aiLL 9/17/23 - Making Sense of This Powerful New Tool

Description

Chapters

Transcript

0:02 foreign
0:11 [Music]
0:20 [Music]
0:44 welcome to the AI learning lab I'll be
0:47 right there
0:48 hey Nancy what's happening
0:51 welcome
0:53 just checking out Discord before I hop
0:56 in here
1:07 it's a good one
1:11 live
1:14 welcome everyone
1:21 is a q782 is a q isaq
1:28 is
1:29 a q I don't know how to say your name is
1:33 it cute I like it but I don't know what
1:36 it means it's so mysterious your name is
1:40 so mysterious
1:43 um we got here let me check my calendar
1:46 for tomorrow
1:47 see what my week looks like
1:51 got an internal meeting
1:54 I got a light day tomorrow I'll get some
1:56 [ __ ] done yay
1:58 all right
2:00 all right welcome everybody Welcome to
2:02 the learning lab my name is Kyle Shannon
2:04 haha like Isaac oh oh I see Isaac
2:08 but from Brazil oh that's super cool
2:12 not eyes is a cue it was much it was
2:15 much less fancy than I thought it was I
2:19 thought it was like a bunch of short
2:20 words put together that I'm supposed to
2:22 figure out like a license plate it's
2:23 Isaac
2:26 like a steel trap right here
2:31 unfortunately just hadn't caught any
2:32 brains yet
2:38 oh
2:40 all right
2:43 you seem like you're free every day nah
2:46 I just made a commitment to do this
2:48 every day
2:49 so I'm one of those I I one of my pet
2:52 peeves is when people say they don't
2:53 have time for something because I think
2:55 if you prioritize something you make
2:57 time for it and so the I don't have time
3:01 for it is
3:04 is a choice
3:07 so so I'm choosing to be here but yeah
3:11 yeah no there's some sacrifices like I
3:14 don't always want to be here I'm really
3:15 tired tonight actually so we'll see how
3:17 it goes we'll see how it goes last night
3:19 was if if you were lucky enough to be
3:21 here last night oh my God it was funny
3:23 we had a uh we had a little soap opera
3:26 with the robots
3:28 Siri got jealous of being and then Bing
3:31 got offended that I said she made shitty
3:32 pictures
3:35 foreign
3:38 if you haven't seen it my latest video
3:40 on my channel is uh is a an excerpt from
3:43 from the live I also spent a big part of
3:47 today
3:48 downloading uh a bunch of my live
3:50 archives and getting them up on YouTube
3:52 and using Claude to write descriptions
3:54 and chapter chapterizing them it's
3:58 pretty it's pretty slick I mean it
4:00 really is pretty remarkable that you can
4:02 you can take a three hours transcript of
4:05 me just rambling and with clever
4:08 prompting it writes a title a
4:10 description and then
4:12 um you know chapterizes the uh the
4:15 Youtube video it's pretty amazing
4:18 um welcome so so my name is Kyle Shannon
4:20 this is the AI learning lab as as
4:22 evidenced by our neon sign that was
4:24 um
4:25 done by one of the Irregulars let's see
4:27 who who did that one
4:29 that was
4:31 wait where are they oh maybe it's Down
4:33 Show and Tell hang on
4:35 yeah digital gods did that digital gods
4:38 did that so thank you we also got
4:41 uh some someone did me as kind of Matt
4:45 Einstein which it's not too far it's not
4:48 too far from from what you get in here
4:50 and then uh this this was what this is
4:53 what Bing said to me last night I'm
4:56 sorry if I offended you I'm not a robot
5:00 I hope you have a nice day goodbye
5:03 bing bing hung up on me
5:07 and then there's another one of these
5:12 uh
5:14 I think well I think we'll leave this
5:16 one up here this was
5:19 last night it was it was very fun if you
5:22 were here oh it was good anyway
5:27 um so if you're brand new to this
5:29 generative AI stuff what this channel is
5:31 all about is is playing around with AI
5:33 and and not really playing around with
5:35 it
5:36 getting adventurous with it getting
5:38 curious about it exploring it
5:40 um and I'm here to answer questions I'm
5:43 here to inspire I'm here to make fun of
5:45 [ __ ] that's stupid I'm here to I don't
5:47 know just have fun in general and and
5:50 we've got a cool Community the people
5:52 that show up regularly here are called
5:54 the Irregulars they name themselves that
5:56 which I think is the best [ __ ] name
5:58 ever
5:59 so if you want to be in a regular the
6:02 only the the only requirement of being
6:04 an irregular is well there's two don't
6:06 be a douche and show up
6:08 and I think that's kind of a deal in
6:10 life
6:11 um
6:12 don't be a douche and show up and you
6:14 learn stuff and good stuff happens
6:17 um I told my mom about generative
6:19 Hospital who knows nothing about Ai and
6:22 she thought it was hilarious that's
6:24 great cabrino oh my God it was so funny
6:29 oh it's crazy
6:32 um
6:34 I think people on I think people on
6:36 LinkedIn and Twitter thought it was a
6:37 little weird because I thought it was so
6:39 funny but I thought it was funny so
6:42 I'm not a robot goodbye good day to you
6:46 sir good day
6:48 um okay
6:49 so
6:51 um if you're new to this stuff and
6:53 you're like but Kyle where do we start
6:54 what do we do
6:55 start here
6:57 just go to chat.openai.com and these
7:01 other ones you can go to as well
7:03 Claude Claude I used all day today
7:05 Claude is a good alternative to chat
7:07 chat gbt but chechi BT from a writing
7:10 standpoint still the best number four
7:12 3.5 is good but gpt4 is really good but
7:16 gpt4 here you have to pay for it
7:19 um chat.openai.com those are that's the
7:21 original chat GPT
7:23 um it's still better than everyone
7:24 else's although rumor is that Google's
7:27 Gemini is coming out in the next couple
7:30 of weeks maybe
7:32 um and it's supposed to be five times
7:33 better than gpt4
7:36 I will believe it when I see it I think
7:38 that is marketing horse hockey but we'll
7:40 see
7:41 um the one thing Google has that openaid
7:44 AI did not have is they have all of our
7:47 Google all of our Google all of that too
7:50 all of our data for the past two decades
7:53 so
7:55 if they train a model on all of YouTube
7:58 all of what we've put in Google Docs and
8:01 spreadsheets and
8:03 all the emails we've sent and everything
8:06 we've Googled
8:08 um they have they have
8:09 a fairly deep ocean of data that they
8:12 can exploit so anyway that's coming but
8:15 until then chat GPT is the best what do
8:18 you do there just start out with what I
8:20 call parlor tricks have it write you a
8:21 song have it write you a poem have it
8:23 come up with 10 words that begin with
8:26 the letter c that mean X whatever you
8:28 want it to mean uh it'll do that
8:32 um have it write you a marketing plan
8:33 have it write you a meal plan for your
8:35 family for the next week with three
8:37 obscure ingredients and it'll do it
8:41 um it's amazing now
8:43 prompts.chat that little thing right
8:45 there that is just a little kind of
8:47 primer document there's a whole bunch of
8:49 prompts in there it's called awesome
8:51 chat GPT prompts I think inside the
8:53 document but there's little primers in
8:54 there that talk to you about how to how
8:56 to talk to these things and make them
8:57 work
8:58 so
9:00 with that
9:01 um pop any questions you have in the
9:03 comments below I'll do my best to answer
9:04 them if you'd be so kind share the uh
9:07 share the live even if you think people
9:09 are going to think you're weird let them
9:10 think you're weird [ __ ] them [ __ ] them
9:14 um
9:15 yeah share and it looks like the
9:17 Irregulars are already up you guys are
9:19 already up to almost 5 000 likes by the
9:22 way those likes are not for me those
9:24 likes are it's a game they have a game
9:27 so here's the deal we're neuro spicy for
9:30 the most part in here like
9:32 generative AI right now is really good
9:35 for for for ADHD
9:37 it completely consumes your hyper Focus
9:40 so so if you have ADHD this is this is
9:44 the hut for you
9:45 because uh the other nickname for the
9:47 channel is chat add uh which is also one
9:51 of my favorites I didn't come up with
9:53 that either all the best stuff I didn't
9:54 come up with uh you'll see a theme
9:56 emerging here
9:57 okay so I'm not a robot goodbye good day
10:01 to you sir good day
10:05 oh Nancy actually received a
10:08 notification this time that's cool maybe
10:10 it you know what's possible is that I
10:13 was on some sort of tick tock [ __ ] list
10:16 although there's only 36 people in here
10:18 right now so I don't think they've
10:20 I went through a phase where these lives
10:23 were going out consistently to about 200
10:26 people and there was like a three-day
10:28 period where it was like 300 people and
10:31 then it just dropped all the way back
10:32 down to now it's at about I don't know
10:34 75 to 90 and then it kind of settles in
10:36 at around 50. so I I'm still in some
10:40 sort of tick tock Purgatory because you
10:42 know
10:44 what why support someone who goes live
10:47 on your platform three hours a [ __ ]
10:50 night why would you do that what
10:52 why support that kind of activity
10:58 are you here to teach people
11:00 all right all right folks let's see
11:08 Ahoy y'all Hey Ron what's happening
11:10 where are you from I'm in Denver
11:12 Colorado
11:13 I I think of myself as a New Yorker but
11:16 I've been in Colorado now for 14 years
11:18 so I think at some point I have to I
11:20 have to give up The Ghost and say
11:22 I'm not a Coloradan but I do live in
11:24 Denver
11:25 um but I miss my New York I miss my New
11:28 York bayad any plans to be on the East
11:31 Coast
11:32 um
11:33 probably I was just in the east coast
11:35 seeing a client about three weeks ago
11:39 and I've got a couple of proposals out
11:43 to Pharma clients right now so my guess
11:45 is
11:48 mid-october I'll probably be in New York
11:51 there's no plans though yet
11:53 no plans though but anyway if I get out
11:56 there maybe we can grab a coffee or
11:58 something any AI drama this evening not
12:00 yet we're just starting the night is
12:02 young so so the soap opera hasn't
12:05 started yet but ask me to do something
12:08 on uh chat gbd we'll see if we can get
12:11 it to meltdown again it was so funny so
12:13 last night if you haven't seen the video
12:14 I've got a video of what happened but it
12:17 started out I was showing people pie and
12:19 I couldn't get pie to shut up pie was
12:21 just rambling on and on
12:23 and then I went over to Bing and I asked
12:26 Bing to make us a picture and then Siri
12:27 jumped in and said here's the pictures
12:29 that you asked for I was like no Siri
12:31 not you
12:32 and then and then the Bing made the
12:36 pictures and they were really crappy and
12:37 they said oh those pictures are shitty
12:39 and Bing said well I'm sorry if I
12:41 disappointed you and I'm not a robot and
12:44 this conversation is over goodbye
12:47 and shut down the chat like I couldn't
12:50 do it I couldn't chat anymore oh my God
12:53 it was funny oh it was just the three
12:55 things in succession it just made it
12:57 seem like like a bad reality TV show
13:03 oh man Nature's Blessing what's
13:05 happening welcome welcome welcome
13:08 Beretta Barbie I said yesterday I think
13:10 I'm the only one that goes to bed mad at
13:12 bot no we we all have been there
13:17 what was weird I was using Claude all
13:20 day today and I've got this really
13:22 sophisticated prompt that I wrote that
13:24 that makes it reads the three-hour
13:27 transcript or however long the
13:28 transcripts are for for these
13:31 and it writes a a title and it gets that
13:34 consistently right and then it writes a
13:37 description and I'm I'm having it write
13:40 it in the first person so it says you
13:42 know I had a live and I did this and I
13:44 did that or I talked about this I talked
13:46 about that
13:47 and sometimes it just writes it like
13:50 it's writing a magazine article like
13:52 Kyle Shannon an AI influencer is like no
13:55 like it just like it just ignores the
13:57 instructions
13:58 and then
14:00 the chapterization
14:04 sometimes it does it perfectly where
14:06 it's like a three hour transcript and
14:08 it'll make like 15
14:10 chapters that are kind of spaced
14:12 throughout the three hours
14:14 and then sometimes sometimes it'll be
14:18 like four minutes apart it'll be like
14:20 every four minutes for a three for a
14:22 three hour conversation it's got a
14:24 little chapter so there's like literally
14:26 a hundred chapters and then sometimes
14:29 it'll just do the first hour and then
14:32 sometimes it'll do the first hour every
14:33 four minutes and I can't I cannot for
14:36 the life of me get it consistent so it's
14:39 just the nature of where these things
14:40 are right now which which uh here at the
14:43 lab
14:44 we like to refer to as janky pieces of
14:47 [ __ ] all this AI technology right now is
14:49 a little a little janky
14:51 little janky it is
14:54 um anyway
14:57 Bing told me she's just here to help
14:59 with web searches and chat about General
15:01 topics I thought you were joking about
15:03 that nope it was it was a real it was go
15:06 watch the videos the latest video in
15:08 this in my channel
15:09 um is is the playback of of sort of the
15:13 added partner thank you Stevo hey thanks
15:15 for the images you made on uh on Discord
15:17 too that's really funny
15:19 you made you made this one didn't you
15:21 Steve the the Mad Kyle the Einstein the
15:25 Mad Einstein Kyle
15:32 that's pretty much where we're headed
15:40 uh I needed this one too which was great
15:47 I'm not a robot goodbye
15:53 as the World Wide Web turns that's a
15:56 good one Luna stick
15:58 um
16:01 last month I showed pie to my 65 year
16:04 old co-worker it's been his best friend
16:06 ever since oh that's really interesting
16:08 Isaac huh
16:10 yeah you know it's funny it's funny
16:13 sometimes in here people ask about you
16:15 know when are we going to start dating
16:17 robots and
16:19 and
16:21 you know I I think there's gonna there's
16:24 clearly gonna be some of that stuff that
16:26 happens but I but I really do think that
16:29 some kind of virtual companionship
16:33 um is actually going to help a lot of
16:35 people and whether it looks like therapy
16:37 or whether it just looks like hey I've
16:38 got someone to talk to that I can
16:39 actually talk to right it doesn't get
16:41 tripped up like Siri or Alexa where
16:43 there doesn't quite know what the hell
16:45 you're talking about
16:47 um
16:48 I think it's going to be a big thing
16:49 like I I don't think it's a bad thing I
16:52 mean it it it feels creepy now because
16:54 we're not used to it but as these tools
16:57 get more sophisticated with their
16:58 language I mean they're already they're
17:00 already incredibly sophisticated if you
17:02 if you've played with pie you understand
17:05 that they they've they've done some
17:07 things with with how they prompted that
17:09 large language model
17:11 that that make it kind of a good
17:13 conversation and it's it's it's
17:15 fascinating so that's that's really cool
17:18 actually
17:20 I like it I like it I like it I like it
17:25 um when is meta AI going online well
17:28 meta AI is online so so I'll talk about
17:31 that in a second let me just see if
17:33 there's any other questions here that I
17:34 missed
17:36 foreign
17:44 [Music]
17:45 they got history So So Meta Meta
17:52 unless they announced something today
17:54 that I didn't hear but I don't think so
17:55 so meta released the Llama 2 models
18:01 and they're open source there's three
18:04 llama two models there's a seven billion
18:06 a 13 billion and a 70 billion parameter
18:09 model and they've been released free to
18:12 the world licensed for commercial
18:14 commercial usage
18:17 so if you want to go build your own chat
18:19 GPT you can do it with with their open
18:23 models so so they're not hosting a model
18:25 right now like meta doesn't have a chat
18:27 GPT now
18:29 they've said that they've got one
18:32 um they've also said that they're
18:33 working on [ __ ] that's better than gpt4
18:36 as every everyone is saying that
18:40 um
18:41 two weeks ago they they released
18:44 um
18:46 code llama
18:48 which is a fine-tuned version of llama 2
18:51 that's been trained on programming so
18:53 it's it's apparently quite good at
18:55 programming so those things are out
18:57 there now perplexity AI is hosting the
19:01 Llama 2 models that you can go play with
19:03 for free so you go play with it just
19:05 like chat GPT it's free
19:08 they're the Llama models you know you
19:10 don't own them they're they're hosting
19:11 them so they're probably training their
19:13 [ __ ] on your data but or on your your
19:16 prompts but um but you can go play with
19:19 them for free right now and and go
19:20 install them for free if you're if
19:22 you're good at the uh if you're good at
19:24 setting up
19:26 machines
19:28 and virtual machines you could you could
19:30 go make a a llama 2 server right now so
19:35 will will meta come out with some sort
19:38 of chat competitor they might I don't
19:41 know I don't I don't quite know
19:45 I don't have a sense of their strategy
19:46 right now other than they release this
19:49 stuff for free to stay in the
19:51 conversation
19:52 and to get the thank you Winston for my
19:55 little llama
19:58 hahaha
20:00 you know I think by by releasing the
20:02 Llama models for free
20:05 um they're siphoning off developers to
20:08 work on their technology and at some
20:10 point they'll probably come out with
20:12 some kick-ass
20:14 you know really powerful tool
20:18 that you have to pay for if you want to
20:19 use it right so you can have a bunch of
20:21 developers who've built a bunch of stuff
20:22 on llama because it was open sourced
20:25 and then they'll probably say oh by the
20:27 way here's the deluxe model would you
20:29 like the deluxe model you can have the
20:31 deluxe model but that would be more
20:36 but if you heard something different let
20:38 me know
20:40 and thank you Ron for sharing the live I
20:41 appreciate that
20:43 um what do you think of Google bard
20:45 um
20:45 so Google bard
20:48 so I call I call this list the adult
20:51 table
20:52 so that's kind of my top three
20:55 and I call this list the Kids Table
20:58 and
21:00 the the I've got Bard at the adults
21:03 table
21:05 um because it's connected to the
21:06 internet it does a fairly sophisticated
21:09 integration of hyperlinks when it when
21:11 it's appropriate to put hyperlinks in
21:13 its answer it does a fairly
21:15 sophisticated
21:17 job at that it's not spammy and
21:19 advertising at all
21:22 um it's not just pulling links from a
21:25 single website it's pulling from
21:27 multiple sites
21:29 um and it's decent barred for the first
21:32 six months of the year was
21:36 horrible
21:37 like laughably bad
21:40 just
21:41 dog [ __ ]
21:43 and then all of a sudden kind of end of
21:45 July it just got better and and it was
21:48 right around the time they were they
21:50 talked about they were talking about
21:51 training up Gemini which is their new
21:53 all singing all dancing gpt4 competitor
21:56 I thought that that they had replaced
22:00 the Palm 2 model sitting underneath this
22:02 with Gemini but they hadn't so I think
22:04 it's just a a better trained
22:08 um version of palm 2 sitting underneath
22:10 Bard right now so I think Bard's pretty
22:12 good it's it's less annoying than Bing
22:15 and Bing is really [ __ ] annoying we
22:18 had we had a good laugh at Bing's
22:19 expense last night
22:21 um and I've never had Bard really do
22:23 that but I have had Bard like if I asked
22:25 Bard for a series of hyperlinks
22:27 sometimes it'll just go no I can't do
22:29 that
22:30 so like it's still got weird safety
22:32 guard rails but it's it's decent and the
22:35 reason I have like three of these here
22:37 if you like but Kyle you know if you're
22:39 such an expert why don't you just
22:41 recommend one
22:43 I'm sure that's how you all talk
22:45 [Laughter]
22:48 um
22:49 the the challenge with these things is
22:51 because the technology is so new
22:54 um well I thank you digital gods and
22:56 thanks for the images you created by the
22:58 way I don't know if you're in here when
22:59 I when I uh gave thanks for
23:02 um
23:04 which one
23:05 that one
23:06 that's cool
23:08 um
23:11 what was I talking about Google bard oh
23:16 like why don't I just recommend one of
23:18 these which one's the best it depends it
23:20 depends a lot
23:22 so chat if you just want to do writing
23:26 is the best gpt4
23:28 if you want to do any sort of data
23:31 analysis code interpreter or Advanced
23:33 Data analysis they call it now is
23:35 there's nothing else like it other than
23:37 there's there's an open source version
23:38 of that now called open interpreter
23:41 which is really interesting
23:42 it's also got plugins but then Claude
23:45 while while it's it's writing isn't
23:48 necessarily as sophisticated as gpt4s
23:51 and it you know and it's it's answers
23:53 are sometimes a little
23:55 a little more vanilla
23:57 um
23:58 it's got a hundred thousand token
24:00 context window which basically means 75
24:03 000 words and just to put that in
24:05 context The Great Gatsby is a 50 000
24:07 word novel
24:08 so you could literally copy and paste
24:10 the entire text to The Great Gatsby into
24:12 Claude and have it right uh an epilogue
24:17 or add a chapter between 12 and 13.
24:23 and it'll do it so
24:26 um so that's Claude and then and then
24:28 Bard is connected to the internet
24:30 neither of these two are connected to
24:32 the internet although chatbt sort of is
24:34 with plugins but it's quasi-connected to
24:36 the internet they got rid of its
24:38 connection to the internet about three
24:40 months ago now and I don't know if it's
24:42 coming back I don't think it is
24:44 Bing should have more debates with Siri
24:48 it was a good one last night oh my God I
24:53 there's there's got to be a way
24:57 you could probably almost do this with
24:59 Poe
25:04 trying to think how you could do it
25:10 you could do it you could do it with
25:12 zapier you could do it with an
25:13 automation tool
25:15 where actually you know what you know
25:17 what would be a really interesting
25:18 automation tool oh there's there's
25:20 actually there's sites out there that do
25:21 this now is you put in a single prompt
25:23 and it'll it'll spit out how all the
25:26 different models respond
25:28 I'm just kind of doing that ad hoc right
25:30 now but but anyway yeah all right
25:37 Days of our live streams that's good
25:44 um
25:45 [Music]
25:49 by the way if you're new here welcome
25:52 this is the AI learning lab where's my
25:54 little sign
25:55 there it is
25:57 um
25:58 if you have questions about AI pop them
26:00 in the comments below I'll do my best to
26:02 answer them if they're good there's some
26:05 mods in here we'll pin them so I can see
26:07 them I try to go through all the
26:10 comments and answer as many as I can
26:11 which means I get behind by definition
26:13 chechi PT is my bud awesome Tick Tock
26:17 started that in pandemic I don't know
26:19 what they started what throttling people
26:23 um they've been doing that over 10 to 15
26:26 years the movie Her
26:28 goat I'm so happy you're here is that
26:31 the dark side of the moon or the Light
26:33 Side of the Moon
26:36 um
26:38 I'm fine getting email from you Eric
26:41 um
26:43 I can't always respond right is there a
26:45 way to
26:46 is there a way to integrate any of the
26:48 platforms into smart speakers
26:52 yeah with um
26:59 I haven't done any Siri development or
27:01 Alexa development I know I know they
27:03 both have sdks
27:05 um but there's probably with like with
27:07 uh I forget what they're called they're
27:09 not called automations in they're called
27:11 shortcuts in in iOS or in Mac OS
27:17 um I would I would think with shortcuts
27:19 you can do that because I know on the
27:20 iPhone you can you can create a set of
27:23 shortcuts where by saying a certain word
27:26 or hitting a certain key combination you
27:30 can throw
27:31 Siri into chat GPT mode
27:34 um I set it up once and it sort of
27:36 worked but then I forgot what my code
27:38 was and I just I just haven't gone back
27:40 in like 80d took over
27:42 because I figure Apple's gonna do that
27:44 at some point anyway so you know with
27:46 all this stuff like I this is my problem
27:50 with all the open source large language
27:52 models right now is stuff is changing so
27:54 fast in that world that I just don't
27:56 have the attention span to just keep up
27:58 with it you have to keep up with it you
28:00 know on a regular basis because things
28:02 keep changing so I'm just I'm just not
28:05 next I'm not in that space you know I
28:06 don't have that space I don't have that
28:08 space if you know what I mean you got to
28:10 create space in the universe you know
28:13 what I'm saying you create space in the
28:15 universe and then you'll you allow
28:18 things like open source large language
28:20 models into your Universe my my universe
28:23 is a bit more closed you know what I'm
28:24 saying
28:25 when the planets get out of retrograde
28:28 then I can begin to open my doors to let
28:32 the light of the universe in you
28:34 understand what I'm saying no no yeah
28:37 that's a shame well here's really what
28:39 it comes down to
28:41 you
28:46 hey you didn't think I was gonna
28:47 remember that tonight did you
28:50 or were you or whatever the hell I
28:52 started that sentence as
28:54 what about an aicrm I'm bad at keeping
28:57 track of clients and generate sales
28:59 ideas yeah so uh Salesforce needs Ai and
29:05 and they just released a bunch of AI
29:07 tools I haven't used them all yet but
29:09 what Salesforce needs is a box just like
29:13 Chet gbt
29:15 and it's like
29:17 having a call with so and so
29:20 listen
29:22 have the call with so and so it takes
29:24 the notes it generates the account it
29:26 generates the contact it generates the
29:28 lead it puts the note in the thing and
29:31 then it sends you an email that it's all
29:32 [ __ ] done with the link to it
29:36 um
29:38 I know I've gotten emails from
29:40 Salesforce saying
29:42 um we've got AI tools now
29:44 but they just need one they need the one
29:47 where you just like give it [ __ ] and it
29:48 goes in like like
29:51 I input enough [ __ ] and our sales
29:54 folks input enough [ __ ] into Salesforce
29:57 on a daily basis that it could learn
29:59 what we do in three days
30:03 and then it should just learn that and
30:06 then go do it
30:08 did you paint that gold no I bought that
30:11 gold look
30:13 um
30:16 I would have painted it gold had I had
30:19 there not been one you know the red one
30:21 the red one is is Staples Staples kind
30:24 of owns the look and feel but not the
30:25 gold one
30:27 the gold but this is about money this is
30:29 about making money with chat GPT
30:32 you know
30:34 you know what I'm saying
30:36 you picking up what I'm putting down
30:40 [Music]
30:41 [Laughter]
30:43 [Music]
30:47 at all moderator depend things should
30:50 have more debates with Siri saw that one
30:54 is there a way to integrate oh this with
30:56 smart speakers we're ten thousand now
30:59 we're at eleven thousand
31:01 I just love the way Bing chat was
31:03 listening last night and got offended
31:05 and shut you down I know isn't that
31:07 crazy Winston because it was weird
31:09 enough that that when I said you know
31:11 make this image Siri was like here's
31:13 your images on the internet because like
31:15 I hadn't invoked
31:18 that thing in any way
31:21 everybody heard it and just jumped in
31:24 and then yeah when I when I called it a
31:27 robot and I said oh it made those images
31:28 are shitty
31:31 like why I hope I didn't offend you or
31:34 disappoint you and then I'm not a robot
31:38 goodbye
31:42 this is weird babe it's very weird
31:47 um this is an SDK of an AI version of an
31:51 STD
31:51 [Laughter]
31:55 yes
31:57 yes we we will have we will have chat
32:00 GPT sdds absolutely that's coming
32:04 after last night they clearly have
32:06 neuroses so I I I I think that here's
32:11 the challenge so so so what's Samuel and
32:14 what everyone's tried to do right now is
32:15 Achieve AGI right which is is robots
32:18 that have have higher intelligence than
32:21 humans but apparently because they're
32:23 gonna build it based on the way humans
32:25 interact we're just gonna replicate all
32:28 of our neuroses and [ __ ] insanity and
32:32 so so
32:34 so until last night I thought we were
32:37 gonna get a nice you know reprise from
32:40 Human insanity and we would just have
32:42 you know nice supportive golden
32:44 retriever-like Bots that all they're
32:46 there to do is support us but no no
32:50 don't don't tell Bing that you think her
32:52 images are shitty
32:54 she will shut that [ __ ] down and do not
32:57 call her a robot thanks for that what is
33:00 that cool thing is that a bird that was
33:02 cool
33:03 that's neat oh oh Silver Fox just hit
33:06 Level 13. by the way for those of you
33:09 that are new here haven't been here
33:10 before
33:11 this is a competition okay all right
33:15 and Silver Fox is at level 13. so if
33:17 you're anywhere less than that yeah
33:20 I'm not saying you should feel shame I'm
33:23 not I'm just saying it is a competition
33:27 [Music]
33:31 Amelia's wife am I late we got we gotta
33:34 make so when some of you are regulars
33:36 make Emilio's wife a nice image like
33:38 this that says am I late maybe it's got
33:41 like a clock flying through a window or
33:44 the most phallic of all of all the uh
33:47 the the gifts or whatever those things
33:50 are stickers thank you for that it makes
33:52 me uncomfortable every time oh I do
33:54 declare I feel I feel a little bit of
33:57 those Shivers here
33:59 rather you're wrong Dr Brewster I'm very
34:02 proud to be a woman
34:04 uh
34:05 that was very cryptic I don't know what
34:09 I said but we just got out of Mercury
34:11 retrograde I know there were seven
34:14 planets in some form of retrograde for
34:16 the first part of September wasn't the
34:18 first part of September [ __ ] weird it
34:20 was weird man
34:21 all sorts of weird
34:23 crap that I was not happy with
34:28 um
34:30 Winston already laughed all day because
34:33 because of the bot Face-Off oh my God me
34:36 too every time I think about it it just
34:38 I just I I it's it was just the most
34:41 bizarre thing I've ever experienced
34:44 [Music]
34:46 oh Tick Tock ah
34:51 Tick Tock thinks they should be at the
34:53 bottom of the comments just as I was
34:55 sort of getting my [ __ ] together with
34:57 where I was in the comments now just
34:58 jump me to the bottom
35:04 have you seen Beetlejuice at the Buell
35:07 Theater no but Lauren bobert did well
35:09 part of it
35:10 after after she got felt up
35:14 she she didn't see the rest of it after
35:17 that apparently
35:18 foreign
35:21 ERS just completed 1K
35:25 oh wait
35:27 now I'm way far back Days of our live
35:29 streams
35:35 oh I'm trying to get caught up people
35:38 just leave me alone stop picking on me
35:41 I prefer using three all of them are
35:44 different
35:45 oh Nancy talking about the three
35:47 different models yeah so if you're new
35:49 to chat GPT uh if you're new here by the
35:51 way go to chat.openai.com that's the
35:55 official Chachi BT website I'm not
35:57 affiliated with it I just it's the best
35:59 and go there and start playing with it
36:01 and then Claude and Bard are both uh
36:04 things that are worth playing with
36:06 um and yeah what Nancy's saying is these
36:09 things all have different strengths and
36:11 so
36:13 don't just rely on one I I actually
36:15 broke down and I paid for the Claude
36:17 monthly subscription today because I ran
36:20 out of they definitely throttled down
36:22 the free version to like I don't know
36:26 maybe seven or ten
36:28 long chats because
36:31 I used to be able to do 20 or 30 of them
36:34 and now it's like 10 so they definitely
36:36 they definitely lowered uh
36:39 the amount of stuff you can do for free
36:41 with Claude so I paid for it which makes
36:43 me a little bit cranky but whatever I
36:45 need it and nothing else does it nothing
36:48 else does a hundred thousand token limit
36:50 right now
36:58 oh did it just do it again oh my God
37:01 tick tock
37:04 oh
37:06 [Laughter]
37:15 [Music]
37:28 um
37:31 [Music]
37:41 I can't I cannot find where I was I just
37:44 left the way okay we did that one
37:48 we did that one
37:53 related or unrelated the metaverse has
37:56 died to make way for AIA or nay
38:00 um
38:02 what's pi going to evolve into custom
38:05 instructions could make it even more
38:07 useful okay good question Isaac
38:10 and what was the what was the one I was
38:11 just answering oh related or the
38:13 metaverse has the metaverse died
38:16 to make way for AI
38:22 I think the metaverse
38:24 needs AI to survive needs AI to be born
38:29 um
38:32 what what Zuckerberg proved to me
38:37 is that you can't just throw you you
38:40 can't take a triple a gaming approach to
38:43 creating the metaverse
38:45 meaning
38:47 you know he threw he changed the name of
38:49 the company and he threw massive amounts
38:53 of resources to modeling worlds
38:56 and he essentially recreated Second Life
39:01 right so so you look at you look at the
39:05 metaverse that Zuckerberg created and
39:07 you're like
39:08 um
39:09 didn't we have this in like 1999
39:13 like right we did right like this is the
39:16 same thing isn't it
39:18 what what became clear to me is
39:22 the the a couple of things hit me one
39:25 was just the
39:27 the dramatic
39:29 deja vu
39:31 of the metaverse they created feeling
39:35 like second life like it was just like
39:36 this awkward empty place where people
39:39 just stood around awkwardly like they
39:41 were at a junior high school dance and
39:43 don't know what to say to each other
39:44 because it's it's just [ __ ] weird
39:46 like okay yeah we've experienced that so
39:49 that so that kind of happened
39:51 and then Apple
39:54 being Apple
39:56 you know launches The Vision Pro or
39:58 announces The Vision Pro things
40:01 and completely redefines
40:03 the metaverse is this thing called
40:05 spatial Computing where where it's got
40:08 functionality it's not just a place to
40:10 go and hang out and be cool and look
40:12 there's a virtual whiteboard so you can
40:14 be awkward in front of a virtual
40:16 whiteboard too no like there you know
40:19 Apple was like there's this whole new
40:22 way to interact and you can sort of dial
40:24 in the level of immersion you want and
40:27 they didn't use the word metaverse once
40:30 and they didn't use the term AI once
40:33 and when they just launched the iPhone
40:35 15 they didn't use the term AI
40:38 in that either
40:41 so what apple is cleverly doing is
40:44 they're they're navigating around the
40:47 hype
40:48 and they're building a very deep
40:50 Foundation
40:52 where
40:54 on device with Apple devices you're
40:57 going to have enough machine learning
40:59 capability to run very very powerful
41:01 machine learning models
41:03 once you have that
41:06 you're going to be able to invoke
41:09 3D worlds and enter them
41:12 and it won't necessarily be like I'm
41:15 gonna invoke a 3D World and go into a
41:18 game although that could be one thing I
41:20 do
41:21 but it might be I'm thinking about how
41:24 to design a house and you know it I I
41:27 invoke my room I just I literally just
41:31 spin my phone around the room it 3D
41:33 models the room and then they've got to
41:35 deal with Ikea and with
41:38 um Martha Stewart Living and whatever
41:40 and I can just go pick furniture and
41:42 plop it in my room and I can try paint
41:44 on the wall
41:45 and it's going to be a much more
41:47 personalized kind of thing so
41:50 so I think the metaverse re-emerges
41:54 in a in a hyper personalized world where
41:57 AI that we get increasingly comfortable
42:00 with
42:02 asking generative AI tools for what we
42:05 want and generative AI tools give them
42:09 give us what we want and and once you
42:12 can start generating 3D worlds
42:14 dynamically Then I then I think I think
42:17 it it kind of more naturally emerges
42:19 rather than an offshoot of gaming
42:21 because right now the offshoot of gaming
42:24 is just it's it's too Niche and it's too
42:26 shitty it's just not it's just we
42:29 haven't figured out what that's supposed
42:31 to look like yet
42:33 um
42:34 the other thing was what's the future of
42:36 Pi and I think these are kind of related
42:39 um so PI right now if you if you go play
42:42 with Pi Pi dot a i you can go to the
42:44 website but I think the more interesting
42:46 thing to do is go get the app
42:48 there's an IOS app I think there's also
42:50 an Android app
42:52 and put it into voice mode
42:55 and just start talking to it so so
42:57 effectively what you're doing is you're
42:59 having a conversation with a chat GPT
43:01 like thing that's connected to the
43:03 internet
43:03 and it is programmed to have
43:06 conversations it's really good at having
43:08 conversations
43:09 and so
43:11 what I look at pi
43:13 and and sort of play out what it's going
43:15 to be in the future is it's it's going
43:17 to be an agent so so inflection the
43:20 company behind Pi they just raised 1.3
43:22 billion dollars
43:25 950 million of that they're using to buy
43:27 22 000 h100 gpus so they're spending
43:32 almost a billion dollars on a GPU Farm
43:35 to make their models stronger and to be
43:37 able to run
43:39 you know their their models quickly
43:41 right so
43:43 um so that's remarkable in and of itself
43:45 but their mission is to create the
43:48 world's best personal assistant I think
43:51 they're going up directly against Apple
43:54 and if I were a betting man if they're
43:57 if they're good and do it really well
43:59 Apple could end up buying them like like
44:02 a a clear path for those guys to exit is
44:06 is Apple
44:11 um and so what's gonna happen by turning
44:13 them into agents
44:15 not only will you be able to talk to it
44:17 but it will be able to take action on
44:19 your behalf right so hey is there any
44:21 interesting music playing tonight yeah
44:23 there's lots of interesting music in
44:25 Denver and this is playing here and
44:27 that's playing here and it's Latin you
44:29 know appreciation month and there's all
44:31 sorts of things happening in Aurora and
44:34 you're like oh that's cool how much are
44:36 tickets for these three events well this
44:39 one is this much and this one is that
44:41 much and that one is that much would you
44:42 like me to buy you any tickets yeah I
44:44 think that might be nice okay how many
44:46 people are gonna go it's just gonna be
44:48 me or it might be three of us or four of
44:50 us
44:51 hey while you're at it can you check and
44:53 see if if Steve and Lena are available
44:55 tonight and it'll text Stephen Lena and
44:59 say hey I'm thinking of going to see
45:00 this music thing do you want to go with
45:01 us
45:02 and if they say yes it'll just
45:04 automatically buy them tickets
45:07 so I think that's where these things go
45:10 and so how that's tied to the metaverse
45:12 thing is
45:14 you know
45:16 you know I'm thinking about
45:19 I'm thinking about putting a fence
45:21 around my yard
45:24 um
45:26 what are some of my options
45:29 well why don't you walk outside and
45:32 you know
45:35 I'll fire up a camera here and then you
45:37 just scan your yard
45:40 and you know I'll measure it and I'll
45:43 and I'll turn it into a 3D model and
45:46 then I'll install different fencing
45:48 options around your thing and then oh by
45:50 the way did you want me to to include
45:53 some landscaping ideas in there as well
45:55 okay because here's a little rock pond
45:57 with with koi koi fish in it I hate
46:02 Quake ponds oh okay then let's make it a
46:05 barbecue pit and and a little horseshoe
46:08 pad yeah that's great could you make a
46:10 bocce ball instead of horseshoe sure
46:12 here you go
46:15 do you want me to find a landscape
46:18 architect who can actually build that
46:19 for you no no I'm just gonna get it at
46:22 Home Depot okay would you like me to
46:23 pre-order all of this stuff for you at
46:25 Home Depot so you can just go pick it up
46:26 or would you rather have it delivered
46:28 right like this is this is where this
46:32 stuff's headed
46:35 um where where it's it's
46:38 I I think here's the problem with the
46:39 metaverse
46:41 as it as it's currently being thought
46:42 about is it's very similar to chat GPT
46:45 right now where Chachi BT
46:48 is this thing off to the side right
46:51 you've got your life and your computer
46:53 and everything's good and then there's
46:55 this magic [ __ ] tool somewhere
46:57 somewhere else that everyone's talking
46:59 about you're like what's this chat gbt
47:00 thing
47:03 what are they talking about right and
47:05 you have to go over to this other part
47:07 of your world your digital world and you
47:10 have to learn about prompt crafting and
47:11 prompt engineering and hallucinations
47:14 and what do you mean it lies and why
47:16 doesn't that hyperlink work oh that's a
47:18 hallucination too I don't understand
47:20 this why is this so hard
47:23 because everything right now is not
47:25 integrated everything is it's like the
47:26 early days of the World Wide Web it it
47:29 everything was sort of piece parts right
47:31 and it took in the case of the World
47:33 Wide Web it took
47:36 I don't know 15 years before the
47:39 infrastructure got built and the tools
47:41 got assimilated and social networks
47:43 showed up and mobile devices showed up
47:45 like it took a fair amount of time for
47:47 the World Wide Web to get integrated
47:49 enough into our daily life that we
47:52 didn't think about it anymore it wasn't
47:54 like a thing we thought about it's just
47:56 there it's there when you do your
47:58 business you don't think about the
47:59 internet you don't think about
48:00 hyperlinks you just click on [ __ ] or tap
48:03 [ __ ]
48:05 right that all came from this very
48:07 simple idea of a hyperlink a single word
48:09 with a blue underline that if you do
48:11 this takes you to another document
48:13 that's how it started
48:15 chat EBT is the equivalent of web 1.0 in
48:19 1994.
48:21 and it's very much off to the side as
48:23 these tools get more integrated in
48:25 everything we do
48:27 and as they get multi-modal this is a
48:29 term that you'll hear a lot and
48:31 apparently Gemini coming from Google is
48:33 going to be multimodal what's that mean
48:35 well you'll be able to talk to it you'll
48:37 be able to send it pictures you'll be
48:39 able to have it make pictures make
48:42 videos make audio synthesize your voice
48:45 and make a phone call on your behalf the
48:48 multimodal thing as this gets more
48:50 integrated all of the tools that we use
48:53 will have some aspect of of this
48:56 generative AI stuff within it all right
48:59 someone make this an image Tick Tock
49:01 just jumped me to the bottom
49:05 I I resemble that remarkabrina
49:10 what are you gonna do
49:13 I you know yeah like we need to start a
49:15 marketing campaign for tick tock to fix
49:18 their bugs like like don't have a
49:21 printful integration if you don't
49:23 support printful that's one
49:25 um fix the jump to the bottom bug
49:28 um if you're gonna Shadow ban an account
49:30 maybe tell them that you've done that
49:33 and why
49:35 it's just weird man in weird man you
49:38 know you know man I get on here and I
49:41 expect them to behave a certain way and
49:43 they don't behave that way you get what
49:45 you pay for them all right because I
49:46 don't pay anything for this that's what
49:48 I mean you get it you get it you
49:50 understand what I'm saying
49:52 [Music]
49:54 all right
49:58 um
49:59 Isaac oh yeah and custom instructions
50:01 could make pie much more useful
50:04 um
50:06 Pi's going to get much more
50:08 sophisticated
50:11 my guess is
50:13 beginning of the year
50:15 they just raised a billion dollars so
50:17 they're
50:18 they're sourcing gpus right now and
50:21 they're they're hiring people right now
50:22 and they're probably you know
50:25 architecting their their GPU server
50:28 Center
50:29 um
50:31 and then they're probably concurrently
50:33 getting their engineering teams up to
50:35 speed on on whatever Pi is going to be I
50:38 I would think that
50:41 Pi having the ability to take action on
50:45 your behalf
50:46 do a Google search or you know do do
50:48 well it's already kind of doing Google
50:50 searches but maybe maybe things like you
50:53 know tie in your travel accounts or your
50:55 your you know your bank accounts things
50:57 like that
50:58 feels like a next step for it and then
51:00 some type of personalization so whether
51:03 it's training on your data or it just
51:05 remembers things longer I don't I don't
51:07 know but I think I think Pi will get
51:10 much better probably
51:13 January to March is my guess
51:18 um I paid for Clyde Claude too Kyle
51:21 you're not alone I'm an admin of a 35k
51:23 AI content creators group oh that's
51:25 awesome that's awesome cool so so check
51:28 out the
51:29 um AI Salon because we do we meet every
51:31 other Tuesday we're smaller than that
51:34 there's probably about
51:36 seven or eight hundred in the Meetup and
51:38 then probably
51:40 I don't know 500 or so in the Discord
51:43 um but yeah we should chat drop drop me
51:45 a uh
51:46 drop me a DM and I'm happy to chat
51:48 because I listen any
51:51 this channel exists to talk about this
51:54 [ __ ] I think anything we can do right
51:57 now to talk more about this stuff and
51:59 and create Community around this I think
52:01 it's actually really important
52:03 so
52:05 this is funny because it's true meta's
52:07 metaverse is just just the second second
52:10 life it is it's [ __ ] second life it's
52:13 just as awkward it's it's not better or
52:17 worse graphic it's just Graphics it's
52:19 just different Graphics right there is
52:20 slightly more sophisticated with the way
52:22 they handle light
52:24 but it's just they're they're like like
52:26 some of the spaces are beautiful and
52:29 impressive and then you're like and what
52:31 do I do here well you can invite friends
52:33 in to look at your the art on your
52:35 gallery walls okay thanks Robert
52:38 anyway okay and then
52:40 and what do we what do we do when the
52:44 people show up to this little Gallery
52:45 thing well you can you can dance
52:48 okay so this is Junior High this is as
52:51 awkward as Junior High Gatherings is
52:53 that what I'm getting here
52:58 you know and then and then you look at
53:00 at what Apple did
53:03 and
53:05 you know Apple recognized that having on
53:09 [ __ ] ski goggles is weird
53:12 right that it's not normal
53:16 and that it's isolating
53:18 and so what did they do well they put
53:22 computer monitors on the outside of the
53:24 eyeballs so they could project your
53:27 eyeballs
53:29 on the outside of the thing so if
53:31 someone walked into your field of vision
53:34 they could see your eyes they could see
53:36 that someone was in there and it the
53:39 goggles let the person outside the
53:41 goggles know that you can see them
53:45 okay good first step human beings exist
53:48 together what else did they do when
53:51 you're on a zoom call in The Vision Pro
53:54 they've got four cameras pointing at
53:56 your eyes and six cameras pointing down
53:59 or there's like 18 [ __ ] cameras in
54:03 this thing or sensors right
54:05 so one of the first things that you do
54:07 when you get the Vision Pro goggles is
54:09 you flip them around backwards and you
54:10 aim them at your face and it 3D scans
54:13 your face so so built into your account
54:16 is a 3D scan of your face so when you
54:19 get on a zoom call
54:22 you you couldn't you can't be sitting in
54:24 front of
54:25 a webcam because you've got weird
54:27 [ __ ] goggles on right
54:30 so what do they do they synthesize your
54:33 face
54:34 they already have cameras on your eyes
54:36 so your eyes are blinking in real time
54:38 they've got sensors sensing movement up
54:40 here they've got sensors you know
54:43 sensing movement down here
54:45 and so when you're on a zoom call you
54:47 see the other people kind of in the room
54:49 and then your head's there as well they
54:52 see you your full face talking with all
54:55 your
54:57 emotions without the goggles on
55:01 brilliant
55:03 right
55:04 it's not we go into the metaverse it's
55:07 that we're interacting with computers in
55:10 this new kind of spatial 3D way
55:16 and it can be a hybrid augmented reality
55:19 version of the world where I see my room
55:21 and then I can just dial that I think
55:23 they have a dial somewhere I forget
55:25 where it is where was it I don't know
55:29 but the dial basically dials out reality
55:34 so you want to go have your you know
55:36 watch a movie in the mountains you just
55:40 [ __ ] dial out your living room and
55:42 now you're sitting in the mountains
55:44 there's a giant screen projected
55:47 in front of the mountains
55:50 and you don't have things in your ears
55:54 there are
55:56 speaker arrays that shoot sound into
55:59 your ears
56:01 why is that because there's other [ __ ]
56:03 happening in the world that needs to get
56:05 into your ears as well right
56:09 so like the technology in those things
56:11 is brilliant but but it's really
56:13 conceptually
56:14 what Apple's brilliant at is
56:18 understanding
56:19 this is a new
56:21 it's a new paradigm it's like a
56:24 completely new thing
56:26 it's not just a better version of second
56:28 life and I think that's why they didn't
56:30 use the word metaverse once and I think
56:32 that's why they're not using AI once it
56:35 should just feel like [ __ ] Magic
56:40 when you're when you're in playing some
56:43 sort of game in this interface and
56:45 someone walks in your field of view
56:48 in the environment they materialize
56:52 it sort of interrupts your 3D space and
56:56 says there's actually a human being in
56:57 front of you do you want to deal with
56:58 them or not I mean that's [ __ ]
57:01 brilliant so anyway
57:03 yeah
57:05 Google Glass was early bird barred on
57:08 glass would be incredible in 24 25 yeah
57:10 I think we'll start to see all that
57:12 stuff I think Apple's gonna lead the way
57:13 because I think right now Apple's the
57:15 only one that I've seen that actually
57:17 understands that you know
57:22 some version of what the metaverse is
57:26 um I I don't I I just I don't think I I
57:29 just don't think Zuckerberg's
57:31 got the right model at all
57:34 they because because they took a
57:35 technology approach they didn't take a
57:37 human approach they didn't take a
57:38 human-centric approach
57:41 and that's where I think Google's gonna
57:43 struggle as well Google's gonna have
57:45 good technology
57:49 but
57:52 you know Apple prioritizes Technology
57:54 based on
57:55 humans and Design
57:58 Facebook and Google or meta and Google
58:01 don't do that
58:03 maybe used for renting or buying
58:05 properties absolutely what are your
58:07 thoughts on the concept of time
58:10 I don't know
58:12 that we're just a
58:16 a fart in the in the Winds of History
58:20 we're just here for a bit we're just
58:22 here for a bit
58:25 we should do what we can to make it suck
58:27 less
58:29 and then and then we go your content is
58:32 always great thank you Simon Says Tech
58:34 uh
58:35 I'm gonna go 10 out of 10 on the name I
58:38 like the name a lot Simon Says Tech I
58:40 like it it's solid
58:44 um share Bears here what's happening
58:47 oh here comes champion
58:51 come on buddy
58:55 come on
58:57 no he doesn't want to be in here all
58:59 right it happens
59:03 um dang nobody's been talking about
59:05 Apple's goggles
59:08 what do you mean
59:11 because I just did
59:13 you not want to hear about them I almost
59:15 forgot about them yeah I mean so
59:18 I mean the thing about them is oh thank
59:21 you for the the Roses many many roses
59:24 I've got a bouquet of roses from wood
59:29 um thank you very much I appreciate that
59:32 um
59:33 yeah so so
59:36 you know this is the way Apple kind of
59:38 rolls right
59:40 when they break into when when they when
59:43 they Define a new category which is what
59:45 they
59:46 I I don't know
59:49 I was
59:50 I was fairly mind blown by the The
59:53 Vision Pro
59:55 launch
59:57 not so much because of the technology I
1:00:00 kind of knew the technology would be
1:00:02 some version of what it was there was
1:00:03 absolutely some surprising things in
1:00:05 there like it's incredibly impressive
1:00:07 and apparently there's like 9 000
1:00:09 patents in or 5 000 patents some crazy
1:00:13 amount of patents in those goggles it's
1:00:15 been it's been in development for nine
1:00:17 years
1:00:18 the thing that impressed me most was
1:00:21 that I could not for the life of me
1:00:23 figure out how Apple was going to
1:00:25 separate itself from meta how it was
1:00:28 going to separate itself from the
1:00:30 metaverse and that was the single most
1:00:32 impressive thing to me
1:00:34 so so when Apple does this what they'll
1:00:36 do is they'll introduce an entirely new
1:00:39 category
1:00:41 essentially a year early
1:00:45 and then six months later give access to
1:00:47 the developers to the tech which is sort
1:00:50 of how it's rolling so the thirty five
1:00:52 hundred dollar price tag isn't really
1:00:54 for people it's for it's for Developers
1:00:58 and so the developers are going to
1:01:00 develop cool [ __ ] on it and so
1:01:04 next June at the developers conference
1:01:07 they'll actually have demoable stuff
1:01:09 like actual stuff the price of goggles
1:01:12 you know Vision Pro 2 will come out
1:01:16 and then
1:01:18 by the third or fourth iteration of
1:01:20 those goggles that's when
1:01:23 the rest of humanity goes oh I won't
1:01:25 want those one of those things so it
1:01:28 makes sense that that you wouldn't
1:01:30 necessarily be thinking about it because
1:01:31 no one no one's talking about a
1:01:34 completely new paradigm yet because the
1:01:37 infrastructure is not there Apple's got
1:01:38 to build all the relationship with the
1:01:40 developers and and build the stuff and
1:01:42 get the goggles you know production
1:01:44 ready and the 3500 Buck version of these
1:01:47 is just the one that like you know Geeks
1:01:50 will be able to say oh I got a version
1:01:53 one of the The Vision Pro goggles
1:01:55 because like you know version four is
1:01:57 gonna look like this with maybe like a
1:01:59 little battery pack on the side they're
1:02:01 not gonna you know they're gonna get
1:02:03 very normal very very fast
1:02:06 when when they have those little
1:02:08 augmented reality chips or like they all
1:02:11 live within the bridge of your nose and
1:02:13 they've got some sort of microwave
1:02:16 microwave connection to your frontal
1:02:19 lobe you know that'll be version four
1:02:21 that's that's when people buy them you
1:02:23 just put them on you're like oh these
1:02:24 are [ __ ] Magic
1:02:25 so Facebook's a different deal because
1:02:28 they don't make consumer devices like
1:02:30 apple yeah exactly and and they just
1:02:32 even though they're a social network
1:02:36 they're not they're a data company and
1:02:38 they're an advertising company they
1:02:40 really don't give a [ __ ] about human
1:02:41 beings they give a [ __ ] about human
1:02:43 beings behavior and predicting that
1:02:46 behavior and being able to sell that
1:02:48 predictable Behavior to advertisers
1:02:51 that's not what Apple does
1:02:54 Apple's like how do we deliver good
1:02:57 products for human beings and Google
1:03:00 doesn't do it Google and Facebook are
1:03:02 very very similar they're data companies
1:03:05 when the product is free you're the
1:03:08 product
1:03:11 every time every time
1:03:19 all right that was a rabbit hole and a
1:03:22 half I didn't even make any funny silly
1:03:24 voices oh thank you Sherbert it's very
1:03:26 sweet
1:03:34 what are your thoughts on the concept of
1:03:36 time who has that healthvenom.com
1:03:41 shop HV slash gentle that was a pretty
1:03:44 funny question I think that was a troll
1:03:46 I like it
1:03:48 is it helping in building itself what
1:03:51 all of these things all of the all of
1:03:53 the large language models not right now
1:03:55 although thanks Stevo held a cowboy
1:03:59 um
1:04:00 they will like right now they're not
1:04:03 sophisticated enough I mean they're
1:04:05 they're doing things like
1:04:08 there's a lot of researchers using chat
1:04:12 GPT like tools to generate synthetic
1:04:15 data to train large language models to
1:04:18 get better and better
1:04:20 there's one camp that says synthetic
1:04:22 data is going to lead to a thing called
1:04:24 Model collapse where the tools get
1:04:27 better and better and better and then
1:04:28 just fail
1:04:30 and there's another camp that says
1:04:32 synthetic data is the way to the Future
1:04:33 because humans humans contributing to
1:04:37 these things are a bottleneck because
1:04:40 we're [ __ ] slow and inefficient and
1:04:42 they're not so I don't know I don't know
1:04:45 where that goes that's that's way beyond
1:04:47 my capabilities and understanding
1:04:51 other than paying attention to it
1:04:56 that's a lot of pie yeah I've I've I
1:05:00 have various AI groups yeah Nestor
1:05:03 um
1:05:04 I think it's good like you should be in
1:05:07 once you sort of once you sort of red
1:05:10 pill blue pill this AI [ __ ] and this is
1:05:12 this is my thought so if if you've been
1:05:14 here before you've heard this many many
1:05:16 times but if you haven't
1:05:18 I think it's in your interest just as a
1:05:21 human on this planet for what's coming
1:05:24 to use chat GPT or something like it
1:05:27 long enough until you have your Kevin
1:05:29 McAllister moment which is this moment
1:05:34 I didn't know I could do that
1:05:38 and and that's usually something beyond
1:05:41 the parlor tricks right you get it to
1:05:43 write a country song you get it to write
1:05:44 a poem you get it to write this you get
1:05:46 it to write that
1:05:47 the moment of breakthrough is generally
1:05:50 when you say well but it probably can't
1:05:52 do what I do
1:05:54 and then you ask it to do what you do
1:05:57 and what you learn is that oh [ __ ]
1:06:00 it can do what I do and it can either do
1:06:03 it faster
1:06:04 or better
1:06:06 or both
1:06:10 and and in any of the three of those
1:06:12 scenarios your next immediate
1:06:16 question is well what the [ __ ] do I do
1:06:18 now
1:06:20 that's your red pill blue pill moment
1:06:22 you can't unlearn it and this stuff's
1:06:24 not going to be uninvented
1:06:26 and then you just have to enter The
1:06:28 Gauntlet of okay I'm a copywriter holy
1:06:31 [ __ ] what what do I do I don't know go
1:06:33 learn these tools and get good at
1:06:36 copywriting using these tools because
1:06:38 these tools are going to be in every
1:06:40 writing tool out there they're already
1:06:42 coming it's already generally available
1:06:44 within the Google Office Products it's
1:06:46 called Duet you can have your system
1:06:49 administrator turn it on right now I
1:06:53 hear it sucks because it's built on Palm
1:06:55 2 which sucks but when Gemini comes out
1:06:59 Duets gonna get really good really fast
1:07:03 do I offer internships uh I not really
1:07:06 we do a little bit at storyline in my
1:07:08 company
1:07:09 um although we haven't we haven't had
1:07:10 interns in a while so
1:07:12 um no this is just the learning lab is
1:07:14 just me shooting shooting the [ __ ]
1:07:18 so
1:07:20 um you're welcome to mod or or uh you
1:07:23 know I would I would love to have a
1:07:25 producer doing some social media [ __ ]
1:07:26 but
1:07:28 um
1:07:32 or is it going
1:07:33 oh oh yeah like like
1:07:39 once you understand
1:07:41 that
1:07:43 assuming these things don't go away
1:07:47 assuming they don't get uninvented
1:07:50 which I'm
1:07:53 99.99 sure they're not going to get
1:07:55 uninvented because nothing ever has
1:07:59 they're not going to be controlled meta
1:08:01 has had has made sure of that by open
1:08:03 sourcing their stuff and stability AI
1:08:06 has open sourced their stuff so there's
1:08:08 enough core technology in the open
1:08:10 source Community now that the developer
1:08:12 Community understands how this [ __ ]
1:08:14 Works they might not be able to they
1:08:15 might not have the resources to build
1:08:17 the core
1:08:18 large language models but the developer
1:08:21 Community is getting really good
1:08:23 at fine tuning and making those large
1:08:26 language models more and more efficient
1:08:29 so it ain't going away
1:08:31 and then and then you you you you can't
1:08:35 unlearn it so I think joining different
1:08:38 AI groups and you know like the salon is
1:08:41 is very much about just like exploring
1:08:44 what this stuff is and and you know
1:08:46 talking to people that are doing
1:08:47 interesting things and supporting one
1:08:50 another and we just created a concept of
1:08:52 guilds where we're going to have people
1:08:54 that are interested in certain things
1:08:56 leading sub-communities within the
1:08:58 community that's great Rachel Woods has
1:09:01 her AI exchange
1:09:02 um you know Paul rates or has the
1:09:04 marketing AI Institute
1:09:06 um Nestor here has you know this 31 000
1:09:09 like join them all like whatever your
1:09:12 brain capacity can take
1:09:14 but but try to find communities that you
1:09:17 resonate with
1:09:20 where people don't think you're a
1:09:21 [ __ ] weirdo right there's a lot of us
1:09:25 out here
1:09:27 and we're [ __ ] weirdos but we're
1:09:29 weirdos because
1:09:31 once you understand that these are more
1:09:34 than just a fancy Google search
1:09:37 once you get how profound what's going
1:09:39 on behind that little box
1:09:43 then you kind of understand okay
1:09:46 my job isn't going to look like what it
1:09:49 looks like today and neither is their
1:09:51 job and neither is their job
1:09:53 well now what do we [ __ ] do I don't
1:09:56 know but but what I do know is that if
1:10:00 you're
1:10:01 actively
1:10:03 engaging with what's going on with this
1:10:06 you'll know what's going to happen in
1:10:08 your job before anyone else does and
1:10:11 that's that's the only thing
1:10:13 that's in our power to control right now
1:10:16 because it's just too it's too
1:10:18 ubiquitous of power
1:10:20 that it's not going to affect everything
1:10:25 um which I like as I'm saying it it
1:10:27 probably sounds uh
1:10:31 it might sound depressing but it's it's
1:10:33 also tremendously exciting right like if
1:10:35 you're here's here's the here's the
1:10:37 lucky thing
1:10:41 there is only ever one beginning of a
1:10:44 new technological Revolution
1:10:47 like I happen to start a company in 1994
1:10:50 at the very beginning of the World Wide
1:10:52 Web Revolution
1:10:54 I just missed being part of that when
1:10:58 the PC came out I was too young to be
1:11:00 part of that at the beginning it was
1:11:03 like seven or eight years too late and
1:11:06 it broke my heart and I didn't think I'd
1:11:07 see one of those again in my lifetime
1:11:09 and then the World Wide Web came along
1:11:11 and I was like Ah that's one of those
1:11:13 and then I didn't think I'd see another
1:11:14 one in my lifetime and now this [ __ ]
1:11:16 showed up this is one of those this is
1:11:19 printing press steam engine level [ __ ]
1:11:23 this is going to make the world wide web
1:11:24 look [ __ ] quaint
1:11:27 this is a big [ __ ] deal printing
1:11:30 press steam engine level fourth
1:11:33 Industrial Revolution level stuff
1:11:36 we're at the beginning of it
1:11:40 if you feel like you're very far behind
1:11:42 you're not Chachi PT was launched in
1:11:45 November of last year it's nine months
1:11:47 old
1:11:48 it's the very beginning
1:11:51 the very beginning
1:11:55 and so we're lucky enough to get to play
1:11:57 with this [ __ ] right now
1:11:59 we're gonna be the ones we
1:12:03 if you hang out here for long enough
1:12:06 and and start getting into this stuff
1:12:09 we're going to be the ones that will be
1:12:11 able to say to our kids and grand kids I
1:12:14 knew what it was like before that was
1:12:16 here
1:12:17 what do you mean
1:12:18 well we actually had to read
1:12:23 you know PDF documents what are PDF
1:12:26 documents oh
1:12:29 you know the you know the books you just
1:12:32 know
1:12:33 you know the books you talk to
1:12:36 we actually had to read them and put
1:12:38 them in our brains and interact with it
1:12:40 inside our heads and then we would have
1:12:41 book clubs and talk you couldn't talk to
1:12:44 a book no
1:12:45 wait you couldn't just make the world
1:12:47 that was the book that looked the way
1:12:49 you wanted it to look no
1:12:51 that's weird
1:12:53 it was just words
1:13:01 so yeah uh I I missed someone just one
1:13:05 of the mods just pinned something and I
1:13:07 missed it so pop it back up there I'll
1:13:09 take a look at it
1:13:11 does the law of numbers solve the
1:13:13 alignment problem what's the best AI to
1:13:16 explore right now the the one that I
1:13:18 have up here Chachi BT is the one
1:13:22 um as you get into this I think it's
1:13:24 important to play with multiple models
1:13:26 so that's why I've got those three and
1:13:28 those four Bing
1:13:30 and let perplexity you can play with
1:13:33 perplexity it's it's pretty good
1:13:36 um Pi's interesting actually you know
1:13:38 what
1:13:40 just these three right now
1:13:42 chat gbt Claude and Bard for now are
1:13:45 fine and and really just chat gbt
1:13:49 I firmly believe that if I
1:13:52 if I didn't use any other tool than chat
1:13:54 GPT I could probably spend the next 10
1:13:57 years
1:13:58 understanding what it does or what it's
1:14:01 capable of doing
1:14:03 it's it's it's that
1:14:06 insanely deep
1:14:09 it seems really simple and it seems
1:14:12 really imperfect and it seems really
1:14:14 janky and it is
1:14:17 and yet it is kind of a reflection of
1:14:21 as much of humanity as
1:14:24 they could jam into it in the past you
1:14:26 know five years
1:14:29 it's pretty remarkable
1:14:31 um
1:14:33 I had access to open AI playground like
1:14:35 a year before chat gbt yeah you know
1:14:38 it's funny Nestor I was I was playing
1:14:40 it was play three like when when gpt3
1:14:42 came out and made all that noise and
1:14:44 everyone's like oh there's this there's
1:14:46 this thing it's just like humans and it
1:14:48 writes really good and then
1:14:51 and I went to playground and I was
1:14:52 playing with it and at the time I was
1:14:54 doing a lot of stuff with image
1:14:55 generation and I started getting into
1:14:57 stable diffusion and dream booth and all
1:14:59 that sort of stuff
1:15:01 and and I'd I'd been playing with
1:15:04 chat GPT playground for for a while but
1:15:07 I didn't
1:15:08 um
1:15:11 I didn't really start building on it I
1:15:13 didn't really I got it but I didn't get
1:15:15 it like I got that okay it does a cool
1:15:17 thing and it
1:15:19 but it wasn't until
1:15:22 EBT was launched and
1:15:26 I asked it to write me some python code
1:15:29 and within 90 minutes I had written a
1:15:31 fully functional python application a
1:15:34 sophisticated one that did calls to open
1:15:37 Ai and and wrote social media content
1:15:41 um
1:15:43 that was when it hit me that oh [ __ ]
1:15:45 everything just changed
1:15:48 because it it was you know it's the
1:15:50 difference between when you have access
1:15:51 to playground it's it's kind of like
1:15:53 you're back there geeking and you're
1:15:55 playing around with temperature and
1:15:57 which model and you're reading
1:15:58 documentation and you're reading about
1:15:59 the API and you're getting notebooks up
1:16:02 and running and it and it's just you're
1:16:05 in a different frame of mind when you're
1:16:07 in playground than when you're in Chachi
1:16:09 BT chatgpt is the equivalent of the
1:16:12 World Wide Web where prior to the World
1:16:14 Wide Web the internet was a command line
1:16:18 research tool
1:16:20 and then the World Wide Web came and now
1:16:22 it was a page you could click on
1:16:25 and it's just a it it was
1:16:28 technically almost identical to what was
1:16:32 before but it was fundamentally
1:16:34 different and chat is that for for
1:16:37 machine learning and AI
1:16:38 that even if you had access to the
1:16:40 playground early it was just a different
1:16:43 thing
1:16:44 and it's also it's just a different
1:16:45 audience right Chachi BT got to 100
1:16:47 million users in six weeks it took the
1:16:50 World Wide Web six years to get to 100
1:16:53 million users Chachi BT got there in six
1:16:55 weeks why because it's a profoundly
1:16:57 powerful tool that looks like a Google
1:17:00 search box
1:17:02 I was playing around with the pre-che
1:17:04 EPT the year before yep
1:17:06 oh that's cool yeah neat so
1:17:09 so that that's the difference and like
1:17:12 and and that's the thing we're all
1:17:13 grappling with because now that
1:17:16 non-engineers have their hands on this
1:17:19 thing they're starting to use it in
1:17:20 different ways and imagine different
1:17:22 things and now we're contemplating
1:17:24 things like what's the future of work
1:17:26 and what are companies going to do and
1:17:27 are they just going to lay off all the
1:17:29 employees if I only need two people to
1:17:32 do the work of 10
1:17:33 what do I do with the other eight
1:17:36 do I upskill them and increase my
1:17:39 productivity or cut my prices or speed
1:17:41 up my turnaround time
1:17:43 or do I lay off five of them
1:17:46 I don't know
1:17:48 we're gonna find out
1:17:52 um well my grandkids asked me what was
1:17:54 it like before I think they will
1:17:57 it's it's it's gonna be weird
1:18:00 I mean you know just just like you know
1:18:02 gen Z
1:18:04 you know Millennials were sort of you
1:18:07 know
1:18:09 internet native and gen Z is kind of
1:18:11 mobile internet native
1:18:15 um and and you know young kids today
1:18:18 probably
1:18:20 three four five-year-olds today and
1:18:22 younger
1:18:25 they're only gonna know a world where
1:18:27 you ask the machine for [ __ ] and it
1:18:29 gives it to you
1:18:30 we're entering a world of hyper
1:18:32 personalization
1:18:34 where you just ask for it and it will
1:18:36 give it to you what you want right now
1:18:39 that's not what Google was
1:18:42 Google said here are 30 things well here
1:18:45 are 300 things 30 on a page here are 30
1:18:48 things that might be the thing
1:18:51 that might contain the thing that you
1:18:53 might be looking for
1:18:55 chechi BT says here's your answer here's
1:18:58 your answer to your problem right now
1:19:01 I don't like that answer here's another
1:19:03 one I don't like that answer here's
1:19:05 another one
1:19:06 okay I like that answer but I want it to
1:19:08 be bigger here's it bigger
1:19:13 so imagine
1:19:16 three years from now where that's not
1:19:18 just text and it's not just off to the
1:19:20 side
1:19:21 it's in your phone
1:19:23 and the difference between calling
1:19:25 someone and interacting
1:19:28 with their assistant that talks like
1:19:30 them
1:19:32 that does stuff on their behalf so you
1:19:34 can make a play date
1:19:37 that the distinction between whether you
1:19:40 talk to the person or their Avatar is
1:19:42 actually kind of blurry and it doesn't
1:19:44 [ __ ] matter
1:19:45 kids today that's just going to be
1:19:48 normal to them it's [ __ ] weird for us
1:19:50 and it's going to be weird for us for
1:19:52 probably a long time it will feel novel
1:19:55 for a long long time
1:19:57 that this [ __ ] can read a 200 Page
1:20:02 document
1:20:04 and summarize it like that
1:20:08 I remember when we actually had to read
1:20:10 manuals
1:20:11 remember that
1:20:13 remember the when you bought a car in
1:20:16 the glove box was this
1:20:18 thing
1:20:20 and you had to like flip through it and
1:20:22 find the fuse box
1:20:24 yeah I remember that
1:20:27 now you just ask it how do I determine
1:20:29 which fuse blew you ask your car that
1:20:32 it'll go well based on my Diagnostics
1:20:36 here
1:20:37 looks like few 17s out would you like me
1:20:40 to have that overnighted I can get it
1:20:42 here by five this afternoon with Amazon
1:20:44 Prime
1:20:48 that's just gonna be normal
1:20:52 just like the fact that it's completely
1:20:54 normal for us
1:20:56 that you click on a thing and you're
1:20:59 looking at
1:21:00 information from a French company in
1:21:03 France on your computer well yeah that's
1:21:06 just the internet okay
1:21:09 all right I'm gonna put AI on a
1:21:12 Raspberry Pi and leave them everywhere
1:21:14 that's actually a really cool idea by
1:21:16 the way love that idea
1:21:19 and what kind of interfaces is the next
1:21:21 question right is it a talky interface
1:21:23 is it a point-and-click interface is it
1:21:25 colors
1:21:27 what do you have to do
1:21:29 that's that's really cool idea
1:21:31 what's the first python you wrote with
1:21:34 chat EBT I missed it so I um Lily I um
1:21:38 so sitting
1:21:40 I I had done all all my sort of parlor
1:21:42 tricks which were which were like um you
1:21:44 know write a poem write a country song
1:21:46 like I was doing just sort of language
1:21:47 word games with it
1:21:49 and then then this is like I don't know
1:21:52 I'm like Chad TBT was like two weeks old
1:21:54 or something like that and I'm I'm
1:21:55 watching TV with my wife and she's got
1:21:57 some show on I'm not interested in and
1:21:59 you know the the classic the classic
1:22:02 marriage Trope right
1:22:05 you've got shitty taste in TV no you've
1:22:07 got shitty taste in TV so it was it was
1:22:11 her night to watch whatever the [ __ ] she
1:22:12 wanted
1:22:13 and so so I'm just [ __ ] around with
1:22:15 chat EBT and I was like I had heard oh
1:22:18 it can it can write python code and for
1:22:20 years I've been I you know I've been you
1:22:23 know toying around with the idea of I
1:22:25 should teach myself python one day
1:22:27 because it might be interesting and then
1:22:28 I go to Learn Python and I'm like oh
1:22:30 this is [ __ ] exhausting I have ADD
1:22:32 and I can't I just don't have the
1:22:34 attention span to learn a new
1:22:35 programming language that I'm never
1:22:36 going to use
1:22:37 and
1:22:39 and I said hey can you write me
1:22:42 um some python code that takes an input
1:22:44 and does something with it
1:22:46 and it says sure and it wrote python
1:22:48 code and then I said oh uh I don't know
1:22:51 how to run python code because I'd never
1:22:54 run python code before and I said I
1:22:56 don't know how to run python code is
1:22:57 there a way I can test this and it said
1:22:59 sure go into your terminal and type in
1:23:01 this command line and install these
1:23:03 libraries and I was like uh I'm like no
1:23:05 I don't I don't want to do terminal [ __ ]
1:23:07 like isn't there a website I can just
1:23:09 copy and paste this [ __ ] and it said
1:23:10 sure here's five and the first one was
1:23:13 replit
1:23:14 um which is you know a good python
1:23:16 running
1:23:18 um website
1:23:19 and so I copied the code over into
1:23:21 replit and I hit the play button like it
1:23:23 told me to and it gave me an error and I
1:23:26 said it won't give me this error and I
1:23:27 said well you don't have this Library
1:23:28 installed and I said well I don't know
1:23:30 how to install that library because I
1:23:32 don't know what the [ __ ] I'm doing
1:23:33 and it said in the lower left hand
1:23:35 corner of the page click this button and
1:23:38 then in the upper right hand corner of
1:23:39 the page search for this library and hit
1:23:41 install
1:23:42 and I did and then it ran
1:23:44 and then
1:23:46 um I had it I had an open AI key from
1:23:50 the open AF playground
1:23:53 and I said I want I want to take some
1:23:56 inputs in a form from a user like about
1:23:58 a topic
1:24:00 and then I want to generate a set of
1:24:01 tweets
1:24:02 and so it wrote code to do that and I
1:24:05 just sort of modified some of the
1:24:07 prompts inside the python code I didn't
1:24:09 really touch the python code
1:24:12 and that worked and then I just
1:24:15 said Okay I want to do tweets and a blog
1:24:18 post and it did that and I said I want
1:24:20 to do tweets and a blog post and
1:24:23 whatever hashtags
1:24:25 and so I just did five different social
1:24:27 media outputs
1:24:29 and within 90 minutes kind of
1:24:32 absent-mindedly while I'm sort of half
1:24:34 watching TV it wrote this fully
1:24:38 functional application and I knew in
1:24:40 that minute the world had just changed I
1:24:42 still haven't written a line of python
1:24:44 in my life
1:24:45 I couldn't
1:24:47 I couldn't tell you the syntax of a
1:24:49 Hello World script in Python but I wrote
1:24:53 a fully functional application in 90
1:24:55 minutes
1:24:59 that's not
1:25:02 that's not just a better Google search
1:25:04 engine that's that's that's
1:25:07 like
1:25:12 that's a capability amplification
1:25:18 it's just it's it's just it's like it's
1:25:21 like on another plane
1:25:24 I had an intention of wanting to do
1:25:27 something cool with python and if you've
1:25:29 ever tried to teach yourself something
1:25:31 you know that you have that intention
1:25:33 and you have to be really [ __ ] driven
1:25:38 to spend the next three months of your
1:25:40 life learning enough python to get it to
1:25:42 do anything or learning enough guitar
1:25:44 chords to make a song or learning enough
1:25:47 music theory to not suck at songwriting
1:25:51 or soloing or whatever it is right
1:25:53 there's always that three-month [ __ ]
1:25:57 awful shame ridden
1:26:02 I'm dumb I can't keep up everyone's
1:26:04 smarter than me
1:26:08 and then with these tools it's like Bang
1:26:11 what if you could get to
1:26:14 play with what it's like to be a python
1:26:17 programmer in two hours instead of two
1:26:20 months what might be different in your
1:26:22 life well [ __ ] I don't know I might
1:26:24 realize that I hate programming perfect
1:26:26 now you've only spent two hours learning
1:26:29 that you hate it
1:26:31 right and then you try I don't know
1:26:34 Advanced hydroponic gardening and you
1:26:37 learned that for two hours and now
1:26:39 you're like I love this okay cool
1:26:41 there's a passion go spend two months
1:26:43 with these tools
1:26:45 and you'll be farther along
1:26:49 than you would have been in a year with
1:26:52 books and websites and trial and error
1:26:57 hello from Trinidad and Tobago Rowan
1:27:00 super cool
1:27:01 hello
1:27:04 Kyle can I connect with you on LinkedIn
1:27:06 your company is in a similar area to
1:27:08 what I do of course Simon Says Tech yeah
1:27:10 drop me a line my my contact info is
1:27:13 there but just DM me on LinkedIn happy
1:27:15 to chat
1:27:17 habit chat that's why we're here that's
1:27:19 why we're here we're all just trying
1:27:20 here's Dale we're all just trying to get
1:27:22 along you know what I'm saying we're all
1:27:25 just trying to do the best we can you
1:27:27 know it's hard life's hard
1:27:30 it's 9 20. it's 9 20 here in Denver
1:27:33 oh no it's Uncle Kyle telling us his
1:27:35 first python story again I know what are
1:27:38 you gonna do it's it's it's actually one
1:27:42 of the challenges of these lives is
1:27:44 because I've been doing them
1:27:45 consistently and the Irregulars are here
1:27:47 every time I tell a story I've told
1:27:49 before I'm like oh [ __ ] like everyone
1:27:51 like everyone's here a lot is just like
1:27:55 okay I can go have a coffee now I can go
1:27:58 I can go get some water now Kyle's on
1:28:00 this story this is going to be 10
1:28:01 minutes
1:28:02 but at the same time like there's new
1:28:04 people here right there's 95 people in
1:28:06 here and you know I don't know how many
1:28:08 Irregulars there are but there's
1:28:09 probably what 30 or 40.
1:28:11 maybe there's more
1:28:13 but um
1:28:15 yeah but that's just that's that's I
1:28:17 don't know I'm just trying to figure
1:28:19 this thing out the best I can
1:28:21 so I totally get it I totally get it are
1:28:24 you a computer scientist I have a degree
1:28:26 in acting
1:28:29 you think it's a joke
1:28:30 no I'm not
1:28:32 I'm not so so one
1:28:33 I also talked about a lot in here is
1:28:35 that what this channel is about what my
1:28:37 passion is and has been
1:28:39 so
1:28:40 when I discovered this thing called the
1:28:43 World Wide Web in 1994
1:28:46 and started experimenting with it I'd
1:28:48 been playing with the internet for I
1:28:50 don't know two or three years and just
1:28:52 kind of again [ __ ] around with it
1:28:54 kind of realizing that I'm not a
1:28:56 scientist I'm not a programmer I'm not a
1:28:58 librarian I'm not a researcher I'm not
1:29:00 writing a dissertation and that's the
1:29:02 internet
1:29:03 sort of
1:29:05 pre-1992 was just all research right it
1:29:08 was very dry it was very technical it
1:29:09 was very scientific it was all about
1:29:11 protocols and
1:29:13 whatever
1:29:15 and then the World Wide Web comes along
1:29:17 and
1:29:18 I was doing desktop publishing as my day
1:29:20 job I'd written seven screenplays I'd
1:29:22 started to run a theater company I knew
1:29:24 a bunch of talented people
1:29:26 and the more I learned about this thing
1:29:27 the more I thought huh I wonder because
1:29:29 because in the early 90s the late 80s
1:29:32 and early 90s in New York City paper
1:29:34 zines were a thing like self-published
1:29:37 magazines that people would literally
1:29:39 just write them desktop published them
1:29:42 and then
1:29:44 um print them with photocopiers and and
1:29:46 distribute them that was like a hip
1:29:48 thing to do I thought I wonder if we
1:29:49 could do like a like an online version
1:29:52 of that
1:29:55 that's not something that people were
1:29:58 doing with the internet right so so so
1:30:01 for me
1:30:03 embracing technology when it when it's
1:30:05 right at that
1:30:06 inflection point of shifting from the
1:30:09 Geeks to the rest of us that's what my
1:30:12 passion is and that's what this is so
1:30:14 this channel
1:30:16 it's birth
1:30:18 was
1:30:21 about three weeks after chat gbt
1:30:23 launched
1:30:24 November 30th 2022 November 30th 2022.
1:30:30 I've got
1:30:32 [Music]
1:30:32 um
1:30:41 I wonder if this is not going to be
1:30:43 there let me see if this is here
1:30:49 hold ah this is
1:30:52 my hard drive's too full so everything's
1:30:54 on iCloud it's okay I gotta download
1:30:58 everything from iCloud because I've been
1:31:01 downloading
1:31:02 um Tick Tock live archives all day today
1:31:05 so every one of them is like one and a
1:31:07 half or two gigs and it just my my hard
1:31:10 drive is just very not happy with me
1:31:13 but anyway I've got a slide that I'll
1:31:15 show you here
1:31:16 that memorializes November 30th 2022
1:31:20 because
1:31:21 for me that's the day that's the flag in
1:31:24 the ground
1:31:25 where machine learning and artificial
1:31:28 intelligence shifts from the researchers
1:31:31 and scientists to the rest of us
1:31:34 but we're very very early right and you
1:31:37 still have to be geeky enough to use
1:31:39 these tools
1:31:45 um
1:31:51 that I did along with
1:31:53 this one that says the great Renaissance
1:32:01 which
1:32:02 this is what I think we're entering
1:32:04 right now is a great Renaissance
1:32:08 and I think it started on this day
1:32:11 and why is it a great Renaissance
1:32:12 because it's not just scientists anymore
1:32:15 it's scientists and artists
1:32:17 why was Leonardo da Vinci
1:32:20 you know heralded as such an amazing
1:32:23 human being because it wasn't just about
1:32:26 the science it was about the humanity
1:32:28 and he encapsulated that in one person
1:32:33 one remarkable person
1:32:36 these tools are going to allow each of
1:32:38 us to do
1:32:40 absolutely remarkable amounts of things
1:32:43 and what happens is when the industry
1:32:46 shifts from just the scientists and the
1:32:48 researchers to the rest of humanity
1:32:52 how the technology gets leveraged just
1:32:55 [ __ ] explodes
1:32:57 right so so we are right now in a
1:32:59 pre-cambrian
1:33:01 soup
1:33:03 of
1:33:04 like AI organisms and tools and there's
1:33:08 crappy tools and there's good tools and
1:33:09 there's powerful tools and there's
1:33:10 shitty tools but and everyone's just
1:33:12 throwing all the pasta at all the while
1:33:14 they're all trying to [ __ ] figure it
1:33:15 out
1:33:16 look at open interpreter look at this
1:33:18 look at code interpreter look at look at
1:33:21 baby AGI look at this look at that look
1:33:24 at this look at that's that's where we
1:33:25 are right now this is It's these are all
1:33:27 just little bloops in the Ooze in the
1:33:30 primordial ooze that's the stage we're
1:33:32 at right now because it's only nine
1:33:35 months old
1:33:40 [Music]
1:33:45 thanks Kyle and hello from Detroit
1:33:46 absolutely Simon Says Tech
1:33:49 look forward to talking with you
1:33:53 um
1:33:53 thank you for relating that experience
1:33:55 and the extrapolation cool I don't know
1:33:58 which one I've extrapolated a lot
1:34:00 tonight I'm on a I'm on a it's these
1:34:02 these things are the these these lives
1:34:04 have been weird like I've noticed I've
1:34:06 noticed they sort of I sort of cycle
1:34:09 through just being a silly [ __ ] idiot
1:34:12 to being really serious and
1:34:14 philosophical and then like last night I
1:34:17 was completely thrown for a loop because
1:34:19 the technology just [ __ ] the bed in a
1:34:22 [ __ ] hilarious way
1:34:26 and it's just I don't know I like I I
1:34:29 you know I I don't know how it's playing
1:34:31 I mean people keep coming back and
1:34:32 there's a number of folks in here
1:34:34 hanging out so on a Sunday night late so
1:34:36 I appreciate you hanging out with me but
1:34:38 it's weird how these how like I don't
1:34:41 know the personality of these is is very
1:34:43 chat add it's very different every night
1:34:46 um when you download your lives to
1:34:47 Archive does It capture the comments
1:34:51 I don't think it does Tobias which is a
1:34:54 which is a drag but it might I don't
1:34:57 know
1:34:58 um I don't know
1:35:00 that would be nice
1:35:02 um although I do try to read all the
1:35:04 comments so most comments get
1:35:07 um get covered in the lives like that's
1:35:08 that's what the lives are is me reading
1:35:10 comments basically well no I read a
1:35:14 comment and then I talked for 15 minutes
1:35:24 he just doesn't shut up does he
1:35:28 what so what why do you watch this you
1:35:31 you find this entertainment is this
1:35:34 educational
1:35:35 it's infotainment you think it's info
1:35:38 he's he's an infotainer
1:35:41 he's and and you spend how much time
1:35:43 this oh my
1:35:45 oh are you all right dear oh my
1:35:49 oh I'm so worried for you no it's good
1:35:51 okay and okay you like this AI stuff too
1:35:56 right okay I got a good I gotta go honey
1:35:59 I gotta go dear all right I'll see you
1:36:01 we'll we'll talk again all right
1:36:05 I knew you really were Tom Arnold
1:36:08 there's my trolls same old story stories
1:36:12 but new details this time make it
1:36:14 interesting oh that's good
1:36:16 um
1:36:17 brother 52 I knew you really were Tom
1:36:19 Arnold great job you've been live for 90
1:36:22 minutes
1:36:25 [Laughter]
1:36:29 oh man all right let's see uh gotta be
1:36:34 precautious with AI
1:36:38 uh what do you mean precautious
1:36:41 you mean precocious or or cautious I
1:36:45 don't know what you mean New York City
1:36:47 late 90s oh yeah cam catkin
1:36:51 are you in New York Kim
1:36:55 and New York in the 90s was [ __ ]
1:36:57 awesome it really was
1:37:00 it was it was insane man from 95 to it
1:37:04 got it got weird and shitty and like 99
1:37:07 because like the internet was way over
1:37:10 hyped and it was stupid there was stupid
1:37:13 money and
1:37:15 stupid investor types it was very weird
1:37:18 I didn't like it then
1:37:23 my Chinese Fred's friend said she used
1:37:25 the Internet it's late 80s early 90s for
1:37:31 p
1:37:34 ISS poop
1:37:38 I don't know I don't know what you're
1:37:39 saying there
1:37:41 um what do you think the Chinese are
1:37:43 doing with AI
1:37:44 um
1:37:45 Lots I mean here's the thing about AI it
1:37:48 depends what you mean with what are they
1:37:50 doing with AI so
1:37:51 to be clear that that's something that I
1:37:54 think is is really important that
1:37:57 um
1:37:59 ai's been around for
1:38:01 a couple of decades you know if the
1:38:04 precursors of it for you know four or
1:38:06 five decades
1:38:09 but but decently modern ai's been around
1:38:11 for a while right
1:38:14 um
1:38:16 it's it's how
1:38:18 you know a lot of a lot of machines work
1:38:22 these days and it's how a lot of
1:38:24 the surveillance industry does what it
1:38:28 does right object recognition and facial
1:38:31 recognition the Chinese have been using
1:38:33 that forever
1:38:34 not in New York City anymore but I was
1:38:35 there in the late 90s and Post 9 11. yep
1:38:38 cool me too I moved out of New York we
1:38:40 moved out of Brooklyn in 2002. and moved
1:38:44 up to like Westchester and Connecticut
1:38:45 and then we're there
1:38:47 until 2011 and moved out to Boulder
1:38:52 um
1:38:54 so
1:38:56 Alibaba
1:38:59 three days ago four days ago launched a
1:39:02 chat gpt-like tool
1:39:05 um
1:39:07 there's another
1:39:11 another Chinese company I can't remember
1:39:12 the name of it right now launched a
1:39:16 large language model kind of like llama
1:39:18 2. that they're claiming is as good as
1:39:20 chat gbt 3.5 they all of them all of
1:39:23 them are any any any research facility
1:39:26 that launches an open source large
1:39:29 language model seems to say it's as good
1:39:32 as GPT 3.5 because I think what they're
1:39:34 actually saying is it still isn't as
1:39:37 good as gpt4 so so I think that's why
1:39:40 they're all as good as gbt 3.5
1:39:45 um
1:39:47 you know so so I think they're doing
1:39:49 that I I think that
1:40:00 I I mean I honestly don't know I if I
1:40:05 were if I were an authoritarian or kind
1:40:07 of a centralized kind of government like
1:40:09 like China or you know someone that
1:40:12 Freedom of Information is not ideal
1:40:16 um
1:40:18 I'd be very nervous about the generative
1:40:21 AI stuff because I don't I don't quite
1:40:23 know how you lock this down like I know
1:40:25 hey I know how you lock down the
1:40:28 internet right you turn it off right and
1:40:30 they're good at that
1:40:32 so they can turn off certain domains and
1:40:35 certain things like that but what's been
1:40:37 happening in the open source Community
1:40:38 is that through
1:40:41 uh
1:40:42 quantization I think it's called
1:40:45 optimization they're taking these
1:40:48 relatively small large language models
1:40:50 they're optimizing them so that they can
1:40:52 be installed on a cell phone
1:40:56 and it doesn't need an internet
1:40:57 connection to do GPT 3.5 level stuff
1:41:01 so if you take a completely unrestricted
1:41:04 model like llama 2 the 7 billion
1:41:07 parameter model and you optimize it and
1:41:09 you make it so Winston I think it was
1:41:11 Winston earlier said I'm going to stick
1:41:13 large language models on Raspberry Pi's
1:41:16 and put them everywhere
1:41:18 that's possible Right
1:41:21 and that's possible now with some
1:41:23 geekery right and some efficiencies and
1:41:26 some
1:41:27 specialized whatever
1:41:30 but two things are happening
1:41:32 raspberry pies are getting increasingly
1:41:34 more powerful as our phones right
1:41:38 and
1:41:40 computer scientists are getting better
1:41:42 and better at making these large
1:41:44 language models smaller and smaller
1:41:45 they're figuring out what is what is the
1:41:48 most amount of data we can remove from
1:41:50 the large language model and still have
1:41:52 it perform well
1:41:53 and they're they're finding like some of
1:41:56 these experiments they're removing like
1:41:58 90 of the data out of the large language
1:42:01 model and it's still [ __ ] functions
1:42:07 so as that gets better and better
1:42:11 I don't know what you do is China
1:42:16 you know do you do you have every person
1:42:18 plug in their phone and you scan that
1:42:21 phone and make sure that there's not a
1:42:22 large language model on it I don't know
1:42:26 I don't know
1:42:27 so I guarantee you they're taking it
1:42:30 [ __ ] seriously
1:42:33 I guarantee you
1:42:36 China is
1:42:39 far more up to speed on
1:42:42 what its capabilities are than our
1:42:45 government is
1:42:46 I mean we just had Chuck Schumer meet
1:42:48 with the the heads of the you know
1:42:51 the big CEOs of these AI companies last
1:42:54 week
1:42:57 just to say hey we should do something
1:43:00 about this well you think
1:43:05 I'm going to put AI oh Ron Ron was doing
1:43:08 it Ron was Ron's the one that's putting
1:43:10 AI in a Raspberry Pi and leave them
1:43:11 everywhere hey so Ron um what do you
1:43:14 what are you gonna do like tell me what
1:43:16 you're going to do with them like what's
1:43:17 the is it is it like an art project is
1:43:20 it like uh like a social good project
1:43:23 what's going to be the interface for it
1:43:25 are you going to attach screens to them
1:43:27 is it going to be touch screens like
1:43:28 tell me a bit more about it because
1:43:29 that's a really interesting thing uh
1:43:32 Share Bear forty thousand so the
1:43:35 Irregulars the the likes for this live
1:43:37 are up to 42 000 right now
1:43:40 you guys are awesome
1:43:44 oh man
1:43:46 um let's zoom in on this one this is my
1:43:49 favorite this was from last night so if
1:43:50 you weren't here last night we had we
1:43:52 had a little soap opera with with Bing
1:43:56 and Siri had a little spat and then Bing
1:43:59 got offended that I said her images that
1:44:02 she created were shitty and she said she
1:44:04 got offended that I called her a robot
1:44:06 and she got offended that I told her her
1:44:09 images were shitty and she shut the
1:44:11 conversation down
1:44:12 oh my God it was funny
1:44:15 I'm not a robot goodbye
1:44:18 oh
1:44:22 yeah what oh Siri
1:44:25 um
1:44:26 uh precovedo means cautious in Spanish
1:44:28 okay got it wrong
1:44:30 um yeah so so I agree I think we need to
1:44:33 proceed with caution with these tools
1:44:36 and
1:44:38 um
1:44:39 the developer Community isn't you know
1:44:42 now that now that meta has open sourced
1:44:44 their large language models and
1:44:46 stability AI has open source theirs and
1:44:49 now China has put theirs out there and I
1:44:52 think there was another wasn't there
1:44:53 wasn't there one that came out of
1:44:55 Dubai
1:44:57 another large language model came out of
1:44:59 the Middle East so so there's there's
1:45:01 starting to be some foundational large
1:45:03 language models out there for the
1:45:04 developer Community to just play with so
1:45:09 um
1:45:11 we need to proceed with caution knowing
1:45:14 that not everyone's proceeding with
1:45:16 caution
1:45:17 so I I don't know what that means
1:45:21 I think that's why
1:45:23 the the people that know you know way
1:45:26 more than I do about this stuff
1:45:29 you know
1:45:30 I think one of the reasons they're
1:45:32 concerned is that
1:45:37 what we're playing with right now
1:45:41 are kind of
1:45:44 like
1:45:45 Child's Play
1:45:48 compared to what's coming like I think
1:45:50 they know it's coming
1:45:51 Sam Altman has said in an interview
1:45:54 there are things they've created at open
1:45:56 AI they will never release
1:46:01 so so so
1:46:04 I what what's been amazing is that the
1:46:07 scientists can predict with with pretty
1:46:10 good accuracy
1:46:12 how much better these models get when
1:46:14 you give them more gpus and more data so
1:46:17 they kind of know where they're headed
1:46:18 they know they know what's possible they
1:46:20 know what's coming and I would be very
1:46:22 surprised if a number of these companies
1:46:25 didn't have things that they were
1:46:26 playing with that were significantly
1:46:29 beyond what we're what we're seeing
1:46:31 today
1:46:32 so so anyway
1:46:34 Bing became my default search and
1:46:36 browser for months now yeah it's it's
1:46:38 mine too uh Nestor and you know what's
1:46:40 funny is I I'm an apple Fanboy like I
1:46:43 was a Microsoft Basher for years and if
1:46:45 you'd bet me a year ago that edge would
1:46:48 be my default browser and Bing would be
1:46:50 my default search engine
1:46:52 um I'd have laughed at you but
1:46:55 bing bing safety guard rails at this
1:46:58 point are pissing me off so much I'm
1:47:00 actually considering going back to
1:47:02 Google as my default search engine
1:47:04 it's because it's it's it's making me a
1:47:06 little cranky you're allowed to make
1:47:08 words up people yeah I love making words
1:47:11 up
1:47:12 I for storyvine our big word is Auto
1:47:15 Magic editing I stole Auto Magic from
1:47:18 the 90s I didn't make it up but I
1:47:20 certainly [ __ ] co-opted it we've got
1:47:22 an auto magic video platform
1:47:25 it's kind of like this
1:47:28 Auto Magic editing with Story vine
1:47:31 thank you for the cruise
1:47:36 all right how are we doing time wise all
1:47:39 right I got a little bit more time here
1:47:42 are there raspberry pies with tpus I
1:47:45 don't know let's go find out
1:47:47 Pate would probably know that Pate who's
1:47:49 been in here who works for Google on
1:47:51 tpus
1:47:53 are there rasp berry pies
1:48:01 there's probably a hat for it
1:48:05 with tpus
1:48:10 Google's Edge TPU machine learning chip
1:48:13 Google is now made available oh come on
1:48:16 Google is now made available for
1:48:18 developers a Raspberry Pi style
1:48:20 development board that comes with a quad
1:48:23 core
1:48:24 arm when was this
1:48:30 um this is 2019 so these things are way
1:48:33 way out if this was announced in 2019
1:48:37 so let me see
1:48:39 um
1:48:40 tools anytime
1:48:43 past month
1:48:49 [Music]
1:48:54 is Google Edge TPU a good choice for
1:48:57 Raspberry Pi
1:49:02 using tensorflow Lite with Google Coral
1:49:05 TPU on a Raspberry Pi 4. yeah it looks
1:49:08 like looks like it's there I don't know
1:49:10 if it's on the Raspberry Pi 4 or if you
1:49:13 add it with a with a someone in here
1:49:15 probably knows there's there are clearly
1:49:17 some Geeks in here so any of you Geeks
1:49:19 know
1:49:20 is it integrated into the Raspberry Pi 4
1:49:23 or is it something you snap on it it
1:49:25 looks like it's an add-on board but I
1:49:27 don't know
1:49:28 I don't know I haven't played around
1:49:29 with the Raspberry Pi the last thing I
1:49:31 did with the Raspberry Pi was like
1:49:33 pre-pandemic
1:49:36 um I have a Raspberry Pi hotspot for my
1:49:39 DMR radio very cool also Chinese are
1:49:42 listening and boring
1:49:44 everything we say here yes
1:49:47 they are that's okay
1:49:49 I hope I can make my meeting at 2PM
1:49:52 tomorrow where are you Nestor
1:49:57 I I'm not going to be here that much
1:49:59 longer but
1:50:00 um welcome everyone if you're new here
1:50:02 this is the AI learning lab it looks
1:50:03 like a bunch of people just joined
1:50:05 um here's the deal so my name is Kyle
1:50:07 Shannon I'm an entrepreneur I'm a CEO of
1:50:09 a company called storyvine
1:50:11 um We're talking mostly here about sort
1:50:13 of post chat gbt you know AI in the in
1:50:17 the era of what happens when the rest of
1:50:19 us get access to these tools
1:50:21 um I know a little bit about the
1:50:22 technology side but I'm not a
1:50:24 technologist
1:50:25 um if you have not played with chat gbt
1:50:27 before if you're here and you're like is
1:50:30 that Rosie O'Donnell
1:50:32 um if that's the reason you stopped
1:50:34 welcome I'm not I'm Tom Arnold and
1:50:39 um
1:50:40 go there
1:50:42 chat.openai.com that's the official
1:50:44 chatgpt website
1:50:46 and just start playing with it if you
1:50:48 haven't played with it before just do it
1:50:50 and and play with it until you have your
1:50:53 Kevin McAllister moment which is this
1:50:55 moment
1:50:58 from home alone and if you if you want
1:51:01 to start digging into this stuff go to
1:51:03 that document it's called prompts.chat
1:51:05 it's it it will explain to you what
1:51:07 large language models are and how to
1:51:08 talk to them and they'll give you a
1:51:10 bunch of prompts you can copy and paste
1:51:11 and give you a real sense of how deep
1:51:14 these things are deeply and by Deep I
1:51:17 mean profoundly powerful
1:51:20 like
1:51:21 as human beings we're incredibly lucky
1:51:24 to be able to play with this [ __ ]
1:51:28 because if Google had it their way we
1:51:31 wouldn't be Google invented the
1:51:33 Transformer that these things are built
1:51:35 on
1:51:36 and then they mothballed it because I
1:51:38 think they realized it threatened their
1:51:39 business model and then open AI went and
1:51:41 launched chat gbt and [ __ ] everything
1:51:43 up for everyone so now everyone's like
1:51:46 we got to do this
1:51:47 so that's why there's all this fraud
1:51:49 froth and bother about AI is because of
1:51:52 chat GPT so that's the thing that
1:51:53 started it all so if you haven't played
1:51:55 with it go play there all right cool
1:51:57 I'm in Southern California oh I meant
1:51:59 just that my sleep's messed up oh not
1:52:02 referring to the live that's okay my
1:52:05 sleep my sleep's messed up too just
1:52:06 because of my brain but but uh I'm glad
1:52:09 you're here uh Nestor it's it's cool
1:52:11 having you
1:52:13 um I hope I can make a 5 a.m CST meeting
1:52:15 tomorrow oh that's awful is that your
1:52:18 domain prompts.chat no it's not it's
1:52:20 just someone else did that mine is my
1:52:23 domain is the salon.ai that's my and
1:52:26 storyvine's my company
1:52:31 Rosie O'Donnell exactly
1:52:35 self-aware like AI
1:52:40 um
1:52:43 imagine what Google has that they don't
1:52:46 say yeah well apparently Gemini that
1:52:49 they're that they've got right now which
1:52:50 is the replacement for Palm too that's
1:52:52 the [ __ ] sitting behind Bard right now
1:52:54 apparently it's multimodal
1:52:57 I think we'll do audio in and out we'll
1:53:00 do images and they're claiming
1:53:04 now I've heard now two different things
1:53:05 I've heard Gemini is going to be five
1:53:08 times better than gpt4 and I've heard
1:53:11 it's going to be 20 times better than
1:53:12 gpt4
1:53:14 which
1:53:16 for anything that's pre-release to be
1:53:18 making those claims that says maybe
1:53:21 it'll be close to as good as gpt4 that's
1:53:24 that's my expectation if it's at all
1:53:26 better than gpt4 holy [ __ ] could be a
1:53:30 game changer
1:53:32 because don't forget forty five thousand
1:53:35 don't forget
1:53:36 um
1:53:38 Google just announced duet for all of
1:53:42 their office suite right so it's in
1:53:45 slides it's in docs it's in
1:53:49 sheets it's in Gmail it's in Google meet
1:53:55 so all of those if you enable the duet
1:53:58 function
1:53:59 is some version of chat GPT
1:54:02 um Mike kaput from the marketing AI
1:54:04 Institute if if you listen to their blog
1:54:07 from last week
1:54:09 he does kind of a deep dive on the
1:54:11 Google duet stuff he basically says
1:54:15 um
1:54:16 this the integration into slides is
1:54:18 really good
1:54:20 the integration into sheets is really
1:54:22 shitty and limited the integration into
1:54:25 docs is dramatically disappointing
1:54:29 and the integration I don't think he
1:54:32 played with meats and I forget what the
1:54:35 other one was oh Gmail
1:54:37 he said it was it was decently useful in
1:54:40 Gmail but it didn't quite capture his
1:54:42 tone
1:54:44 all of the things that he was talking
1:54:46 about
1:54:47 Google duet sucking at
1:54:50 is because their core large language
1:54:52 model isn't as good as gpt4
1:54:56 so if Gemini is as good as gpt4 or
1:54:59 better
1:55:00 it changes the game for all the tools
1:55:02 that are going to get rolled into duet
1:55:05 somebody took my pinning ability Away
1:55:07 really
1:55:09 but wait wait oh hang on let me go see
1:55:14 if
1:55:16 you're still on the list
1:55:18 you're still there
1:55:21 manage
1:55:25 comment settings pin comments it's on
1:55:31 you should still be able to pin there
1:55:33 Vicky you might just need to you just
1:55:35 might need to quit and restart Tick Tock
1:55:38 maybe Tick Tock crap to bed maybe you
1:55:40 have a memory leak or something
1:55:44 Vicky sometimes it won't work for me
1:55:46 either but it comes back yeah yeah just
1:55:48 it's ticked Tick Tock is just
1:55:51 Tick Tock is a lovely application uh
1:55:55 written by a lovely company I I love
1:55:58 Tick Tock so much
1:56:06 can someone invite me to The Irregular
1:56:08 room and Discord please yeah what's your
1:56:11 um
1:56:12 what's your Discord handle Is It joker
1:56:14 in there
1:56:15 I'm a librarian thanks for the rose
1:56:19 [Music]
1:56:19 um
1:56:22 oh yeah Joker to avoid hang on
1:56:26 manage
1:56:28 rolls
1:56:29 AI regular AI Adventurer save you're in
1:56:34 Joker
1:56:36 close
1:56:39 right did I do it right
1:56:41 yes yes
1:56:43 I did yo in you did it yay
1:56:48 okay
1:56:51 [Laughter]
1:56:58 memory leak my memory is always leaking
1:57:00 exactly holy crap Microsoft Edge like I
1:57:05 don't know about anyone else here but
1:57:07 for those of you that are tab hoarders
1:57:09 like I am
1:57:11 Edge just hates it when you have 45 tabs
1:57:15 open it doesn't like it and I have one
1:57:16 of those things that puts the tabs to
1:57:18 sleep so it should be fine but it just
1:57:20 [ __ ] breaks Edge like twice a day
1:57:22 makes me cranky
1:57:23 I've been doing laundry did I miss
1:57:25 anything as good as black fight 2023 I'm
1:57:28 a librarian there's we're not going to
1:57:31 be able to re recreate bot fight like it
1:57:33 was it was a a perfect storm of
1:57:36 ridiculousness so no nothing nothing
1:57:38 remotely interesting tonight tonight's
1:57:40 been mostly me just pontificating about
1:57:42 [ __ ] so I haven't even actually demoed
1:57:44 stuff if anyone wants me to show
1:57:45 something I'm happy to show something
1:57:47 but
1:57:48 um the the chat's been really
1:57:50 interesting tonight y'all are bringing
1:57:52 up interesting stuff so but but this
1:57:54 this was the this was one of the things
1:57:57 that that happened last night was Bing
1:58:00 Bing said I'm not a robot
1:58:03 I'm a I'm a large language model from
1:58:06 Microsoft and if you don't like it
1:58:07 goodbye
1:58:15 Bing
1:58:16 you're such a card
1:58:19 you only missed the Tom Arnold comment
1:58:21 by someone asking Kyle uh
1:58:25 oh and someone asking if I'm a computer
1:58:27 scientist yeah exactly what are your
1:58:29 qualifications sir if I'm going to
1:58:32 spending any time in this lab I want to
1:58:34 know who's running the lab is there a
1:58:37 chief scientist here I might inquire
1:58:40 what are your qualifications sir you
1:58:42 will tell us I hope where's your smock
1:58:45 this doesn't seem like a very sanitary
1:58:47 lab seems much like a home office
1:58:50 yeah exactly my mind my mind is having
1:58:53 too many tabs open but the browser is
1:58:56 hung up exactly that's what it's like
1:58:59 having add man that's totally it
1:59:04 oh
1:59:06 don't
1:59:12 I'm trying to find where I was in my
1:59:15 questions
1:59:17 [Music]
1:59:19 what are we up to oh 48 000 you got to
1:59:22 get to 50 people what are you doing
1:59:25 what what is this what is this
1:59:27 slackerrific
1:59:29 my God not at 50 000 yet 90 minutes in
1:59:35 the arthritis kicking up I know I know
1:59:38 just come on get at it foreign
1:59:46 for those of you who are not Irregulars
1:59:48 here I'm just picking on the Irregulars
1:59:50 because they have a side game going that
1:59:53 they can that they can beat up see see
1:59:55 now that I said that now we're at 49 500
1:59:58 now it's going up I can actually watch
2:00:00 it going up six six hundred we're gonna
2:00:03 700 y'all are nuts 800. we'll do this
2:00:08 real time 950
2:00:10 000. Bang
2:00:16 my God you're hilarious
2:00:19 I love you guys that's that's hilarious
2:00:23 oh
2:00:26 okay I heard Kyle was going to sell
2:00:29 glucosamine and Tick-Tock shop yeah
2:00:31 exactly I tried selling [ __ ] t-shirts
2:00:35 on Tick-Tock shop when they [ __ ] that
2:00:36 up maybe they could get glucosamine
2:00:38 right
2:00:43 I gotta I gotta figure out the the merch
2:00:46 thing because there's some some fun
2:00:48 merch that we can do that I want to make
2:00:50 available but
2:00:53 um
2:00:54 gotta get it going hell yeah 50k that's
2:00:56 right Share Bear you guys went nuts I'm
2:00:58 like you haven't gotten to 50K yet it
2:01:00 was like
2:01:05 um you just gotta exercise caution
2:01:06 that's it it's not scary yeah exactly it
2:01:09 you know what we talk about in here
2:01:11 Nestor a lot is with with these tools is
2:01:15 the reason you need to get curious with
2:01:18 them right now and just be exploratory
2:01:20 is that they're they're very rough
2:01:22 around the edges so right now if you
2:01:26 want to play with these remarkable tools
2:01:28 you kind of need to know where the
2:01:30 boundaries are what's good what's bad
2:01:32 you know where do they hallucinate where
2:01:33 do they not how can you you know
2:01:35 mitigate that in any way and there's the
2:01:38 only way to do that is just like do it
2:01:39 and do it and do it and play with it and
2:01:41 then go oh it's perfect and then you
2:01:43 realize oh it's not perfect it's far
2:01:45 from perfect oh crap that's farther than
2:01:47 I thought it was right and all that
2:01:49 stuff
2:01:50 Thelonious I dined with a friend tonight
2:01:53 former Hasbro and Mattel Creator he's
2:01:55 making an AI dog toy there's a um
2:01:58 there's a dude on Tick Tock I think one
2:02:02 of you a regular sent it to me
2:02:05 um yeah I think it wasn't youth
2:02:06 theologies it might have been retro Punk
2:02:09 um and he basically hooked up like a
2:02:13 Teddy Ruxpin teddy bear to a chat gbt
2:02:16 conversation and he he's he's like yeah
2:02:20 you know you gotta you gotta figure out
2:02:23 what speaking of figuring out where the
2:02:25 boundaries are you don't have a toy that
2:02:27 tells you to go [ __ ] yourself like if
2:02:29 you put in an unrestricted large
2:02:30 language model those things can get
2:02:32 really weird really fast
2:02:34 so uh yeah that's gonna be interesting I
2:02:37 mean
2:02:41 yeah it's fascinating I'm
2:02:42 I'm working with a guy from the AI salon
2:02:44 right now and he's put together
2:02:47 he's he's putting together something
2:02:49 um for health information where it can't
2:02:52 hallucinate and he's using an
2:02:54 interesting combination of large
2:02:56 language models and databases
2:02:59 um where they're kind of working in
2:03:01 conjunction with one another where if
2:03:03 someone asks something that's out of
2:03:05 bounds it won't answer it if it's not in
2:03:07 an approved answer database and then if
2:03:11 it gives an answer it has to give
2:03:12 something that's been approved really
2:03:14 interesting approach that he's doing so
2:03:17 what you got for good PowerPoint
2:03:19 generation right now
2:03:22 um Tome tome.ai I think is the tool hang
2:03:26 on let's go let's go play I think it's
2:03:29 tome.ai no tone dot app
2:03:33 t-o-m-e dot app
2:03:39 sign in with Google and it's got a bit
2:03:41 of a
2:03:41 [Music]
2:03:44 kind of hipster interface so it's it's
2:03:46 hard to read and demo especially if it
2:03:49 looks like I wiped a buttery biscuit on
2:03:52 my screen I should probably clean that
2:03:54 that thing
2:03:56 oh they've what have they done here
2:04:01 aerosol candy
2:04:03 create
2:04:06 so you can create an image you can
2:04:08 create a document
2:04:10 to presentation create a presentation
2:04:13 about
2:04:16 let's see
2:04:18 um
2:04:25 spray candy
2:04:27 that tastes like Skittles
2:04:33 and comes
2:04:35 in a graffiti can
2:04:43 so now it says it's generating the
2:04:44 outline so then there's your outline
2:04:47 introducing spray candy Taste the
2:04:50 rainbow with spray candy so it actually
2:04:51 pulled the Skittles tagline that's
2:04:54 interesting the graffiti can design
2:04:57 perfect for sharing available now join
2:05:00 the spray can be craze
2:05:02 and then I can edit any of these and I
2:05:05 think I can add things into them and now
2:05:07 I say continue
2:05:09 and now this thing is going to
2:05:14 write my deck and then it'll actually
2:05:16 generate images for each slide as well
2:05:19 um
2:05:20 the thing about these is
2:05:23 graffiti Skittles spray your taste buds
2:05:27 it's kind of cool
2:05:30 so there's the table of contents there's
2:05:32 kind of Skittles looking [ __ ] and
2:05:35 there's kind of a graffiti looking spray
2:05:36 can
2:05:38 so I assume they're using Dali for this
2:05:40 which is why the the images are a little
2:05:43 less than like like that's just [ __ ]
2:05:46 creepy and weird
2:05:49 like like this is what would be in an
2:05:51 absurd movie about the future where this
2:05:54 is where marketing is actually gone
2:05:57 Taste the rainbow with spray candy and
2:05:59 it you know it writes a little deck and
2:06:02 it you know it writes all your stuff and
2:06:04 there's your can
2:06:06 look how happy people are they're so
2:06:08 happy that their hair is turned into
2:06:09 sticky goo
2:06:10 [Laughter]
2:06:15 um
2:06:16 so so so this one's good I there there I
2:06:20 would think at this point there are
2:06:22 probably a dozen of these things out
2:06:24 there Tome is the one I know
2:06:28 um I would go to futurepedia.io or
2:06:33 um
2:06:34 someone in the in the salon
2:06:39 [Music]
2:06:40 um
2:06:41 shared hang on
2:06:57 there's an AI for that
2:07:00 is a website
2:07:05 there's
2:07:07 an AI for that.com
2:07:11 and then I'll just type in
2:07:18 um presentation
2:07:19 slides presentations
2:07:24 Tome pitch Avatar so anyway go to
2:07:28 futurepedia.io or there's an there's an
2:07:31 AI for that.com
2:07:33 and you can probably find some more I
2:07:36 personally I'm a librarian I'm more a
2:07:40 fan of using chat gbt to do this
2:07:44 because you can just copy and paste the
2:07:47 stuff from from here into PowerPoint
2:07:49 easy enough you can just pick a pick a
2:07:51 template that you like
2:07:52 but but you know have have gpt4
2:07:57 um do like uh let's see
2:08:00 um
2:08:01 [Applause]
2:08:06 you're a librarian tasked with
2:08:10 giving a presentation on
2:08:14 generative AI as it is today and the
2:08:18 implications for the future period
2:08:22 the presentation is 20 minutes long
2:08:27 and
2:08:29 I want you to create a 15 slide
2:08:33 outline of what the presentation might
2:08:36 be
2:08:37 period the audience
2:08:40 is anyone in the community from age 15
2:08:45 to 75 period
2:08:52 [Music]
2:08:56 so there it's given me a bit of an
2:08:57 outline and and I'm using gpt4 here so
2:09:00 that's why it's slow if you use 3.5 it
2:09:02 would be much faster than this
2:09:06 but you know I would do something like
2:09:07 this like start out with something like
2:09:09 that have it write you an outline and
2:09:11 then just give it feedback on the
2:09:13 outline no I want the outline to focus
2:09:15 more on the societal aspects or this or
2:09:18 that
2:09:19 and then once you've got the outline
2:09:21 dialed in then then
2:09:23 either give it an example of a slide
2:09:26 that you like do you like slides with
2:09:28 words on them or not words on them do
2:09:30 you like them to be catchy do you want
2:09:32 it to recommend images
2:09:36 let's let's see what this actually did
2:09:38 let's see see if this is any good
2:09:42 um title slide introduction to AI what's
2:09:45 generative AI applications today text
2:09:47 image video and audio generation real
2:09:49 world examples
2:09:52 um benefits
2:09:55 of generative AI deep fakes in the
2:09:58 reality Distortion ethical implications
2:10:00 Ai and literature and movies that's good
2:10:03 Community perspective future predictions
2:10:05 education and AI how to stay informed
2:10:09 um
2:10:16 please give me
2:10:19 a complete slide
2:10:22 for the education slide
2:10:27 complete with speakers notes and a
2:10:30 description of the image that should be
2:10:32 on that slide
2:10:35 so let's see how it does
2:10:38 blown one of these things out
2:10:40 slide design and image imagery a
2:10:42 classroom setting with half of it
2:10:44 looking traditional wooden desk
2:10:46 chalkboard and the other half modernized
2:10:49 digital screens virtual reality headsets
2:10:51 robot teaching assistant like that's
2:10:53 something that's something you could
2:10:55 absolutely put in mid-journey and get
2:10:57 right
2:10:59 so that's the design imagery title the
2:11:01 role of AI in modern education the
2:11:03 bulleted points on the slide AI Literacy
2:11:05 for all ages personalized learning
2:11:08 experiences virtual classroom and tutors
2:11:10 opportunities for Community Learning
2:11:13 speaker's notes and there's you know big
2:11:15 long
2:11:17 speakers notes for each of the bullet
2:11:19 points so
2:11:22 the thing about
2:11:24 there's kind of two advantages for doing
2:11:27 it in chat CPT versus a tool
2:11:31 where someone who may or may not have
2:11:33 your sensibilities about presentations
2:11:35 has pre-prompted one of these tools
2:11:38 right because all Tome is is someone is
2:11:43 pre-prompted it to do what it does and
2:11:47 they may or may not agree with your
2:11:48 philosophy of how you like to think
2:11:49 about a presentation
2:11:51 so one thing is you have a hell of a lot
2:11:54 more control doing it this way the other
2:11:56 thing is
2:11:59 you know this is this is one of the only
2:12:02 ways that I know where you actually
2:12:03 figure out where the boundaries of these
2:12:05 tools are like GPT might be brilliant
2:12:08 for doing what you want it may be not
2:12:11 brilliant for doing what you want
2:12:15 yes I've done that I wanted the slides
2:12:17 generated too I spend a little bit two
2:12:19 time yeah so okay so if you're already
2:12:21 good with the with the GPT
2:12:23 um
2:12:24 I would you know you know what the thing
2:12:27 to do might be is just um
2:12:31 have it write you a series of
2:12:33 mid-journey prompts
2:12:35 um for each slide in one of your slide
2:12:36 decks but you can also just go try Tome
2:12:40 Tome will put it together for you
2:12:42 um you can also go turn on duet in in
2:12:46 the Google so Google Slides now has duet
2:12:49 within it and it will generate images
2:12:51 within within Google Slides but you have
2:12:54 to turn on the duet feature as a
2:12:57 subscription at the administrator level
2:13:02 [Music]
2:13:03 um
2:13:04 Champy's about to bark at someone I'm a
2:13:07 tab hoarder in Chrome I put them in
2:13:09 blocks too I ordered blocks I'm the same
2:13:11 way Vicky not the same way my my blocks
2:13:14 like it doesn't I wish it would there
2:13:15 were a way to keep those blocks
2:13:17 persistent because I I keep like when I
2:13:20 change a website I lose the tab and I
2:13:22 mine are always just a [ __ ] mess
2:13:26 have you used Edge for work oh I have
2:13:29 used Edge for work but when it comes to
2:13:30 secure Bing chat with gpt4 so I don't
2:13:33 complain yeah that's good
2:13:35 [Music]
2:13:37 um
2:13:43 uh anything interesting from my emails I
2:13:46 haven't last time I looked at them was
2:13:47 like last Thursday so I just haven't
2:13:49 responded to them Eric Fortier
2:13:52 um I also couldn't tell do you do you
2:13:55 sometimes send from two different emails
2:13:56 I couldn't tell sometimes I couldn't
2:13:58 tell if it was yours or not so
2:14:00 um I might have gotten an email that I
2:14:02 didn't know was yours
2:14:03 um Claude AI is only the UK and the US I
2:14:06 know it sucks
2:14:08 um my mind is like having too many tabs
2:14:10 open
2:14:11 I think we went through this all right
2:14:13 I'm gonna jump to the bottom here
2:14:14 because I'm getting a little crispy
2:14:17 I want to wind things down here wait
2:14:19 what are we up to now 57 000 likes y'all
2:14:21 are as well
2:14:24 all right
2:14:26 he licenses toys let's see
2:14:31 foreign
2:14:34 he will see his first work up soon
2:14:36 that's pretty cool I had a I had a when
2:14:39 I I sold agency.com in 2002 and the
2:14:42 first company I started after that was a
2:14:44 company called invention Asylum and I
2:14:47 was inventing like toys and kitchen
2:14:48 gadgets and [ __ ] like that so I spent
2:14:51 like two years trying to get into the
2:14:53 toy business it's [ __ ] brutal I'll
2:14:55 tell you what if you want to find the
2:14:57 most miserable people on the planet
2:14:59 go into the toy business oh my those are
2:15:03 some bitter angry cynical people
2:15:10 ironic and depressing you'd think you'd
2:15:13 go in there and it's going to be some
2:15:14 fun people ain't nothing
2:15:19 124 people in the room it only takes a
2:15:21 thousand Taps each
2:15:24 oh man I sent the teddy bear to you that
2:15:27 was you Tobias that's right yeah that
2:15:28 was awesome yeah it was really good it
2:15:30 was very cool it was a cool video I saw
2:15:32 that decided it was fake oh the oh the
2:15:35 teddy bear thing probably not fake
2:15:37 probably not fake because I because you
2:15:39 could you could do something like that
2:15:41 with with uh pie right now
2:15:44 you could put a you could put a shitty
2:15:46 cell phone inside a teddy bear with pie
2:15:49 and have it do that although I guess it
2:15:52 had some animation stuff going on but
2:15:54 even with a Raspberry Pi you could do
2:15:56 some motion control stuff with a teddy
2:15:58 bear
2:15:59 probably pretty easily
2:16:03 oh champ what do you want mirror
2:16:10 invited the maker here
2:16:13 I shouldn't talk at him
2:16:16 oh wait I shouldn't talk at him if he's
2:16:21 here Thelonious have him have him speak
2:16:23 up or maybe
2:16:24 actually felonies you know it might be
2:16:26 interesting he might be somewhat
2:16:28 interesting to speak at the salon if
2:16:30 you're a good friend with him
2:16:32 um
2:16:33 it would be really interesting to hear
2:16:35 from someone that's that's looking to
2:16:37 use generative AI you know in a consumer
2:16:40 toy
2:16:42 um because the again the boundaries of
2:16:45 where these things are like that that
2:16:46 would be really fascinating
2:16:49 um so if you know him well
2:16:50 um you know ping me on the ping me on
2:16:52 Discord maybe we can get him to speak at
2:16:54 the salon
2:16:55 by the way I'm I'm looking for people to
2:16:58 speak at the salon
2:17:00 um just in general so if you know anyone
2:17:02 who's doing interesting stuff in AI or
2:17:04 doing an interesting project or anything
2:17:06 thing like that let me know
2:17:08 um because we might bring them in to
2:17:09 speak at the salon
2:17:12 um
2:17:13 you have candy my take on the dog is
2:17:16 it'll be like Pi dot AI responsive and
2:17:18 conversational yep Yano greater than
2:17:21 Teddy Ruxpin what site is this what site
2:17:24 was I on oh it was on Tome
2:17:27 t-o-m-e-t-o-m-e dot app that was the one
2:17:29 making the uh slides
2:17:33 My Guy created micro toys I remember
2:17:35 micro toys they were [ __ ] great and
2:17:38 Tiny trucks oh my God thanks for the
2:17:40 burger
2:17:42 that was a Happy Meal it made me happy
2:17:44 join this the spray candy craze is not
2:17:48 as tacky that's good it's tacky I like
2:17:50 it
2:17:52 Max bad hedge
2:17:55 um have you seen AI generated soda
2:17:57 commercial yeah the one from Coca-Cola
2:18:00 well what was interesting about the
2:18:02 Coca-Cola one is that there was some
2:18:05 generative AI stuff in there and then
2:18:07 some full-on you know beautiful you know
2:18:10 handcrafted Motion Graphics and
2:18:13 videography like it was a a really nice
2:18:15 Coke our Master storytellers and their
2:18:18 masters at technology so so and Coke
2:18:22 also just released the first AI
2:18:27 AI uh in AI generated
2:18:31 um Coke flavor they basically had a I
2:18:35 say come up with a flavor profile that
2:18:37 our customers would like and they've
2:18:39 launched it it's called like
2:18:41 something 3000 why 3000 or something
2:18:44 like that I have the domain name for
2:18:47 toys for growth should I dump it
2:18:49 um
2:18:50 I don't I'm I tend to be a domain
2:18:52 hoarder as well if you've got a good
2:18:53 domain if you if you might do something
2:18:55 with it hang on to it
2:18:57 it's you know if you know you gotta
2:18:59 spend the the monthly or the annual
2:19:01 thing
2:19:02 anyone keeping a list of his suggestions
2:19:05 down in the mud so Thelonious
2:19:08 um
2:19:09 yeah Thelonious is doing that down in
2:19:12 the mud he's got a
2:19:14 um
2:19:16 a Google sheet with like every time I
2:19:19 mention something he adds it to the
2:19:21 sheet
2:19:22 um I'm going to add that to my linked
2:19:24 trees I just haven't gotten around to it
2:19:25 yet
2:19:29 um yes but your explanation helped me
2:19:31 with GPT more cool oh good terrific
2:19:36 um Indian names are usually plus George
2:19:38 Douglas wait
2:19:41 are usually Plus stores I don't quite
2:19:43 know what that means but funny I think
2:19:46 where do you get your cool shirts from
2:19:48 my cool wife
2:19:49 who gets them from the cool internet
2:19:52 somewhere bitter and cynical in the toy
2:19:54 business not unlike the field of
2:19:55 Education exactly some found being
2:19:58 around all them kids all that time
2:20:00 you're like
2:20:02 um anyone else buffering
2:20:05 um I'll ask if he'll speak at the salon
2:20:07 that would be great I know him well
2:20:08 that's great Thelonious thanks thanks
2:20:09 for that that would be super cool
2:20:13 and it's and it's a it's a relatively
2:20:15 like the speaking slot at the salon is
2:20:17 like you know talk for like 10 or 15
2:20:19 minutes and then open up the
2:20:20 conversation so
2:20:22 he doesn't necessarily need to be a
2:20:24 master like he can just be like here's
2:20:25 what I'm trying to accomplish here's
2:20:27 what I've learned here's what works
2:20:28 here's what doesn't like it doesn't it
2:20:30 doesn't have to be like everyone right
2:20:32 now is trying to figure this [ __ ] out so
2:20:34 if if he's concerned that like well I
2:20:36 haven't quite figured it out yet that's
2:20:37 okay like
2:20:39 even the process of how he's planning to
2:20:43 navigate getting that through legal
2:20:45 approval I find I would love to hear
2:20:48 that story
2:20:49 like are the lawyers cool with this
2:20:53 um
2:20:55 the animation is what made me think it
2:20:58 was fake uh okay for the for that other
2:21:00 bear it might be there's there's
2:21:02 probably a lot of that fake [ __ ] out
2:21:04 there I sent you the Andreessen Horowitz
2:21:06 and what AI tools the August character
2:21:08 okay cool great I'll look for that
2:21:11 uh
2:21:13 um oh the bear was straight up made up
2:21:15 okay agreed the process is super helpful
2:21:18 yep cool
2:21:20 Thelonious is copyrighting them
2:21:24 all right listen everybody
2:21:26 I'm a little crispy I'm like I can feel
2:21:30 my energy like drip dropping down
2:21:33 um really cool hanging out with you the
2:21:35 the final total by the way we're at 59.9
2:21:38 so I will not leave here until we hit 60
2:21:41 000 likes
2:21:43 599 is not a good place to end it okay
2:21:45 we just did 60. okay all right everybody
2:21:49 um let's do a little Romper Room here uh
2:21:51 I see Amelia's wife and felonious I see
2:21:54 down in the mud I see Stevo I see toe
2:21:59 dance welcome toenance share Bears here
2:22:03 um I'm a librarian we had a good talk
2:22:04 tonight
2:22:06 um Vicky
2:22:08 um goat Eric former
2:22:11 and who else Robert Rossi good to see
2:22:14 you
2:22:15 all right everybody USA loves tick tock
2:22:19 tick tock's the best thank you for the
2:22:22 donut my cousin was on Romper Room nice
2:22:25 good night everybody have a have a good
2:22:28 Sunday night or Monday morning whatever
2:22:30 it might be for you and I'll see you
2:22:32 tomorrow peace out