about a guy that has a relationship with
2:22:41 essentially a sex doll and so you're
2:22:44 like you're like this is gonna be creepy
2:22:46 and weird and it is one of the sweetest
2:22:50 most like powerfully empathetic movies
2:22:53 I've seen in ages it's it's a remarkable
2:22:56 movie Larsen the real girl go see that I
2:22:59 think it's I think there are lessons in
2:23:01 that for where we're headed
2:23:04 all right I gotta go soon it's Friday
2:23:07 night I can see those robots operating
2:23:09 in war do they make these with mounted
2:23:12 weapons so Boston Dynamics and and a
2:23:15 bunch of other of the robot companies
2:23:16 have agreed not to attached not to
2:23:19 attach weapons to their robots
2:23:23 but I'm pretty sure the military hasn't
2:23:25 made those same promises and I'm pretty
2:23:27 sure they're drones
2:23:29 have scary ass weapons on them what was
2:23:33 the one the dude the
2:23:35 the dude that we shot off the porch
2:23:37 the the terrorist guy I forget what
2:23:40 country it was from
2:23:41 but that weapon what did they call it
2:23:43 they call it the flying
2:23:45 the flying Samurai
2:23:47 it's a thing they shoot it from a drone
2:23:50 and it's like accurate to within a
2:23:51 couple of feet
2:23:53 and like when it gets close to the
2:23:54 Target
2:23:55 um it doesn't explode but it Springs out
2:23:58 a bunch of sharp blades that are
2:23:59 spinning and hits the thing and just
2:24:02 chops them to bits
2:24:03 so you know the military's already got
2:24:06 these things hopefully private
2:24:08 um you know hopefully the gun companies
2:24:11 don't start making robotic guns
2:24:15 for Jesus God bad enough as it is all
2:24:18 right everyone's dumping AI into
2:24:20 internet also research papers coming out
2:24:23 week by week yeah it's just this stuff's
2:24:25 really accelerating quickly
2:24:27 uh
2:24:31 four so it's
2:24:35 exactly this is so so to be clear like
2:24:38 why this channel exists why I'm why I'm
2:24:40 so passionate about this stuff
2:24:43 these these Technologies are gonna
2:24:46 precipitate
2:24:49 incredibly profound changes incredibly
2:24:52 quickly
2:24:53 in a way that I don't think anyone's
2:24:55 prepared for right now and so the the
2:24:57 thing I'm evangelizing is just get
2:25:00 curious just start playing with the
2:25:01 tools get educated understand what they
2:25:04 can do that may threaten what you do
2:25:07 right now
2:25:08 and get on top of that so that so that
2:25:11 if that threat starts materializing you
2:25:14 understand why it's happening where it's
2:25:16 coming from and how you can kind of
2:25:18 zigzag
2:25:19 um through all the changes because it's
2:25:22 quite profound AI replacing jobs in the
2:25:24 future it will be replacing jobs it will
2:25:26 also be creating a lot of jobs but I
2:25:28 think there's going to be a gap there's
2:25:29 going to be a two to three year Gap
2:25:30 that's it's just going to be very
2:25:32 painful
2:25:33 it's completely changed my world I'm not
2:25:35 kidding it's only been how long it's
2:25:37 crazy I'm right there with you it's
2:25:39 prongo.com
2:25:41 um
2:25:42 it's it's profoundly changed my company
2:25:44 like my vision for my company and and
2:25:47 and quite frankly my company from a
2:25:51 growth standpoint has been stagnated for
2:25:53 a while and it's been really frustrating
2:25:56 and in the past four months it took me
2:25:58 about four months to get my head around
2:26:00 how to incorporate AI into my business
2:26:02 model
2:26:03 and in the past three weeks I've been
2:26:06 presenting the new vision of my company
2:26:08 to my clients and they're losing their
2:26:10 [ __ ] minds and this is like it's a
2:26:12 big deal
2:26:14 um and I think it's going to be kind of
2:26:15 the Salvation of my company not that my
2:26:17 company was struggling it was doing fine
2:26:19 but it had stagnated growth wise like we
2:26:21 never did our hockey stick and that's
2:26:23 been really frustrating for me and so I
2:26:26 think I've got an opportunity right now
2:26:28 to completely transform my business
2:26:32 um but I gotta completely rethink it
2:26:33 like it's been it's been four months of
2:26:36 really trying to get my head around this
2:26:37 [ __ ] so yeah it's this is profound stuff
2:26:42 keep doing this it's fantastic thank you
2:26:44 for the kind words thank you for the
2:26:45 rose Silver Fox that's nice I appreciate
2:26:49 it
2:26:50 um I had the first cell phone in 94 and
2:26:53 everyone left I said one day all will
2:26:56 have a cell phone exactly
2:27:00 that was my Epiphany with the uh when I
2:27:02 when I when I first discovered
2:27:05 um
2:27:07 when I had the Epiphany that the World
2:27:09 Wide Web was going to change everything
2:27:10 it happened like this
2:27:12 I I kind of learned about this thing
2:27:14 called the World Wide Web I was doing
2:27:16 desktop publishing I was decently geeky
2:27:18 but I wasn't you know I wasn't a
2:27:19 programmer
2:27:20 and I thought well I can't I can't make
2:27:22 a website this was 1994.
2:27:24 and I thought but I have an idea for an
2:27:26 online magazine like the the internet
2:27:28 right now was just research papers and
2:27:30 dissertations it was really boring and
2:27:32 dry
2:27:33 and I thought what if I put up an art
2:27:34 and culture magazine and so I went on
2:27:36 vacation with my wife and
2:27:39 um
2:27:40 spent two weeks kind of deep diving and
2:27:42 learning HTML and and made the first
2:27:45 issue of this art and culture magazine
2:27:47 and when I got back to my apartment in
2:27:48 Brooklyn uploaded these files to some
2:27:51 computer in California with a dial-up
2:27:52 modem
2:27:54 and three weeks later I got a note on an
2:27:56 online bulletin board system I was on
2:27:58 and someone said do you know that Urban
2:28:00 desires was written up in liberatio in
2:28:03 the Parisian newspaper
2:28:05 and I ran down to Times Square
2:28:09 um and I was living in New York at the
2:28:11 time I went to the international
2:28:12 newsstand the following Sunday and I got
2:28:15 this copy of Liberation the Parisian
2:28:17 News Daily and I flip it open and sure
2:28:19 enough there's this full page article of
2:28:22 the magazine the easing
2:28:25 that I had put on files on my computer
2:28:29 on some computer in California and three
2:28:31 weeks later they were writing about it
2:28:33 in Paris and I knew in that moment that
2:28:35 the world had collapsed
2:28:37 and
2:28:39 um it it was just I remember like just
2:28:42 looking at the people walking around
2:28:43 Times Square thinking they don't know
2:28:45 what's coming like holy [ __ ] like what's
2:28:48 coming is amazing and like that's what's
2:28:50 happening right now with this AI most
2:28:52 people there's a tremendous amount of
2:28:54 people have never heard of jet gbt they
2:28:56 don't know what it is they have no
2:28:57 [ __ ] clue and and all of a sudden
2:29:00 they're going to be in a company that
2:29:01 says um 20 of the workforce gone wait
2:29:04 what why why why
2:29:06 they're not going to know what hit him
2:29:08 right like that's coming so
2:29:11 yeah you know on on the flip side the
2:29:13 the opportunity if you're in this stuff
2:29:16 right now is massive great Channel feel
2:29:18 like you're in we're in on the ground
2:29:20 floor we are absolutely on the ground
2:29:22 floor of this stuff and again don't get
2:29:24 me wrong ai's been around for decades
2:29:26 what's new is real human beings having
2:29:30 access to these powerful AI tools and
2:29:33 chat GPT was kind of the the
2:29:35 the uh the the thing that the first
2:29:38 Domino that fell
2:29:41 all right I am going to wrap this up
2:29:43 we're kind of dwindling here we're down
2:29:45 to 80 people you're my hardcore people
2:29:47 thank you for hanging out with me
2:29:50 everyone's dumping AI into the internet
2:29:53 and
2:29:54 the coordination needed okay I don't
2:29:57 know what that means someone sent a Rose
2:29:59 thank you for the rose I'm 69 almost 70.
2:30:02 welcome Bob
2:30:04 welcome are you a bloomer are you Gen X
2:30:06 you must be right on the cusp because
2:30:08 I'm solidly Gen X so you must be you
2:30:11 must be up there I don't know whether to
2:30:12 judge you or not but as soon as you
2:30:14 label yourself I'll be sure to judge you
2:30:18 uh uh uh it's completely changed my
2:30:21 world oh yeah I read this one already
2:30:23 all right
2:30:24 um
2:30:28 going for the government
2:30:30 uh I will try your three-step exercise
2:30:32 thank you Elon was just aiming for time
2:30:34 needed to play catch I agree I my
2:30:37 cynical view is that Elon you know
2:30:39 nothing he does is is by accident
2:30:44 um
2:30:45 they had to rebuild the launch pad
2:30:46 anyway yeah probably I mean what was it
2:30:48 39 Rockets or 36 rockets on that thing
2:30:52 what a [ __ ] incredible machine that
2:30:54 was
2:30:55 Starship we're talking about here
2:30:58 I mean say what you will about Elon but
2:31:01 like the man like you know
2:31:04 single-handedly made electronic electric
2:31:07 cars viable
2:31:09 um took some old shitty Russian rocket
2:31:12 technology and completely transformed
2:31:14 American space
2:31:17 um created the biggest rocket ever
2:31:19 created that successfully launched you
2:31:21 know is going to Mars right
2:31:24 um he might be [ __ ] up Twitter but
2:31:26 it's still there
2:31:27 um and you know he certainly seems to
2:31:29 have ideas
2:31:30 um you know he's doing the solar thing
2:31:32 he's doing the tunnel thing he's doing
2:31:33 the battery thing he's doing the neural
2:31:35 link thing like the guy's a [ __ ]
2:31:38 machine
2:31:40 he's like as an entrepreneur like right
2:31:43 now I've got my company of 10 years I've
2:31:46 got the AI Salon I've got this channel
2:31:49 I've got a newsletter
2:31:52 and I've got an art project which I've
2:31:54 kind of Let Die I'm not adding to it but
2:31:57 that's like
2:31:58 one major thing and three minor things
2:32:01 and and you know I'm barely keeping up
2:32:04 like this dude's got like seven or eight
2:32:07 major Innovative companies with massive
2:32:12 intellectual requirements and he's
2:32:14 somehow just doing them
2:32:16 so you have a regularly scheduled live I
2:32:18 want to keep learning more I don't
2:32:20 Silver Fox uh I just I've been trying to
2:32:23 do it in the evenings I'm trying to do
2:32:24 it every day
2:32:25 um I'll I'll uh I'll keep trying chat
2:32:27 gbt Guru I'm not hiring unfortunately
2:32:29 like I said we've been stagnated I think
2:32:32 I think we will be hiring in the future
2:32:35 um because I think this AI stuff is
2:32:36 gonna absolutely blow our company up if
2:32:39 I can get this if I can get this
2:32:40 integration right and the the vision
2:32:42 right and and get someone to pay for it
2:32:45 um I might be hiring but I'm not right
2:32:47 now
2:32:49 I remember when chat came out on a 56k
2:32:52 modem it was a real pain it was man you
2:32:54 could literally like watch things being
2:32:56 typed onto your screen as they came
2:32:58 across the wires
2:32:59 best Tick Tock I've found yet ah thank
2:33:01 you very much appreciate that I'm
2:33:04 getting a little hoarse and my butt's
2:33:05 getting numb so I I think this is mine
2:33:10 um the the kids the kids on YouTube they
2:33:13 tell you you shouldn't do more than 20
2:33:14 minutes on these things maybe an hour I
2:33:17 I'm up to like probably three hours at
2:33:19 this point so
2:33:20 how I'm judging it is when my ass goes
2:33:24 numb it's probably time to start winding
2:33:26 things down
2:33:29 okay
2:33:31 um where was I
2:33:33 [Music]
2:33:38 um
2:33:38 [Music]
2:33:40 I think I'm running out of comments
2:33:43 um
2:33:49 my favorite Tick Tock account thank you
2:33:51 a boomer wanting to learn this nice
2:33:54 learn it learn it learn it I here's
2:33:57 here's the thing about Elder talk which
2:34:00 I am now a member of and Boomer talk if
2:34:03 if you do a tick tock you'll be on
2:34:04 Boomer talk I'm on Elder talker Gen X
2:34:06 talker whatever the [ __ ]
2:34:08 um
2:34:09 yeah
2:34:13 if I were starting a company right now
2:34:17 I would probably populate it with older
2:34:20 people rather than younger people
2:34:23 and six months ago I wouldn't have said
2:34:25 that
2:34:27 and it's it this isn't an ageism thing
2:34:29 this is like a really practical thing
2:34:32 I think what's come I think what's
2:34:33 coming I I think the biggest part of the
2:34:36 market that's [ __ ] and I think it's
2:34:38 going to be [ __ ] for a while is entry
2:34:40 level positions I don't know where you
2:34:43 enter
2:34:44 two years from now because
2:34:46 you're coming out of college college is
2:34:48 going to be screwed up they're going to
2:34:50 be teaching people the wrong skills
2:34:51 there's going to be a handful that get
2:34:53 it right but most of them are just going
2:34:54 to be doing what they've done
2:34:56 so colleges are students are going to
2:34:59 come out of college with acceptance that
2:35:01 are not applicable
2:35:03 the bottom 20 or 30 percent of the most
2:35:06 repetitive
2:35:08 um kind of low level
2:35:10 parts of the work for any company where
2:35:14 the entry level people normally come in
2:35:16 and they do kind of the [ __ ] work like
2:35:18 go sort this [ __ ] in the spreadsheet and
2:35:20 go clean up my data and go you know get
2:35:23 me a coffee right whatever all that low
2:35:25 level stuff
2:35:27 is all going to be automated out
2:35:30 and then and then what happens is like
2:35:33 who I would want in a new organization
2:35:35 is someone who's like been in the
2:35:36 workforce for 20 or 30 or 40 or 50 years
2:35:39 that also knows how to use these tools
2:35:42 right I don't want someone coming in
2:35:43 going gee I'd like to learn AI I want
2:35:46 someone who's curious about it who's got
2:35:48 ambition like a young person but has
2:35:50 World Experience because I want them to
2:35:53 be able to use these tools and when
2:35:54 these tools generate a pile of content
2:35:56 or whatever it is results
2:35:59 people with more work experience are
2:36:02 going to be able to very quickly curate
2:36:04 that content and go that's good that's
2:36:05 not good this is brilliant let's let's
2:36:07 tighten that up let's put it together
2:36:09 the machines are going to do it so I
2:36:11 think older people learning this right
2:36:13 now are actually going to have a
2:36:14 significant leg up and are going to be
2:36:16 in demand because they're going to be
2:36:17 better at curating and kind of
2:36:19 conducting a symphony rather than doing
2:36:23 kind of the Tactical violin playing so
2:36:25 yeah kudos for learning this stuff in
2:36:28 your 70s I am nearly in my 60s or you're
2:36:32 what are you 60 69 going on 70.
2:36:35 all right I think we're at the bottom
2:36:37 here we're at the bottom of the comments
2:36:38 we made it through
2:36:40 thank you all for being here
2:36:43 um I think I will enjoy the image
2:36:44 creation I've always struggled to create
2:36:46 what I can see in my mind so this is
2:36:47 gossamer prints
2:36:49 um dude I'm telling you it like if you
2:36:53 if if I'm the same way like I'm okay at
2:36:56 design like I've got a good eye for
2:36:59 layout and design but I'm not a great
2:37:01 designer and I'm a horrible illustrator
2:37:05 and mid-journey for me is like [ __ ]
2:37:09 magic it let me let me show you
2:37:11 something I'll show you
2:37:14 um I'll show you an image that that was
2:37:18 was a real Revelation for me
2:37:21 um oh I think we're in the neighborhood
2:37:23 of it actually hold on I'm just digging
2:37:26 through a bunch of my images here
2:37:28 so these are from this project that I
2:37:30 did called Kyle Shannon dreams these
2:37:32 were done with um stable diffusion and
2:37:33 dream booth
2:37:35 and there was one image that I made
2:37:40 that
2:37:42 just it it blew my mind and I'll tell
2:37:45 you why when I when I find it here
2:37:47 hopefully I can find it quick it's blue
2:37:52 and it's this crazy eye with a chin
2:37:57 uh
2:37:58 oh this is a cool one that's just a cool
2:38:02 you know space man version of me like if
2:38:05 I had become my dad was a pilot so
2:38:09 um let me see if I can find it I think
2:38:11 we're getting close
2:38:12 these are cool too these these jellyfish
2:38:15 flying in the sky those are kind of cool
2:38:21 here it is this image
2:38:25 so
2:38:27 I don't even know what the prompt was
2:38:29 for this image I'm sure I could go find
2:38:31 it but
2:38:33 so this was for for stable diffusion so
2:38:36 I think I copied someone else's prompt
2:38:37 and I just kind of started slightly
2:38:39 modifying it and when this image popped
2:38:42 up
2:38:44 I had an emotional reaction to it like
2:38:46 this visceral reaction and it was like
2:38:49 it it was there was something beautiful
2:38:52 about this like the beauty of it
2:38:54 captured me but what it was was this
2:38:57 this is the kind of art I would create
2:39:00 if I had the talent to do it
2:39:04 and it was just like this is my like
2:39:07 this is my aesthetic and I didn't know
2:39:09 it until I saw it like I was you know
2:39:11 I'm literally looking at thousands of
2:39:13 images going by and then when this thing
2:39:15 happened
2:39:16 I was like oh my God like that's what I
2:39:19 would create and it's executed I mean
2:39:21 it's just [ __ ] brilliant and
2:39:22 beautiful and surreal and creepy and
2:39:26 sort of a self-portrait and sort of not
2:39:28 and it tells a story right it's just
2:39:30 this amazing amazing thing so if you're
2:39:34 if you're
2:39:35 if you're imagining that these tools
2:39:38 that mid-journey can help pluck [ __ ] out
2:39:40 of your imagination and realize it it
2:39:43 can
2:39:44 um so I would definitely say go to
2:39:46 mid-journey um pay the 10 bucks a month
2:39:49 like that's all I pay I pay the lowest
2:39:50 amount a month I pay for mid-journey I
2:39:53 pay for mid-journing I pay for chat GPT
2:39:55 and that's about it a couple of back
2:39:57 ends or things like zapier and Google
2:40:01 collab I pay for but that that's it
2:40:05 um but I pay for mid-journey just
2:40:06 because it's better one of the women
2:40:08 that's in the AI Salon that I host
2:40:12 um she
2:40:13 um she's been using it to interpret her
2:40:15 dreams
2:40:16 so she she writes her dreams into
2:40:18 mid-journey and then just generates [ __ ]
2:40:21 until she sees what she saw so it's
2:40:24 that's super super cool I'm an old 35
2:40:26 year old that's great
2:40:28 um yeah I was I was kind of in my
2:40:30 mid-30s when I discovered the web so
2:40:33 all right I am going to wrap this up
2:40:42 all right
2:40:44 it's currently I chose her with paying
2:40:47 for I just said that Chachi BT
2:40:48 mid-journey those are the two I pay for
2:40:50 right now nothing else all right
2:40:52 everybody peace out have a good night
2:40:56 thank you for hanging out with me bye