aiLL 8/16/23 - The Future Is Here - AI Is Already Changing Lives

Description

Chapters

Transcript

0:06 mm-hmm
0:16 hey Silver Fox what's happening
0:19 back in the home office
0:21 back in Denver headquarters
0:25 back home I mean his life was happening
0:28 I just figured out how to how to
0:30 actually invite you guys
0:33 so all you regulars I should do that so
0:37 that you're not just relying on tick
0:39 tock
0:42 take Doug's a bunch of dum-dums
0:47 encourage more viewers to share yeah
0:49 what she said
0:52 share share the live
0:55 we're just getting started so
0:57 people are just coming in but if you're
0:59 willing to share the live pop that out
1:01 there
1:05 you'll be happy to know I reset my Apple
1:09 watch
1:10 and it's not full anymore
1:13 I put in a beta I have a feeling that's
1:15 what that that's what happened
1:17 I put in a beta and I think it it
1:20 exploded it exploded
1:22 ah I got up at uh
1:26 four this morning in New York
1:30 and wait hang on
1:34 the partner just sent me something that
1:36 I don't think is right
1:44 oh yeah it is
1:47 yeah
1:49 yeah yeah yeah that's correct
1:54 that is correct ma'am you are correct
2:00 all right
2:01 let's see let's get back here I did some
2:04 housekeeping on the old chart the old
2:06 charity chart cleaned some [ __ ] up
2:10 we're gonna have that we're gonna have
2:11 that we're gonna have that
2:15 I guess we'll still have that oh yeah we
2:17 got that oh as as people get in here uh
2:20 for wait for a second I thought you were
2:23 an Indian oh hang on there's it's
2:25 probably the lighting is
2:27 changed and this filter is making me
2:30 look weird
2:32 it's pretty apparently making me look
2:34 weird enhanced filters
2:37 filters
2:39 [Applause]
2:41 many people say
2:44 there we go is that better
2:46 a little a little more better
2:49 you better you better you bet
2:52 all right
2:55 welcome everybody
2:57 um what you call it so we've been
3:00 scammed
3:02 the AI learning lab is just a humble
3:04 little Channel
3:06 are we all we want to do is learn about
3:08 Ai and the scammers are like no
3:11 we're going to take your nice people and
3:13 we're gonna get money from them we're
3:16 going to have them invest in our
3:18 multi-level marketing scheme so if you
3:22 see an AI learning lab asking you to
3:25 join some [ __ ]
3:27 first of all if I ever do that I'll tell
3:29 you here that I'm gonna do it and then
3:31 we'll it'll be like an agreement
3:34 so first of all I've never done that and
3:37 I won't do that and if I you know if I
3:39 do I'll tell you and
3:42 so if someone's asking you check the
3:44 name if it's got two L's for the word
3:46 learning
3:49 that's not how you spell it
3:51 so not only are they scammers they're
3:53 not very smart no but they are because
3:56 the A's next to the L and then the L's
3:58 next to the L and they're all lower case
3:59 so it all just looks like the same thing
4:02 so two L's at the beginning scammer
4:05 hopefully that's the only one uh have I
4:08 reported it to tick tock yes have Tick
4:10 Tock done anything no
4:13 does that make me cranky yes
4:17 do I think Tick Tock are dum-dums yes
4:20 are they not really doing it by intent
4:23 yeah yeah no they aren't and and the
4:25 other thing I think they accidentally
4:26 slipped up on was they they accidentally
4:28 copied over videos of me talking about
4:31 AI that was that was probably just a
4:33 slight oversight as well
4:36 although Tick Tock sees nothing wrong
4:38 with it
4:41 now how do they don't how they don't I
4:43 I'm reporting and I'm like these guys
4:45 are copying me there's pictures of me
4:47 and then on both accounts and I'm
4:51 reporting theirs anyway that's that
4:55 um let's look at some other stuff
4:58 so welcome everyone my name is Kyle
5:00 Shannon
5:02 um this is the AI learning lab my uh my
5:06 remit in life is to hang out and talk
5:08 about AI stuff lately this is a
5:12 self-imposed remit uh I think this
5:15 stuff's really important my background
5:16 I'm an entrepreneur I've started 12 or
5:18 15 companies in my day
5:21 um if you're new to generative AI if
5:24 you've not played with chat GPT go there
5:26 right now that top URL chat.openai.com
5:29 that's chatgbt
5:31 and you may hear people on Tick Tock
5:33 going you can make money
5:35 you won't hear that from me except in
5:37 that stupid voice because that's what
5:39 they sound like and they're idiots
5:41 so we'll talk about that but this shit's
5:44 really important so go play go there and
5:47 play and if you don't know what to say
5:49 there you can go to prompts.chat and
5:51 it'll tell you what to say there okay so
5:54 I I switched things up here if you've
5:57 been here I this was this list had seven
5:59 on it before I've knocked it down to
6:01 these three with the prompts.chat thing
6:03 this is like if you're just getting
6:04 started the reason I have the other two
6:08 chat GPT here
6:11 if you pay the 20 bucks a month you get
6:13 gpt4 you get code interpreter and you
6:16 get plug-ins
6:18 I pay the 20 bucks I think it's worth it
6:20 but I'm not getting paid by them so I
6:21 don't give a [ __ ] if you do it or not
6:23 these two Bing is connected to the
6:26 internet it can also make images for
6:28 free thanks dragonfly I appreciate that
6:29 and then Claude the one over there
6:31 that's a competitor to these two from
6:34 ex-employees of this first one a bunch
6:38 of ex-openai people started anthropic
6:40 and their Claude model has a hundred
6:42 thousand token context window so you
6:44 could paste in a
6:45 let's say you wrote The Great American
6:47 novel and you just wanted a little
6:48 editing pass done on it you could paste
6:51 the whole thing into the Claude prompt
6:52 and start interacting with it it's
6:55 amazing so so the first one is if you
6:58 haven't done anything just go there
7:00 if you want to sort of play with other
7:03 things like connected to the internet or
7:04 images or long long
7:07 chat contexts as they call them go there
7:11 now
7:12 there's also some other opportunities
7:14 feel free to uh what do they call it
7:17 screenshot this
7:19 um so Bard is Google's entry into this
7:21 mess
7:22 um they are not great but not bad but it
7:26 is connected to the internet is pretty
7:28 good at that it can also read images so
7:30 you can upload images and have it talk
7:33 to you about it then you got
7:36 labs.perplexity.ai that's uh meta's
7:39 llama 2 models you can play with pi is
7:41 from a company called inflection Poe is
7:44 from quora all of those these are just
7:47 interesting other things to experiment
7:48 with and then if you want to do pitches
7:50 pretty pitches I would love to do some
7:53 pretty pitches Kyle where do I go what
7:54 do I do
7:56 there's four of them the bottom two are
7:59 stable diffusion but they're kind of
8:01 friendly versions of stable diffusion
8:03 mid-journey in my opinion's the best but
8:05 you got to make it in Discord which is a
8:07 pain in the ass but it's the best that's
8:09 the one I use I pay for it that's 10
8:11 bucks a month Adobe flyer fly fly or fly
8:14 Adobe flyer fly
8:17 uh charges you you have to have a
8:21 creative uh whatever it's called
8:23 Creative Cloud license uh but that's
8:25 adobe's version that's commercially
8:28 um it's not trained on copywritten stuff
8:30 so if you're in a corporation that one's
8:31 probably going to be the one you'll end
8:33 up playing with Leonardo's a web
8:35 interface clip drop is from stability AI
8:37 themselves there's some pretty cool
8:39 tools at clip drop and then if you
8:41 really want to geek out
8:43 um stable diffusion like automatic 11 11
8:45 just go go to Reddit and go to the
8:48 stable diffusion subreddit and see all
8:50 the geeky projects there there's a lot
8:52 of development going on with stable
8:53 diffusion uh which makes it super cool
8:55 but you gotta you gotta be a little
8:57 geeky to do it so these these allow you
8:59 to do it without being super geeky and
9:00 then openart.ai promptbook look it's
9:04 designed for stable diffusion but this
9:06 is a great document and it'll teach you
9:08 about how how these things work in
9:09 general
9:10 um and and how deep they are thank you
9:12 Adam appreciate the Rose The Rose is
9:14 sweet oh the other thing we do here
9:17 which I think is kind of cool
9:19 is I answer questions
9:25 so you give me your questions you put
9:27 them in the comments there and then I'll
9:29 answer them and then what's going to
9:30 happen is this see I make predictions
9:32 too I'm going to predict that before the
9:35 night is up
9:37 someone's gonna go why aren't you
9:39 answering my question Kyle this is very
9:41 frustrating a little feedback on the
9:44 channel this is what we got last night I
9:46 like the content here it's fantastic but
9:48 let me give you a little feedback on the
9:49 channel I asked my question hours ago
9:52 and you haven't answered it yet so
9:53 that's a little you know I would go
9:55 ahead and not do that if I were you I'd
9:56 go ahead and answer my question
9:58 so that person probably came in like an
10:01 hour after I started and then I was
10:03 already behind because I answer all the
10:05 questions I go through and answer them
10:06 all so I there might be a better way to
10:09 do this and at some point I'll do it but
10:10 at this point I'm just I'm I'm right now
10:12 an old man stuck in my ways and I'm I'm
10:15 just a curmudgeon and it's just the way
10:17 I'm doing it and uh at the point at
10:21 which I take twerking lessons things are
10:23 going to change in this channel it's
10:24 going to be a lot more spinning lights
10:27 probably a lot more strappy
10:30 garbs
10:34 I would think some spandex would be in
10:36 order but right now you get t-shirts and
10:39 a shitty haircut
10:41 um
10:42 another big topic conversation what the
10:44 [ __ ] you drinking Kyle that's just water
10:46 but what do I have in it it's not booze
10:49 no
10:50 no
10:52 apple cider vinegar
10:54 all right here's something this is not
10:56 an AI tip but this is
10:58 [ __ ] insane
11:00 if you've got acid reflux
11:03 the reason you have acid reflux is not
11:06 because you have too much acid it's
11:08 because you don't have enough try if you
11:11 have it well partially it's what you eat
11:13 like if I eat
11:15 really like like uh
11:18 something like fatty with dairy like ice
11:21 cream will do it just I don't know
11:23 certain combinations of food it's always
11:24 shitty food certain combinations of
11:26 shitty food I'll get it
11:28 and uh
11:30 if you drink apple cider vinegar it adds
11:33 acid to your stomach the little flap
11:35 that makes that happen it opens if your
11:37 stomach is too low in acid it's
11:39 counterintuitive because as it comes up
11:41 you're like I got too much acid no
11:43 to little acid not too much
11:46 I didn't know that
11:48 I didn't learn that from chat gbt but I
11:50 did learn that from the internet and if
11:52 it's on the internet
11:54 it's true that's the law
11:59 and you know does it taste like [ __ ] oh
12:01 yeah that's why I washed it down with a
12:03 little coffee mmm
12:06 all right so pop your questions in the
12:09 comments below I will do my best to
12:10 answer them and uh keep up but I will
12:13 fall behind on the keeping up
12:15 there are some regulars in here welcome
12:17 welcome welcome good to see you all
12:21 right what have you all said while I
12:23 have been rambling on Sonny and Cher I
12:27 don't know what I was singing but it was
12:28 probably sunny and Cher for a second
12:30 there I thought you were an Indian not
12:32 really eyes look too old to see that
12:35 well oh our eyes too old to see that
12:37 well the bear hello
12:40 she bath
12:42 cyber security 101 damned impersonator
12:45 are they not really doing it by Insane I
12:49 reported it to but the only choice
12:51 choices are they're impersonating me or
12:53 a celebrity wait a minute I'm a
12:54 celebrity
12:55 tell them I'm a celebrity
12:58 in my own mind they don't have that
13:01 option but
13:03 you don't need to tell them that
13:05 I was a I was in New Jack City I played
13:09 white boy number one
13:11 not white boy number two I got the I got
13:14 the lead when it came to all the white
13:15 boys in New Jack City white boy number
13:17 one right here I was driving the car
13:25 it's true
13:28 I still get checks from uh who's it from
13:34 um
13:36 Warner Brothers Warner Brothers
13:39 Holt yeah Warner Brothers
13:43 I got I got a residual check for seven
13:45 dollars and ninety cents
13:47 for what was the period it doesn't say
13:50 does it say
13:52 residual is 10.
13:55 year to date I've made twenty two
13:57 dollars from New Jack City
14:02 that's that's a celebrity right
14:05 all right
14:06 so Kyle was a celebrity exactly oh
14:09 yesterday
14:10 wait yesterday was sheep wait oh sure
14:12 bear not she bear working on a name yeah
14:16 sure bear
14:18 yeah I I okay sure bear not she bear
14:22 well I don't I'm reading fast what are
14:24 you gonna do all right well you come up
14:26 with a name and then
14:27 I'll keep saying Bears she bear sure
14:30 bear or the bear
14:32 depending on my mood and when you get it
14:34 nailed down we'll go there we'll go with
14:36 that
14:37 Sam corn hello good evening good evening
14:40 what day is today Wednesday witness day
14:42 Wednesday
14:45 what's interesting is gpt4 is a 32-2
14:48 token window normally but code
14:50 interpreter must have more yeah well Sam
14:53 so
14:55 gpt4 chat gpt4 still only has four
15:00 thousand maybe it has eight the
15:04 API I don't think it has 32 yet I think
15:07 the 80 the API only has 8 or 16
15:11 it might have 32 I don't know
15:14 but then yeah code interpreter you can
15:16 upload up to 100 megabyte file so code
15:19 interpreters got a significantly bigger
15:21 context window
15:22 here's what here's what I think is going
15:24 on with with uh with code interpreter is
15:28 it spins up a virtual machine inside
15:30 your chat session and I have a feeling
15:32 it's just it's like embedding whatever
15:34 your document is into a little mini
15:37 Vector database which is basically it's
15:40 basically gptizing your
15:44 whatever documents you upload into its
15:47 sort of local memory for that virtual
15:48 machine I think that's what's going on
15:50 there
15:52 um so so it doesn't have to play by the
15:54 same rules it can it can sort of Chunk
15:55 Up
15:56 what's what's things are in the vector
15:58 database there there can be a hell of a
16:00 lot more of it there than just the the
16:03 context window can do in a single chat
16:05 so
16:07 that's speculation but I think that's
16:09 what's going on there but yes the the
16:10 net effect for you as an end user is
16:13 code interpreter
16:14 you can have a lot more data in there so
16:17 you can either go to Claude
16:19 and do up to 60 or 70 000 words just in
16:23 the chat or with an upload or you can
16:26 use code interpreter the challenge with
16:28 code interpreter is because it's doing
16:29 the autonomous writing
16:33 um
16:33 the writing Python and interpreting it
16:36 and then seeing if it got it right and
16:38 it it sometimes you want it just to do a
16:41 simple thing and it starts a whole chain
16:42 of things and it takes a while so it can
16:44 be a pain in the ass so I've been
16:45 finding Claude pretty useful to just pop
16:48 a long piece of text in there and do
16:50 something with it
16:52 all right
16:53 so I had nine million data points
16:55 spreadsheet I wanted to put in code
16:56 interpreter wait had a 9 million data
16:59 point spreadsheet I wanted to put in
17:00 code interpreter so bad earlier did you
17:02 do it Jesse or is that too I don't know
17:05 what that translates into in terms of
17:07 megabytes maybe that's too much for it
17:09 if you got that sort of stuff like start
17:11 playing around Lang chain
17:14 because you can pop that [ __ ] in Lang
17:16 chain and go go to town
17:19 fun suggestion is Moises which which is
17:22 an AI music tool for splitting out
17:25 tracks and guessing chords so that's
17:27 cool I like music I like music music's
17:29 cool let's go to Moises Moises have you
17:32 seen Moises no but I used Moises isn't
17:35 that a lotion all right
17:38 Moises wait Moises
17:43 the musicians vocal remover and much
17:46 much more
17:49 You've Got Mail
17:51 start free yeah yeah same yeah you
17:55 palooka
17:56 oh sorry
18:00 um
18:00 [Music]
18:06 all right what do I do new
18:09 your playlist is currently empty
18:12 you're gonna upload a separate track a
18:16 master track
18:18 demo playlist
18:22 mode
18:30 [Music]
18:42 now listen
18:46 [Music]
18:47 let's champion
18:51 you singing Buddy you singing champion
18:58 [Music]
19:02 foreign
19:06 here
19:11 let's see
19:13 oh mute oh see
19:16 oh let's see so so it splits it into
19:18 stems basically oh that's pretty slick
19:22 so wait I can solo this
19:27 is
19:30 soon
19:41 [Music]
19:52 it's pretty slick that's pretty slick
19:54 it's a little muddy but it's
19:57 that's pretty slick
20:02 yeah like it didn't separate out the
20:04 violin
20:08 that's cool all right I'll definitely
20:10 check that out
20:11 moises.ai Studio Dot moises.ai
20:15 Moises it really just does sound like a
20:17 skin condition but
20:21 but thank you for that who's who gave me
20:24 that uh Sam corn thanks you Sam that was
20:27 cool that was very cute got coffee I do
20:30 got coffee Tobias was shaking
20:32 I got coffee and I got apple cider
20:35 vinegar water
20:37 because I know how to live
20:40 I finished my board deck I got back from
20:44 New York this morning at like
20:46 I don't know nine Denver time
20:50 and uh took a little nappy then I did a
20:54 little grocery shopping and I did a
20:55 little work
20:58 and then I got on the I got on a tick
21:01 tock I got on a tick tock with all you
21:03 fine fine people do me a favor if you
21:06 would we just tick Tick Tock didn't blow
21:08 out the algorithm tonight because they
21:09 don't like us so if you would share the
21:12 live and hit the like button see if we
21:14 can get some more people in here oh
21:15 they're just a few more just popped in
21:17 but still let's get some more folks in
21:19 here let's we're gonna make it a party
21:20 we want to make it a party and if you
21:22 haven't followed me what the [ __ ] is
21:23 your problem follow
21:25 follow share like I don't bite it just
21:30 takes me a while to answer questions and
21:31 I talk stupid
21:36 is he really live he hasn't answered my
21:39 question I don't think he's really live
21:41 why does he act like a dumb dumb
21:44 all right
21:47 is this even live why won't you answer
21:50 me
21:53 they won't do it time check 813 Colorado
21:58 813 Denver Colorado
22:02 do you know where your fat dog is
22:04 yes I do
22:06 yeah why is this Geezer Gen X not
22:08 answering me yeah yeah yeah see yeah
22:12 yeah see ya paluka he won't answer any
22:15 of these questions I don't think he's
22:16 really answering it he's probably an AI
22:19 I heard his GPT five
22:25 welcome to Moviefone
22:27 anyone live in New York in the 80s
22:30 welcome to Moviefone
22:33 press one for the Terminator Perez toe
22:38 wow is it it plus Health course two and
22:41 one oh is this it plus Health course 201
22:44 you know I'm versatile I'll give I'll
22:47 give advice on anything as as we like to
22:50 say about chat GPT and all these large
22:52 language models they're mansplaining as
22:54 a service
22:55 they will confidently answer you even if
22:57 they're wrong
23:00 aren't men the best
23:03 which we're just we're just [ __ ]
23:05 insufferable but you know we try we do
23:08 we do what we can nope Tick Tock just
23:10 jumped me to the bottom because it knows
23:11 I love that so much
23:17 all right we don't have too many
23:19 questions so it didn't it's not too
23:20 painful yet
23:21 uh I think that's true though it's
23:24 healthy yes indeed ahem we are the
23:28 Irregulars oh that's good I like that
23:29 the followers of the AI learning lab
23:32 welcome Irregulars
23:34 that's true we're probably all old and
23:37 half of us are constipated half of us
23:39 are just struggling to make it to the
23:40 toilet that's good that's solid that's
23:43 solid
23:44 I think I think I think you did it
23:46 Emilio's life I think we've got the name
23:49 for the for the Posse the Irregulars
23:52 it's good
23:56 I think that one's gonna stick
23:59 just like just like the poop in your
24:01 underwear
24:02 [Laughter]
24:04 he's a celebrity oh I'm a celebrity
24:07 oh wait Sonny and Cher was my response
24:10 to tick tock asking us to share oh yeah
24:13 okay you are freaking awesome thank you
24:16 Bruce B R Bruce B extrapolate is the
24:20 wrong word LOL but you know what I mean
24:22 okay I don't know when did I use
24:24 extrapolate I'm normally pretty good
24:26 with using words in the right context so
24:28 if I used it in the wrong context don't
24:30 believe you about New Jack City we'll
24:33 need the orb to con to confirm all the
24:37 orb actually the orb thing do I have it
24:40 here
24:41 I think I
24:43 um did I do a screenshot hang on let me
24:45 see if I can show it here
24:48 if I go effects
24:51 and I go green screen
24:54 and I go that
24:58 I can show you my leaves that yeah it's
25:00 here look
25:02 uh
25:07 all right
25:09 so there's the
25:11 there's the uh World ID verified at orb
25:14 Anonymous it's one of the cool things
25:17 about this technology is
25:20 I'm I'm verified but this is this is
25:23 like what they call zero proof so
25:26 I don't need to provide anything other
25:28 than this in whatever systems developed
25:32 to support this but it doesn't have my
25:34 name on it it just says if if you're the
25:37 owner of this particular wallet
25:40 you know that blockchain address if you
25:42 flip it over it's got the block the
25:44 public blockchain address on the back
25:46 um
25:47 as long as you have that you're verified
25:49 human anonymously which is pretty slick
25:52 it's pretty slick I like it it's slick
25:54 six six click
25:57 how do I get rid of this thing oh there
26:00 you go did it ah look at that old man
26:03 figured out the Tic Tac he figured out
26:06 that technology he's pretty good at it
26:09 it was surprising I don't know I don't
26:11 even think his fingers work how's he
26:13 tapping on the screen
26:18 I just subscribed to YouTube excited
26:21 great thank you Grandma sing where are
26:24 you where are you
26:25 like I was I was gonna say where are you
26:27 calling in from this isn't radio in the
26:30 70s hey girl where are you calling in
26:32 from Fantastic Poughkeepsie oh wow
26:36 I went to school there once got kicked
26:39 out if you know what I mean
26:42 is chat GPT creative were you in here
26:45 the other night and asked that same damn
26:46 question and I answered it I I answered
26:48 that question two nights in a row for
26:50 like 45 minutes
26:54 don't be [ __ ] trolling me
26:58 I don't I I'm
27:01 thank you Wes oh wesn radio wsn radio
27:04 welcome you still need dashes in your
27:07 name
27:11 um because it's Westerner Westerner
27:13 wrestner ad IO and cure
27:18 [Laughter]
27:21 wsn radio
27:23 the sound of smooth jazz here in the
27:26 evening
27:28 pull up a
27:29 nice glass of apple cider water sip it
27:34 back
27:35 feel the smooth smooth tones of Kenny G
27:38 here we go
27:43 I'm gonna get me a radio DJ Gig I'm
27:45 gonna do it we're gonna bring back radio
27:50 so here's what I did I put it it just it
27:54 it was too [ __ ] complicated to have
27:55 seven things on a thing so I put
27:59 Claude and Bing because I I just feel
28:03 like those three are the big three right
28:04 now plus the prompts.chat I put them
28:06 here so for people that are just getting
28:09 started those are the things to look at
28:13 the alternates barred
28:16 um the Llama models pie and Poe I just
28:18 put them on their own separate things so
28:20 this is like if you want if you want to
28:22 dig deeper or or as Apple says you want
28:25 to try some deep Cuts you know
28:27 you know here's your hits let's go to
28:29 the b-sides
28:32 and then I put the image things with the
28:37 prompt book for images all on the same
28:40 page because that made sense and then I
28:42 just stuck future PD at the bottom
28:44 because
28:45 future PD is okay but like
28:48 it's just I don't pay attention enough
28:50 to it so that's future PD is just the
28:52 thing that I do when people ask me [ __ ]
28:55 I don't know anything about I'm like go
28:57 research it for yourself it's right
28:58 there so anyway that's what's going on
29:00 with that so good eye good eye good eye
29:02 The Irregular is coming in hot with the
29:05 uh
29:06 with the observations
29:09 telling me I'm using words wrong
29:11 pointing out changes in the
29:13 what is this called the little window
29:19 um I dropped the Trump indictment PDF
29:21 into code interpreter which was great
29:23 best way to learn about oh that's a
29:24 great idea actually yeah those big long
29:26 indictments there's a lot of them
29:28 there's uh what is it 91 total charges
29:31 now yeah pop those indictments in there
29:33 and have it summarize it asking all
29:34 sorts of [ __ ] that's a great idea
29:38 you know what I've been doing with it is
29:40 um
29:41 I've been doing terms of service I you
29:43 know I I have I am not a manual reader
29:46 as we've talked about on this channel
29:47 before and so terms of service are even
29:49 less interesting me than manuals and
29:53 um
29:54 and so every time one of these
29:58 especially these AI things one every
30:00 time they they say we've updated our
30:02 terms of service which I normally would
30:03 ignore that now I'm like let me see what
30:05 you [ __ ] are up to and I go pop it
30:08 into code interpreter or uh Claude and
30:11 and uh
30:13 and and ask I just say summarize it in
30:16 like you know the the most important
30:19 10 things in in 10 short bullet points
30:21 and it'll do it it's pretty pretty dang
30:23 good what's line chain oh wait Bing is
30:26 sometimes gets very slow yeah exactly
30:30 um so I'll get the Lang chain in a
30:32 second Bing gets slow
30:35 uh well I think all of these things as
30:37 people use them they're using up lots
30:39 and lots of gpus that might be just
30:41 Microsoft kind of throttling the amount
30:44 of uh compute power they're willing to
30:46 throw at the free service they haven't
30:49 launched
30:51 um co-pilot the the chat GPT rolled into
30:53 Microsoft Office and Microsoft Windows
30:55 that hasn't launched yet they're gonna
30:57 charge 30 bucks a month per head per 30
30:59 bucks a month per head
31:01 for for that service
31:03 um which is decently reasonable for
31:05 businesses but for individuals it might
31:07 be a bit steep on top of you know Office
31:09 365 subscription
31:12 um that might just be a throttling thing
31:14 so that that's probably what that is uh
31:16 the thing was spelled moisey m-o-i-s-e-y
31:18 yeah that's correct
31:20 um Lang chain is a you have to uh
31:24 you have to be a little geeky to use it
31:27 but what Lang chain is is it's a chain
31:29 of different
31:31 um components that kind of string some
31:33 things together basically what happens
31:34 with Lang chain is
31:36 the the simplest way I can describe it
31:37 is you can chat GP you can gptize your
31:41 own data
31:42 so you embed your data and then you
31:45 store it in what's called a vector
31:46 database and and the vector database is
31:49 basically where you can search like chat
31:50 GPT
31:52 um and then there's other components to
31:54 it like you can search the web and
31:55 things like that so basically you're
31:56 making a little application that you can
31:59 upload your own data to and then you can
32:00 have your own version of chat GPT where
32:03 your gpting your own data so if you've
32:04 got all of your customer service
32:06 documentation for your company you could
32:08 pop it into line chain and create
32:10 yourself a little chat bot for either
32:11 customer service people or even for your
32:14 customers to actually find answers to
32:16 [ __ ] based on your documentation so it's
32:19 pretty powerful pretty popular
32:22 um and and very very flexible it does
32:24 require you to use your Noggin and your
32:27 coding skills it's not insurmountable
32:29 like it's it's in the neighborhood of
32:31 complexity of getting stable diffusion
32:33 up and running except there's a few more
32:35 components to it but it's not bad if you
32:38 want to learn about it just go to Reddit
32:39 and go to the Lang chain subreddit and
32:42 I'm sure there's all sorts of tutorials
32:44 there and you know Google collab
32:46 notebooks that you can just fire up and
32:47 and turn it on I'm sure so
32:50 all right so happy your flight brought
32:52 you safely back to home base thank you
32:54 Firefly I appreciate that my dogs are
32:56 singers too it's just so amazing like
32:58 you know we've had a bunch of dogs some
33:00 dogs can't watch TV
33:02 um some dogs can't hear music we had one
33:06 dog that watched TV it's just like when
33:08 the TV it goes on the dog was like hmm
33:11 dog and run across the TV but like oh
33:15 and then uh champ just if it's acoustic
33:18 guitar
33:19 um or brass bands like like when they
33:22 play the the national anthem in the F1
33:24 races on the weekend he sinks to those
33:26 two so he likes marching bands and
33:28 singer songwriter he's he's an Americana
33:31 kind of dog with the exception of the
33:33 international
33:34 uh national anthem
33:37 exception
33:40 and the other kind of music he just
33:41 doesn't care about classical music
33:43 occasionally if it's got if it's mixed
33:45 in a way where it sounds a little
33:46 acoustic he'll sing to that it's got
33:48 some some uh what's it called uh if it
33:52 just sounds acoustic e
33:54 all right
33:55 let's see sup dude hey Stevo what's
33:58 happening good morning hello you Hi how
34:00 are you welcome welcome welcome
34:02 is that puppy here on the T-shirt it is
34:05 indeed
34:07 I was taking a little nap on the puppy's
34:09 bed where he he looks out the window
34:12 and so it is uh
34:15 not a good idea in a black shirt come
34:18 for the information stay for the silly
34:20 voices
34:21 you've come to the AI learning lab but I
34:23 haven't learned anything yet shut up
34:25 it's gonna be good soon it's gonna get
34:27 better
34:29 it may not
34:31 love the dog hair on your shirt it's a
34:33 fashion accessory it is
34:34 how are your board meetings usually oh
34:37 how wait how are your board meetings
34:38 usually go how many fidget toys do you
34:40 bring Well normally I'm presenting some
34:43 of the stuff and then uh and I've got my
34:46 screen share on so I'm usually pretty
34:48 busy and you know
34:50 actually having to pay attention
34:52 um they usually go about an hour and a
34:53 half
34:54 um and we started we're like an 11 year
34:57 old company at this point so our board
34:59 members know us we know them so we can
35:01 be pretty efficient with them
35:03 and it's not a contentious thing we had
35:05 a we had a
35:06 a contentious investor for a while but
35:09 we bought him out so
35:11 that's good that's gone that that
35:13 headache is gone so we just got people
35:15 trying to do the rat thing we're trying
35:17 to figure some [ __ ] out
35:22 all right
35:23 is oh just jump me to the bottom King
35:26 gang it just yeah freaking Rock Em Sock
35:29 Em
35:35 [Music]
35:39 is that really apple cider vinegar or AI
35:42 juice oh wouldn't you like to know
35:45 wouldn't you like to know now that I've
35:47 had my Iris scanned maybe I got special
35:49 drinks
35:50 maybe I do maybe I'm part of a club
35:53 you're not a part of you have to
35:56 you have to have your own Anonymous
35:57 World ID to be part of the club I'm a
35:59 part of
36:01 whoa
36:05 these large chain molecules taste
36:07 delicious
36:09 all right I feel like we live in a world
36:12 a weird time warp or something no one
36:14 around me wants to talk about AI oh
36:17 April strangeways so yeah so here's the
36:19 deal with that [ __ ] them
36:22 I like I've got I've got I mean here's
36:25 the thing I was an entrepreneur in the
36:26 early days of the world wide web and I
36:28 got I got this crack addicted on
36:31 the early days of the web this is like
36:33 94.95 and I was a screenwriter and I was
36:36 in theater companies like I had you know
36:38 actor and writer friends and they all
36:40 just thought I was [ __ ] weird like
36:41 what the [ __ ] are you doing what are you
36:42 talking about why do you what is this
36:44 world wide web artist Consortium thing
36:46 why do you always like why are you
36:47 always talking about this and then I I
36:49 started agency.com with this big agency
36:52 and it got really big and it got really
36:53 famous and and everyone was like ah hey
36:56 you know I I should have listened to you
36:58 like five years later they're like this
37:00 is really amazing this sense of this
37:02 stuff's amazing I should have listened
37:04 to you I'm like okay yeah I mean yeah
37:06 you could I was
37:08 I was talking about it incessantly all
37:10 you had to do was not be a douche
37:12 all those same people why are you
37:15 talking about this AI stuff this AI
37:16 stuff's awful it was terrible I don't
37:18 know why you're talking about this I
37:19 mean [ __ ] them
37:22 [ __ ] him
37:23 yeah it's weird it's weird but here's
37:25 here's the here's Here's my thought on
37:27 it is we've got you know 45 or 50 years
37:30 of Hollywood is essentially telling us
37:33 that the robots are going to kill us
37:36 we've got technology that is going to be
37:38 disruptive is gonna is gonna
37:41 um cost people jobs I think it's also
37:44 going to generate a lot of jobs it's
37:45 going to generate a lot of prosperity
37:46 but it's going to be disruptive it's
37:48 going to be a mess
37:50 um and and it is
37:52 um it's a lot of change and it's very
37:55 existential right if you're a copywriter
37:58 the first time
38:00 us chat GPT to do something that you
38:02 know how to do that maybe you spend many
38:05 hours of your day doing and it does it
38:07 like that
38:09 and then you start getting the prompting
38:10 dialed in and it does it even better and
38:12 then at some point it does something in
38:15 five seconds better than you can do it
38:17 and at that point you go [ __ ]
38:19 what what's my role it's like it's it's
38:23 terrifying right and it's you know I
38:25 think about voice actors like all of the
38:28 you know they can now take three seconds
38:30 of your voice and synthesize that and
38:32 right now the synthesized voices are a
38:34 little you know robotic and monotone and
38:37 shitty
38:39 um but but you know I saw one that came
38:40 out I don't know a week or so ago that's
38:43 got all these different emotions in it
38:45 and when they start dynamically
38:47 assigning emotions based on the script
38:50 this stuff's gonna start getting more
38:52 and more realistic it's going to start
38:53 having the Cadence of a human it might
38:55 not have the Cadence of an add freak
38:57 like me talking a million miles an hour
39:00 but at some point it will
39:03 and so so like I get why it's scary and
39:07 and here's here's my other thought like
39:09 I get I get this scared thing hey thank
39:11 you very much for the rose
39:13 um I get oh w-e-s-n w-e-s-n now with
39:17 underscores I've changed my name I gave
39:20 you a rose and I'm hoping you call me
39:22 out welcome to wesn it's the hotline
39:26 how are we doing today what's your
39:28 request
39:30 Jane wants to know
39:32 all right
39:34 um
39:35 so thank you dragonfly so I get the um I
39:39 get the fear thing
39:42 um
39:45 I also get that some people
39:48 after having tried it are going to
39:50 decide not to use it I totally respect
39:52 that like what I'm not down with right
39:54 now is people that are sitting on the
39:56 sidelines [ __ ] all over it that have
39:58 strong opinions about Ai and they
40:00 haven't actually used it
40:02 if you've got strong opinions about AI
40:04 but you've actually used it cool like I
40:06 that's actually dialogue I'm really
40:08 interested in
40:09 but most of the people that have strong
40:11 opinions about AI haven't played with it
40:14 my thought is this this shit's coming
40:17 whether we want it or not like I kind of
40:19 think of it like
40:21 um
40:22 you know people deciding that they're
40:24 not going to use AI like like the
40:26 writers saying you know don't use AI to
40:28 write screenplays or the actors saying
40:30 you know you can never use AI in a
40:31 feature film it's kind of like the State
40:33 of Florida saying we're no longer
40:35 accepting hurricanes
40:37 it's a new law we just passed it okay
40:39 great
40:40 nice law great sentiment
40:43 but the hurricanes are still coming
40:47 I'm gonna prepare for those at all and
40:49 that's kind of my that's that's where I
40:51 am with this stuff is is like I don't
40:52 have judgment whether people like it or
40:54 not I don't give a [ __ ] like I get why
40:56 some people might [ __ ] hate it I
40:58 think especially creative people
41:00 you know a lot of creative people
41:03 they're they're self-worth doesn't come
41:05 from the output it comes from the
41:07 process if you love illustrating if you
41:09 love 3D modeling
41:11 this [ __ ] is gonna piss you off like I
41:15 get no enjoyment out of that it's just a
41:16 machine that squirts out [ __ ] even if
41:19 it's as good as what I do or better than
41:21 what I do
41:22 I get my enjoyment out of the process
41:24 great you won't use this stuff that's
41:27 cool but just know you're going to be
41:29 competing against people that are
41:31 and then but and and everyone can make
41:34 that choice and and you know there'll be
41:36 people that are not using AI that still
41:38 have good flourishing careers but those
41:40 careers are just going to be you know
41:41 narrowed you know increasingly down
41:45 um but yeah it is it's we're in a weird
41:47 weird time warpy kind of here's where we
41:50 are April we're in the beginning right
41:52 chat EPT is only nine months old
41:56 um
41:59 this [ __ ] ain't getting uninvented
42:01 there's there's way too much investment
42:03 way too much competition the last couple
42:05 of nights people have been saying oh
42:07 there's been this article that Chachi PT
42:08 may be out of money by the end of the
42:10 year okay they might be
42:12 or by 2024 whatever it was
42:14 um but I don't think that's gonna happen
42:16 and there's there's 40 companies
42:19 sitting in the wings waiting to get
42:21 their technology out there
42:23 um so so there's you know it's coming
42:26 it's coming so
42:28 um the other thing you can do April is
42:30 find community so I created a community
42:32 called the AI Salon we actually had some
42:34 people from in here came to the salon
42:36 and last night they were like hey I came
42:38 to the salon it was awesome it is it's
42:39 really cool so if you go to the salon.ai
42:42 there's a it's a link tree and the first
42:45 link talks about you know what the salon
42:48 is the second link is the Meetup the
42:50 third link is the Discord
42:52 um so you could join a group like that
42:54 it doesn't cost anything it's free
42:57 um and uh you could join a group like
42:59 that or find something locally find
43:01 people that are talking about this stuff
43:03 make new friends who who are exploring
43:05 this stuff and getting their heads
43:07 around it because we're all trying to
43:08 [ __ ] figure it out and if you've got
43:10 people that are like I don't know why
43:11 you do that that's a waste of time
43:16 you're enabling Beelzebub if you if you
43:19 got people like that just [ __ ] them
43:24 all right it's called Moises uh I don't
43:28 know the URL because I use the app
43:29 usually Moises is moises.ai
43:33 that music app
43:35 Studio dot
43:37 m-o-i-s-e-s dot a i
43:45 Silver Fox
43:47 um yeah same in my surroundings yeah
43:49 where people are like
43:50 no we
43:52 you're a dum-dum if you talk about AI
43:54 That's stupid that's stupid
43:57 that's stupid I like I like I like real
44:00 work I like I I'm a writer and I like
44:04 real words if yeah if you got those
44:06 people um
44:08 yeah the glitches is in the Mage Matrix
44:12 raffle
44:14 um hello everyone how far behind are we
44:16 well root source
44:18 I started this
44:21 I don't know maybe half an hour
44:23 yeah half an hour-ish behind but you
44:27 asked that question probably 15 minutes
44:29 ago so when you ask the question we we
44:31 had just gotten started but when I'm
44:33 answering it now we're behind I have 56
44:36 messages down below that I have not seen
44:38 yet
44:39 so so for the new people here welcome
44:41 yeah we got a few more people here my
44:43 name is Kyle Shannon AI learning lab we
44:44 talk about [ __ ] here if you haven't
44:46 played with chat GPT there's the URL
44:48 that's the official URL
44:50 I'm not getting paid by them I don't
44:52 give a [ __ ] if you use it but if you
44:53 haven't used it go use it go play right
44:55 now if you don't know at all what to put
44:58 in there go to this one prompts.chat
45:00 that'll teach you how to use it
45:02 and then these other two are just the
45:04 kind of two main alternatives to chat
45:06 GPT Claude and Bing Bing is chat gbt but
45:09 it's connected to the internet from
45:11 Microsoft and it's it's been hobbled a
45:14 bit so it's it's Bing is a pain in the
45:17 ass right now but it is connected to the
45:19 internet and Claude you can put in giant
45:21 text blocks and it reads it all right so
45:25 those are go play there if you're new
45:28 um so welcome everybody welcome welcome
45:29 welcome and then if you have questions
45:30 about AI pop them in the comments below
45:32 and I am behind but I do try to get to
45:35 everything it's just I I never succeed
45:38 this is like life you know you try to
45:41 get to your to-do list you have six
45:43 things on it you're like this seems
45:44 reasonable you get to the end of the day
45:46 there's 17 things on your to-do list and
45:49 you didn't get two of the six done
45:51 the [ __ ]
45:52 this is why I don't like to-do lists I I
45:54 see them as monuments of Shame
45:59 I have a philosophy philosophy about
46:02 email inbox zero and to-do lists if it's
46:06 important enough it'll come back
46:11 uh
46:12 I have a quick question is there an
46:14 add-on to help AI summarize articles and
46:16 books they're so yeah
46:20 um
46:21 articles so so probably let's here let's
46:24 go play I'll go show you how plugins
46:25 work
46:26 so if you pay your 20 bucks a month to
46:29 chat gbt
46:31 you can make money
46:32 look they added back the little
46:35 suggestions here that's all fancy all
46:38 right so now we're in chat GPT if you
46:40 don't pay your 20 bucks a month you get
46:42 GPT 3.5 if you do pay your 20 bucks a
46:45 month you get gpt4 Plus Code interpreter
46:48 plus plugins and at some point they may
46:50 bring back connection to the internet
46:51 but they haven't but there's plugins
46:54 that are connected to the internet okay
46:57 um so
47:01 if we go to plugins uh let me let me
47:03 show you one other thing because
47:06 someone's gonna someone because of this
47:08 probably gonna go okay I'll sign up for
47:10 the 20 Buck a month thing I don't see
47:12 plugins where are the plugins you have
47:14 to actually enable the plugins and and
47:15 code interpreter so you go to your
47:17 settings and then you go to Beta
47:20 features and you actually have to turn
47:21 them on so if you don't see them do that
47:24 okay
47:26 now that what did I just do what the
47:29 hell did I just do
47:30 um okay
47:32 we go to plugins so when you when you
47:34 have plugins you've got this drop down
47:36 menu and here's all the plugins I have
47:39 installed now just because they're
47:41 installed I'm not actually using them in
47:43 any given chat I can enable up to three
47:45 of them
47:46 so the one I have enabled right now is
47:48 web pilot
47:50 um
47:51 but what plugins do is they're like
47:53 little micro services that
47:56 they've basically described what the
47:58 micro service does and then when you
48:00 interact with chat GPT it it basically
48:03 says are any of the enabled plugins you
48:07 know needed to to service this query
48:10 and then if it is it'll fire off the
48:12 plug-in and it does what it does so web
48:14 Pilot's a good one so what we can do is
48:17 let's go find
48:19 let's go find an article well here's
48:22 here's a good one so so this is zapier's
48:24 terms of service so this could be
48:26 it could be a link to a PDF of a book it
48:29 could be a link to an article like this
48:33 um
48:33 for a PDF of a book you you I I I'll
48:37 give you another option but if it's
48:39 articles or in this case terms of
48:41 service
48:42 I'm going to copy the URL here
48:47 um
48:48 I'm gonna say
48:50 um
48:55 [Music]
48:57 please give me the five most important
49:01 things I am agreeing to in these terms
49:03 of service comma
49:07 in a
49:10 short bulleted list period
49:15 and so now what it's going to do is it
49:16 should fire off the the web pilot
49:18 plug-in oh wait no hang on I didn't give
49:21 it the URL oh wait oh actually yeah
49:24 please it's good it said of course
49:26 please provide the link yeah dumb dumb
49:27 okay there's the link now I'll give it
49:29 the link
49:30 all right so now fired off the web pilot
49:33 plugin
49:34 and it's going off and it's reading that
49:36 article
49:38 and now it should just give us those
49:41 five bullet points back
49:43 based on the provided content acceptance
49:46 of updated terms thank you for the
49:48 hearty heart Clint appreciate that
49:51 to access the service you must register
49:53 for a zapier account
49:55 some features may require payment
49:59 use of your service you can only use the
50:01 service for lawful purposes
50:03 and then data ownership and privacy this
50:06 is the big one
50:07 you retain the ownership of your content
50:10 including content from third-party
50:12 services however
50:14 by transmitting the content through the
50:16 service you grant zapier a limited
50:19 license to access use process and store
50:23 content for specific purposes like
50:25 providing and improving service to
50:26 zapier's privacy policy and data
50:29 processing addendum
50:30 also applied to your data okay
50:34 um so I'm going to say
50:38 please write a snarky poem about number
50:41 five
50:43 foreign
50:45 this is also how to use chat gbt it's
50:48 just for your own entertainment
50:50 in the realm of debt data fast and wide
50:52 number five stands with swagger and
50:55 pride you think you think you own all
50:58 with a click and a type oh dear user
51:01 that's just hype you upload and share
51:04 thinking it's free but number five
51:06 Chuckles it all comes to me with every
51:09 bite every tweet every song your data is
51:12 with five all days all day long
51:15 um you might think it's yours that cute
51:17 little meme but fives in the shadows
51:19 plotting a scheme privacy at scoffs a
51:22 little slide in a grin in the digital
51:24 age that's paper thin so next time you
51:27 upload think twice maybe Thrice
51:30 for number fives watching and it won't
51:32 be nice in the world of data the king of
51:35 the hive beware dear user of sneaky
51:38 number five
51:39 oh chat EPT you're just awesome all
51:43 right so so that's that's a way to do it
51:45 another way to do it is you can use code
51:48 interpreter when you use code
51:49 interpreter
51:51 you're actually going to upload a file
51:54 so you're gonna hit that plus button and
51:55 you can upload that's where you could
51:57 upload say to a PDF of a book
52:00 or you could go to Claude
52:02 claude.ai
52:12 and then they've also got a little
52:14 upload thing here you can just paste an
52:16 entire novel right in your chat window
52:18 and go ahead and do that
52:20 so that's how I would do that if I were
52:22 you hope that helps I hope that helps I
52:25 don't know if I help sometimes I know I
52:27 help other times I don't know if I help
52:29 because I don't I don't I can't see I
52:33 don't follow to the bottom I don't know
52:34 what people saying until later and I
52:36 realized oh I [ __ ] up
52:39 all right I really enjoyed the AI Salon
52:41 from Tobias it was incredibly
52:42 informative there were so many brainiacs
52:44 yeah there's a lot of smart people there
52:45 one of the things we do in the salon
52:48 is we have values and we we reinforce
52:51 them a lot so things like generosity and
52:53 curiosity and uh and empathy
52:57 like having empathy for people like a
53:00 lot of times people come to the salon
53:01 and they're like I just learned about to
53:03 say I should I'm [ __ ] scared like am
53:05 I gonna lose my job like it's it's
53:08 terrifying and and then you know the the
53:11 people that kind of lean into it and say
53:12 okay I'm willing to be curious and I'm
53:14 willing to and then and then those are
53:16 the ones that usually end up like
53:18 sharing what they learned like here's
53:19 what I learned and here's what I'm doing
53:20 now and now I'm really excited and you
53:22 know seeing that transformation is
53:24 amazing so thanks for sharing that
53:25 Tobias did you see the Lance Elliott
53:28 article on Forbes 8 15 about mega
53:30 personas no but that sounds interesting
53:33 let me go find it so I can read it later
53:35 thank you
53:39 um Lance
53:42 Elliott
53:44 Forbes
53:45 Mega personas
53:48 I think I think personas are interesting
53:51 the Bold promise of Mega personas oh
53:53 let's go you know what we could do yeah
53:56 we can this is this isn't going to pay
53:58 well is it no good okay
54:00 so we're gonna go do what we just talked
54:02 about
54:04 we're going to plugins I've got web
54:06 pilot selected
54:09 and they're gonna go
54:11 um
54:12 based on this
54:15 and then I put in the URL oh crap stop
54:18 generating
54:19 um
54:21 based on the above
54:24 article
54:26 um
54:27 let's say
54:32 please give me three short bullet points
54:35 that capture the main ideas of the piece
54:41 only one message at a time please oh I
54:45 [ __ ] up hang on
54:47 a
54:55 based on the article below please give
54:58 me three short bullet points to capture
55:01 the essence of the article
55:17 I don't know what that's from
55:21 that might be for my my smooth jazz
55:24 Channel
55:25 wesn how can I help you oh smooth jazz
55:28 88.
55:33 it's midnight do you know where your
55:35 doobie is based on the article by Forbes
55:37 introduction of introduction of Mega
55:40 personas in AI
55:45 imprompt engineering this scales up the
55:48 concept of multi-personas where wait
55:51 in prompt engineering for generative AI
55:54 the approach scales up the concept of
55:57 multi-personas where instead of
55:59 instructing the AI to pretend to be a
56:01 few distinct entities it could be
56:03 directed to simulate hundreds or even
56:05 thousands of personas
56:07 simultaneously holy [ __ ] that sounds
56:10 cool
56:11 Mega personas can be particularly loose
56:13 God what is with my tongue Mega personas
56:17 can be particularly useful for
56:19 simulating large-scale surveys
56:22 or group oriented analysis I might need
56:24 this for a project I'm working on right
56:26 now
56:27 who's the April strangeways why thank
56:30 you for sharing this I had not seen this
56:32 this is really good caveats and
56:34 considerations how does it do it does it
56:36 say I'm sure it does in the article
56:38 uh but uh one could instruct the AI to
56:41 simulate responses from a large group
56:43 Saving Times and research yeah I'm
56:45 actually getting ready to do this I'm
56:46 creating a bunch of synthesized data so
56:48 if if this project comes through
56:51 um they want to take input from 2 000
56:53 people and sort of process it in real
56:55 time so in order to develop the prompts
56:58 for that
57:00 I want to I want to sort of describe a
57:03 bunch of different kinds of employees in
57:06 a company and describe the company and
57:08 then have it generate you know simulated
57:10 responses from 2 000 people and turn
57:13 that into a CSV file that I'll then use
57:15 to prompt and dial in my prompts
57:19 um so I'm gonna go read that article
57:20 thank you for that that's really awesome
57:22 see you see what's happening here it's
57:24 there's a level of generosity here that
57:26 just you don't find this any old wear
57:29 this isn't
57:30 if you give me 19.95 1995 today I'll
57:35 teach you how to make money
57:40 oh well I suppose I could at this point
57:44 I could put together you can make money
57:46 with Chad gbt instruction booklet and
57:49 probably sell it for 19.95 because I'm a
57:52 celebrity as we've discussed earlier
57:54 this evening yeah super cool dude with
57:57 excellent stage or video presence super
57:59 thank you Firefly that's awesome
58:01 um it's a Chrome extension what is
58:04 oh wait
58:06 Oh Thomas curl check out give me
58:08 summary.ai it's a Chrome extension cool
58:11 okay great yeah yeah so if if uh if
58:14 folks have written extensions that do
58:16 like article summary book summary in
58:18 fact Bing sort of does it too although
58:20 it it was shitty for the longest time
58:22 let me see if that's gotten better
58:25 I'll show you this
58:26 so I think I've still got the article
58:28 here
58:30 um so let me go all right so I'm in Edge
58:34 so in Edge
58:36 there's right up here is a Bing sidebar
58:39 so I can I can flip open the Bing
58:42 sidebar we're doing this real time so
58:44 this may or may not work so this is a
58:46 little mini version of chat GPT in
58:49 theory it's aware of this this web
58:52 window so in this web window I'm going
58:54 to put in that Forbes article
58:58 all right so there's the article so in
59:00 theory
59:03 what's it now what's it saying here new
59:05 Bing with images you can tell it to make
59:06 images generate page summary all right
59:09 so I just clicked a button it
59:10 pre-prompted me to generate a page
59:11 summary
59:13 the early days of Bing this was all
59:15 [ __ ] up sometimes it would recognize
59:16 what was in that page sometimes it
59:18 wasn't
59:20 Mega personas are a new technique for
59:22 prompt engineering
59:23 the author introduces things the
59:25 challenges in caveats so clearly the
59:27 author used that because that's the same
59:29 term we heard before
59:30 so that's working pretty good
59:34 so I'm going to say
59:35 um
59:37 write me an example prompt
59:43 that would generate one of these Mega
59:45 personas based on what the author says
59:56 if it gives me a safety guidelines thing
59:58 I'm going to friggin smack it
1:00:01 one possible example prompt
1:00:04 that could generate a set of these
1:00:10 you are to pretend that you're a lawyer
1:00:12 who is a personality type
1:00:15 you are blank age-old in a location you
1:00:18 have been practicing for years
1:00:24 oh I see what it's doing okay so what
1:00:26 it's okay so the oh I've done this
1:00:28 before okay I this is actually really
1:00:31 cool can I make this bigger
1:00:33 um no I can't so so basically what they
1:00:36 did is they wrote kind of a Mad Libs
1:00:39 prompt you know you're a lawyer of
1:00:42 personality type of age
1:00:44 you're you're in a location you've been
1:00:47 doing it for so many years you've got
1:00:49 this satisfaction level
1:00:51 you're this gender you're this race
1:00:53 ethnicity
1:00:55 and then
1:00:57 each one of those things is a variable
1:00:58 they put them in square brackets
1:01:00 and then basically you say you know
1:01:02 create you know variables for for all of
1:01:06 those things and then answer that
1:01:07 question for a lot of different folks
1:01:09 and stick that in a in a CSV file yep I
1:01:12 could see that working that's super cool
1:01:14 that's super cool so another option here
1:01:16 other than the Gimme summary AI is use
1:01:19 the edge browser with their Bing sidebar
1:01:22 which now looks like it works again so
1:01:25 Bing getting better that I didn't use
1:01:27 that for a bunch of months like six
1:01:30 months because it just sucked but
1:01:32 that's the thing about this this
1:01:34 technology right now kids all you
1:01:36 Irregulars out there
1:01:38 um is this stuff is changing all the
1:01:42 time so the thing that was laughably bad
1:01:44 like like Google's barred was
1:01:47 embarrassingly bad for months six seven
1:01:50 months and now it's not
1:01:53 and
1:01:55 that feature in Microsoft Edge was
1:01:58 really bad and now it's not
1:02:00 um do me a favor if you would just go
1:02:02 ahead and share this live if you're
1:02:04 willing to do that if you're willing to
1:02:05 be weird in front of your friends that
1:02:07 are like why are you sharing some old
1:02:09 man talking about AI is he even
1:02:11 qualified
1:02:14 he's he seemed like he was a little
1:02:15 touched when I watched it what are you
1:02:17 doing with your life if you're willing
1:02:19 to put yourself out there
1:02:20 then more people can come see this
1:02:24 okay I'll do like the I'll be like this
1:02:27 okay we're gonna talk about the
1:02:29 artificial intelligence because it's I I
1:02:32 like the artificial intelligence it's
1:02:35 all right well I like intelligences you
1:02:37 can probably imagine from
1:02:42 what was I saying all right
1:02:45 yeah
1:02:48 it's the Chrome extension hola buenos
1:02:51 noches
1:02:55 Julio H welcome
1:02:58 I I said that with kind of a fake actor
1:03:03 accent I don't know how I did
1:03:07 um but but I think it says good evening
1:03:09 salute so hello or good day or uh from
1:03:13 Colombia hello from Colombia that's
1:03:15 awesome welcome welcome welcome welcome
1:03:16 I do not speak Spanish I just speak
1:03:19 there's a famous Steve Martin from one
1:03:21 of Steve Martin's early albums he goes
1:03:23 those French have a different word for
1:03:25 everything
1:03:26 he's like he's like he's like I was I I
1:03:29 get to the hotel I get in the cab I
1:03:31 would like to go to the hotel please
1:03:32 driver turns around
1:03:37 and he goes so then you try to ask them
1:03:39 with a French accent I would like to go
1:03:41 to the hotel please
1:03:43 [Laughter]
1:03:46 oh Stevie Stevie was funny back in the
1:03:50 day
1:03:51 before he got all too wrapped up in that
1:03:53 banjo I mean he's good at the banjo but
1:03:55 you know it's not quite as funny as the
1:03:57 arrow through the head come on
1:04:04 all right you have a great radio voice
1:04:06 thank you very much Alyssa thank you
1:04:08 very much welcome
1:04:10 welcome to wesn smooth jazz88
1:04:17 don't ask the same DM questions yeah
1:04:19 don't do it I'm too subscribing for you
1:04:22 thank you very much
1:04:24 appreciate that you high if we could be
1:04:26 wait if we could be just be in real time
1:04:29 then other people wouldn't need to ask
1:04:31 over and over I know but that would
1:04:33 require me to change how I do things and
1:04:35 I'm not willing to do that right now if
1:04:37 someone someone I think was Emilio's
1:04:39 wife said there's probably a way to do
1:04:41 this where if we could pull the API feed
1:04:44 this would actually be cool if we could
1:04:47 pull the API feed
1:04:49 from the chat
1:04:51 send it to open Ai and summarize what
1:04:54 the questions are especially if it's
1:04:56 like you know they're kind of Stack
1:04:57 ranked like here's the ones people are
1:04:59 talking about the most and maybe you
1:05:01 have one column with things I should
1:05:03 answer and another column was snarky ass
1:05:06 comments that I can make fun of
1:05:08 see that would be cool and you know Tick
1:05:11 Tock could do that but you know they're
1:05:12 not gonna so someone on here wants to do
1:05:15 a little development project you go go
1:05:17 have chat GPT teach you how to code then
1:05:20 you can figure out if there's a tick
1:05:21 tock API then you can write that
1:05:23 application it could be I don't know it
1:05:26 could be piped into my ear through some
1:05:28 sort of Bluetooth connectivity device
1:05:30 I'm telling you we could billion dollar
1:05:33 idea right there you just go nuts with
1:05:34 that thing and then well it wouldn't
1:05:37 really solve anyone else's problem but
1:05:39 mine avocado jumpsuit saw a startup that
1:05:42 uses AI to read and evaluate business
1:05:43 plan decks for VCS VC Associates scared
1:05:47 yes Ted yes
1:05:49 um
1:05:50 the the uh the folks that read
1:05:53 screenplays
1:05:56 like coverage on a screenplay
1:05:59 it's like all the associate agents at
1:06:02 William Morris like their job's done
1:06:05 like that that that ship has probably
1:06:08 sailed because you could take the
1:06:11 coverage documents that have already
1:06:12 been written one of the things these
1:06:14 things are really good at is mimicking
1:06:15 other documents so you could just say
1:06:19 um you know take this screenplay
1:06:21 put it into one of the uh plugins or
1:06:24 code interpreter take this screenplay
1:06:27 follow this format for tone and
1:06:29 structure and give me coverage of the
1:06:31 screenplay
1:06:32 you could knock out 100 of those in an
1:06:34 hour
1:06:37 so
1:06:38 here so here's a lesson so if you're one
1:06:40 of those people at William Morris that
1:06:42 reads screenplays
1:06:44 be the first one to get good at this
1:06:46 [ __ ] because at some point someone's
1:06:49 gonna go who knows how to do coverage
1:06:51 with AI I do okay come here you're high
1:06:54 you you get to keep your job
1:06:59 AI will not replace your job
1:07:02 someone using AI will
1:07:05 it's from Paul rateser from the
1:07:07 marketing AI Institute
1:07:10 that's the thing
1:07:12 that's the thing it's a thing
1:07:14 is it a good thing nope is it a bad
1:07:17 thing not for the person that keeps
1:07:19 their job
1:07:21 that's why this is here
1:07:23 get curious get learning
1:07:26 but what if I don't like it Kyle I don't
1:07:28 know then don't just don't be surprised
1:07:32 when this [ __ ] comes and smacks you
1:07:34 upside the head I didn't think it come
1:07:36 out could come after consulting jobs
1:07:39 I have a master's degree I get paid 200
1:07:43 000 a year sorry not gonna need your
1:07:46 services anymore this thing just does it
1:07:49 now is that really true no but some
1:07:52 amount of people in every knowledge job
1:07:54 in every creative job are going to be
1:07:56 displaced
1:07:58 within five years there'll be lots of
1:08:00 new jobs so a lot of those people are
1:08:02 going to migrate to new jobs
1:08:04 but the ones that don't make the
1:08:06 transition won't and it's going to be
1:08:07 really painful
1:08:10 it's gonna be really painful and I don't
1:08:12 like it I don't like that part of it but
1:08:15 I'm just messenger
1:08:17 are you even qualified to teach this
1:08:19 class when's the test oh there will be
1:08:21 tests there will be tests shortly after
1:08:25 the twerking launch is when we start the
1:08:27 testing Google vertex AI has some
1:08:30 amazing image question answering that I
1:08:32 haven't seen anywhere else I haven't
1:08:34 played with vertex AI let me see that
1:08:37 this is tonight I'm like learning more
1:08:38 from you than I'm teaching which is
1:08:40 awesome
1:08:43 um Google vertex AI
1:08:48 yeah I went on to watch him he's
1:08:50 supposed to be some sort of expert or
1:08:51 something and then everyone in the in
1:08:53 chat do more than he did every time they
1:08:56 said have you heard of this he's like no
1:08:58 I haven't heard of that
1:08:59 I don't even know what he does
1:09:01 and then it seemed it seemed like it
1:09:05 someone said he was an actor
1:09:07 it seemed like he's an actor
1:09:09 he doesn't know anything about
1:09:11 technology I think he's acting like an
1:09:13 expert that's what I think it is I think
1:09:15 he's acting like an expert
1:09:17 and he's pulling a wool over everyone's
1:09:19 eye acting like an expert and I bet he's
1:09:22 I bet he's signing them up for a
1:09:24 multi-level marketing scheme I bet this
1:09:26 is a pyramid scheme I knew it I knew it
1:09:29 I didn't trust him when I saw him there
1:09:31 was something about him I didn't like it
1:09:33 not one bit okay let's go look at ver
1:09:36 vertex
1:09:38 uh
1:09:40 uh unified data platform that offers
1:09:42 fully managed ml tools for text so I
1:09:46 assume this is something where you can
1:09:47 upload your own data
1:09:49 new customers get 300 in free credits
1:09:52 cool build deploy and scale machine
1:09:54 learning models for faster fully managed
1:09:57 ml tools and use case go to the console
1:10:01 huh oh sorry this is machine learning
1:10:03 developers data scientists and data
1:10:05 Engineers yeah see there's where you
1:10:06 lost me a bit
1:10:07 this is probably really powerful but
1:10:10 it's you're gonna generative AI Studio
1:10:12 try now
1:10:15 language speech and vision that this is
1:10:18 I'm starting to have PTSD of of the past
1:10:21 five years when whenever I would heard
1:10:23 about tensorflow like Google would say
1:10:25 there's a new version of tensorflow
1:10:26 that's really easy and I'd be like oh
1:10:28 good I can go play with machine learning
1:10:30 [ __ ] and I'd go there and be like I
1:10:31 don't have a math degree
1:10:35 I'm not an engineer
1:10:38 test tube
1:10:40 text prompt code prompt
1:10:43 oh this looks pretty cool so this is
1:10:45 basically just like
1:10:46 open ai's playground so yeah this looks
1:10:49 pretty good all right I'll mess around
1:10:50 with that so I got I've now got three
1:10:51 things to go check out this is cool
1:10:53 tonight thank you for that Sam
1:10:56 uh I'm 50 and about blind LOL can we see
1:11:00 a list a little larger for the moment
1:11:02 absolutely sorry about that
1:11:05 so I understand we've got we got
1:11:08 limitations but we got some superpowers
1:11:10 there what's your name oh April
1:11:13 um April we got some superpowers like
1:11:15 we've been on the planet long enough
1:11:16 that we can use these tools in powerful
1:11:18 ways so just screenshot that give that a
1:11:21 screenshot so there's your
1:11:23 there's your basics
1:11:25 here's your bonus category of uh large
1:11:29 language models Bard which is connected
1:11:31 to the internet that's Google's entry
1:11:34 uh perplexity AI is good on its own
1:11:36 especially if you're doing research and
1:11:38 want citations and then lab stop
1:11:40 perplexity is meta's llama 2 models that
1:11:43 you can play with pi is super chatty so
1:11:45 screenshot this
1:11:47 po you can create Bots you can play with
1:11:49 multiple large language models within
1:11:51 the same interface
1:11:53 so that's that
1:11:55 then this is all of your image not all
1:11:59 of them this is a handful of image
1:12:01 things mid journey is the big one the
1:12:03 that I think is still the leader of the
1:12:05 pack Firefly is not trained on
1:12:08 copywritten data that's from Adobe and
1:12:10 their generative fill is incredibly
1:12:11 powerful
1:12:13 um I mean Firefly is not bad and they've
1:12:15 got text tools there that are pretty
1:12:16 interesting Leonardo is stable diffusion
1:12:19 under the hood but a really cool web
1:12:21 interface on top clip drop is from
1:12:23 stability AI so from the makers of
1:12:26 stable diffusion this is like their
1:12:28 playground of experimental tools
1:12:31 and then the prompt book openart.ai
1:12:33 prompt book that'll teach you how to
1:12:35 talk to these image generation things
1:12:37 it's designed for stable diffusion but
1:12:39 the basic things still apply and then
1:12:41 futurepedia.io is
1:12:44 um
1:12:45 whatever you call it it's just a big
1:12:47 directory of AI tools all right so
1:12:49 screenshot that and then the final
1:12:51 little piece is uh that's the link to
1:12:54 the AI Salon that I um host
1:12:57 um there's a a whatever you call it a uh
1:13:01 link tree there with with links the
1:13:03 Meetup we meet every other week and then
1:13:05 there's a Discord link there and then
1:13:07 storyvine is uh my company and that's my
1:13:11 social media tag my tag my my app you
1:13:15 can at me there you can add me there
1:13:17 hang on the dog's line and I'll be right
1:13:19 back
1:13:23 what are you doing
1:14:21 foreign
1:14:31 one of my boys microwaved me some
1:14:34 chicken tenders
1:14:35 with blue cheese sweet
1:14:38 mm-hmm
1:14:40 okay
1:14:42 don't eat with your mouth full
1:14:45 [ __ ] rude
1:14:56 okay
1:15:02 how's it going with the workshop
1:15:03 planning um
1:15:05 I had a big client meeting this week and
1:15:08 I got a board meeting next week so I did
1:15:09 the board meeting
1:15:11 I've got a
1:15:14 15-minute
1:15:16 thing I got to record for virtual Work
1:15:18 Workshop
1:15:21 I've got a talk that I'm giving to a big
1:15:23 Pharma company
1:15:25 on the third I
1:15:27 here's my commitment
1:15:30 I think
1:15:36 um
1:15:37 I knew I shouldn't have put crazy glue
1:15:39 in my drink
1:15:43 um
1:15:45 I think the last week in August I can
1:15:47 probably start thinking about a workshop
1:15:49 like I've already done a workshop for um
1:15:52 for a big big company so let me put
1:15:54 something together let me put something
1:15:56 together I gotta get my [ __ ] together
1:15:57 and I'll just do it thank you for asking
1:16:00 thank you for
1:16:01 keeping me honest
1:16:03 writing them down cool
1:16:05 only voted three times in my life not
1:16:07 into politics yeah I'm I've voted but
1:16:10 I'm
1:16:14 I'm more into politics now that the
1:16:16 country is this polarized I think we
1:16:17 gotta we gotta fix some [ __ ]
1:16:20 I don't like where we are at all
1:16:26 okay y'all I'm glad he spends time and
1:16:29 actually answers questions in depth oh
1:16:31 thank you Tobias I appreciate the kind
1:16:33 words gpt4 is incredibly expensive to
1:16:35 run they spend seven hundred thousand
1:16:37 dollars a month hosting it I think it's
1:16:39 700 000 a day
1:16:42 would that be right that could be right
1:16:45 yeah I think it's I think it's
1:16:48 has Sam Altman described it it is
1:16:50 eye-wateringly expensive
1:16:54 wonder if they will include it with my
1:16:57 with M3 licenses and up I don't know
1:17:00 what M3 licenses are Chris Ward I assume
1:17:02 you're talking about Microsoft
1:17:04 bubble IO is creating apps but how
1:17:08 to get higher for SAS developer
1:17:12 um
1:17:15 I don't know
1:17:21 you know if you're if you're a SAS
1:17:22 developer I mean I would I would say
1:17:24 just lean into
1:17:26 like get good at really quickly spinning
1:17:29 things up right because
1:17:31 you know you can use these these tools
1:17:34 to speed up your coding
1:17:37 you know I would I would I would
1:17:39 definitely play around with Lang chain
1:17:41 if you're if you're a SAS developer I
1:17:43 would play around with Lang chain and
1:17:44 get that up and going because what you
1:17:46 can do then is you can
1:17:48 you could find
1:17:50 a sector of businesses that really want
1:17:52 to be able to use their own data but
1:17:54 don't have the technical chops to do it
1:17:56 so you could either build your own SAS
1:17:58 product for that or you could go to
1:18:00 those businesses and say hey I can build
1:18:02 you a your business GPT really really
1:18:04 quickly that might be a way to do it
1:18:07 again my My Philosophy on you know you
1:18:11 can make money is
1:18:14 it's it's it's completely 180 degrees
1:18:17 the wrong approach
1:18:20 you don't make money with the tool the
1:18:22 tool is not the thing right you can make
1:18:25 money with a shovel
1:18:27 how do you make money with a shovel I
1:18:28 don't know you figure out who needs a
1:18:30 hole
1:18:31 what kind of hole do they need what kind
1:18:33 of shovel do you need
1:18:35 and then if you can dig that hole better
1:18:37 than other people can dig that oh
1:18:38 they'll give you money and then you can
1:18:40 make money with a shovel
1:18:42 so so
1:18:43 if you're already a SAS developer what I
1:18:46 would say is you know figure out where
1:18:48 there's a business need
1:18:51 and then if you can find a way to
1:18:54 you know demonstrate that you're good at
1:18:56 that you can either go get a job with
1:18:57 the company or you can start your own
1:18:59 thing or you can start a Services
1:19:00 business or you can start an education
1:19:02 business
1:19:03 right do workshops do
1:19:06 put put stuff out there
1:19:09 all right Sam corn wow how can they
1:19:11 afford to pay 700k a month no it's is it
1:19:15 let's see how much it is I think it's
1:19:17 700k a day from that article is that
1:19:20 right let's see
1:19:28 how much does Google spend it or not
1:19:31 hang on
1:19:35 what is going on here
1:19:38 okay
1:19:40 how much does open AI spend
1:19:43 why is that not typing in there good
1:19:45 damn it how much does open AI spend her
1:19:53 day in compute
1:19:59 100K per day 3 million a month according
1:20:02 to estimations
1:20:13 chat GPT
1:20:16 yes 700 000 per day that's from The
1:20:18 Business Insider article from April 20th
1:20:21 seven hundred thousand dollars a day
1:20:24 now they've lost users so maybe it's
1:20:28 down to just half a million a day
1:20:31 so so you thought your electric bill was
1:20:34 high
1:20:38 seven hundred thousand dollars a day a
1:20:42 day a day
1:20:44 so let that do
1:20:46 the people come in and they say I want
1:20:48 you to write me a little poem about my
1:20:50 doggy it's like watching four dollars
1:20:53 [Laughter]
1:20:58 the other thing in addition to Lang
1:20:59 chain that's super useful if you're
1:21:01 interested in coding with AI is hugging
1:21:03 face yeah so hugging face is a site that
1:21:06 it's like GitHub where but there's lots
1:21:08 and lots of Open Source models like lots
1:21:11 of different Frameworks AI Frameworks
1:21:13 for everything from images to animation
1:21:15 to music and everything in between
1:21:18 um both GitHub and hugging face assume
1:21:20 that you know what the [ __ ] you're doing
1:21:22 with tech if you don't you're gonna have
1:21:24 to
1:21:25 get over that hurdle the good news is
1:21:27 you can use Chachi PT to create
1:21:29 tutorials for you so like one thing you
1:21:32 can do is you can take the readme file
1:21:33 from a GitHub repository or from a
1:21:35 hugging face thing take the readme file
1:21:37 pop it into chat EBT and tell it to give
1:21:40 you a step-by-step instruction for how
1:21:42 to get that thing up and running and
1:21:43 it'll do it you can even say I'm on a
1:21:45 Mac or I'm on a PC or I'm on Linux
1:21:47 whatever it might be if you're on Linux
1:21:49 you're probably okay with command line
1:21:51 [ __ ]
1:21:53 um but yeah hugging faces is is really
1:21:55 good and hugging face also lets you um
1:21:57 they host a lot of these models so you
1:21:59 can actually go play with different
1:22:00 models there my experience has been that
1:22:03 a lot of times
1:22:05 the hosted models are sometimes
1:22:07 maintained and sometimes not and they
1:22:09 sometimes work and they sometimes don't
1:22:11 and it's really confusing to sift
1:22:13 through all of the [ __ ] so it it's
1:22:15 definitely experience to get up to speed
1:22:17 but it's it's an amazing uh resource so
1:22:22 uh if you if you're doing any like the
1:22:23 person that was talking about SAS before
1:22:25 like yeah you should be you should be
1:22:27 all over the sub subreddits for all
1:22:29 these different projects
1:22:31 um you should be all over GitHub you
1:22:33 should be all over hugging face you
1:22:34 should be finding people like David
1:22:36 Shapiro
1:22:37 um who's who's you know building really
1:22:40 sophisticated
1:22:41 um AI tool AI systems out of these tools
1:22:45 and doing YouTube videos about it he's
1:22:47 amazing
1:22:48 Julio H okay bye friend Salon Julio I'm
1:22:51 sure you've already gone but thank you
1:22:52 for coming Chris word trying to fly away
1:22:55 I can create a team's bot that can do my
1:22:57 work for me
1:22:59 um
1:23:01 I
1:23:02 I think you could probably do that like
1:23:04 I I don't think you're that far from
1:23:06 that so
1:23:09 um
1:23:10 I mean let's let's go
1:23:12 let's go where do we want to go I'm
1:23:15 gonna try Bard for that one oh wait did
1:23:17 I just delete one of my slides I did
1:23:20 that was stupid that was dumb dumb
1:23:22 he was a dumb he was he had a little
1:23:24 slide there had good information on it
1:23:26 and he just deleted it I don't know what
1:23:28 he did
1:23:30 um what am I doing oh teams okay so I'm
1:23:32 gonna go to bard.google.com let's see
1:23:51 hello
1:23:54 hello
1:23:59 dang it
1:24:00 um
1:24:02 can you find latest
1:24:08 documentation on
1:24:11 creating teams Bots
1:24:16 I'm getting lots of hearty Hearts I
1:24:18 don't know I'm getting lots of hearty
1:24:19 Hearts but thank you for the hearty
1:24:20 Hearts I appreciate the hearts the heart
1:24:22 makes me happy
1:24:24 and I'm happy
1:24:26 I'm not programmed to assist with that
1:24:27 well that sucks jerk let's go try Bing
1:24:32 that I'm not gonna use you
1:24:34 so that's what you would call a safety
1:24:36 guard rail where
1:24:38 these systems are designed to not give
1:24:40 you what you want because the lawyer
1:24:42 said it's bad all right
1:24:45 um
1:24:55 can you find me the
1:25:00 [Music]
1:25:00 um
1:25:02 all right it's it's not like in the the
1:25:05 oh wait I would like access to your
1:25:07 microphone hang on allow
1:25:09 let's try this again
1:25:11 can you find me the latest documentation
1:25:14 for the teams API
1:25:17 as I want to write oh well it just took
1:25:21 that
1:25:28 um
1:25:31 sure here's the latest documentation
1:25:35 all right
1:25:36 if I wanted to
1:25:40 if I wanted to write a bot could you
1:25:42 assist me with that
1:25:49 if I wanted to write a bot could you
1:25:51 assist me with that
1:25:59 oh of course Microsoft has lots of
1:26:02 resources oh this is Microsoft right
1:26:04 this is Bing
1:26:08 msteams chat bot in 10 minutes creating
1:26:11 a chat bot with the power of teams
1:26:12 virtual agents tips and tricks
1:26:15 um
1:26:20 I would like to simulate
1:26:23 being at my desk oh stop responding
1:26:28 I would like to simulate I would like to
1:26:30 simulate being at my desk and pretending
1:26:34 that I am working
1:26:36 how would I go about creating such a bot
1:26:40 this it might this it might kick back
1:26:42 and go oh wait oh God damn it
1:26:47 I am thinking about writing about that
1:26:49 can simulate someone sitting at their
1:26:51 desk working period how might I go about
1:26:54 that
1:26:57 all right let's see what it does
1:27:00 looks like something went wrong start
1:27:02 over yeah I had a feeling
1:27:18 yeah it
1:27:25 I'd like to write a bot for Microsoft
1:27:27 teams that simulates someone sitting at
1:27:30 their desk doing work how do I do that
1:27:39 I just have a feeling anything and kind
1:27:40 of like it
1:27:57 can you help me write the code for the
1:27:59 first version
1:28:10 no it said no
1:28:13 um
1:28:15 yeah I I what what I would do
1:28:18 what I would do
1:28:23 is I would go to um
1:28:27 I would I would go find the go find the
1:28:30 documentation
1:28:32 and then
1:28:35 turn it into a PDF and pop it into code
1:28:38 interpreter and and it could help you
1:28:40 write it could probably help you write
1:28:41 those Bots or go to Claude
1:28:44 um or just take enough of the
1:28:46 documentation into into gpt4 and have it
1:28:48 have it code the bot for you I think you
1:28:50 could probably spin something like that
1:28:52 up in less than an hour I would think
1:28:56 so I like it is leonardo.ai any good
1:28:59 yeah it's really good it's really good
1:29:01 I'll go show it to you
1:29:03 so leonardo.ai is one of the ones I have
1:29:06 on my little list
1:29:07 it's one of the ones I like because it's
1:29:10 it's just it you don't have to program
1:29:12 anything so it's all the fancy stable
1:29:14 diffusion stuff
1:29:17 um but it's if it has a wizard it's got
1:29:20 to be good it's got a wizard I like a
1:29:22 wizard everybody's a good wizard
1:29:24 you like a wizard
1:29:30 uh
1:29:32 um
1:29:35 welcome to leonardo.a.i
1:29:38 so it's got a cool dark background gives
1:29:41 you a platform overview this is all new
1:29:42 I haven't seen this sort of teaches you
1:29:45 about the Leonardo Academy
1:29:49 let's go
1:29:51 so what you got here is across the top
1:29:53 you have different styles
1:29:55 so like there's a cool sort of anime
1:29:58 pastel dream
1:30:01 um absolute Beauty here's like
1:30:04 deliberate that looks like 50s poster
1:30:06 art right here's like a little you know
1:30:10 Pixar or
1:30:12 you know cartoony kind of thing
1:30:16 isometric sci-fi buildings paper art
1:30:19 cute anime vintage photography right so
1:30:22 you got different styles so you can
1:30:24 generate within that then you have like
1:30:26 here's all the recent [ __ ] that's been
1:30:27 generated any one of these you can see
1:30:29 what prompt they used
1:30:32 right so you can just scroll down so you
1:30:34 don't you don't even really need to
1:30:36 do like retro style theme so I can click
1:30:39 on this
1:30:41 and then can I generate with this no
1:30:43 it's not letting me some of them you can
1:30:45 generate with
1:30:49 like here's a seamless pattern
1:30:52 why am I not allowed to generate with
1:30:53 this model I don't get it
1:30:57 it's that
1:31:00 text to image generation that did I do
1:31:03 that
1:31:06 let's see Future model
1:31:08 um let's say
1:31:10 seamless pattern
1:31:13 of
1:31:16 chubby
1:31:17 black and white
1:31:19 cartoon dogs
1:31:23 on a on an azure
1:31:28 background
1:31:31 see what it generates
1:31:33 you're currently on the free plan
1:31:36 well you don't need to shame me about it
1:31:38 good Lord
1:31:39 I'm just trying to do my thing here
1:31:43 so the I think this is some things I did
1:31:45 before did I do these
1:31:47 yeah like I did this 50s wallpaper
1:31:49 pattern I did all these puppies
1:31:52 those puppies is cute oh look it did it
1:31:55 chubby black and white dog
1:31:58 um editing canvas crisp upscaler
1:32:02 let's see this one's cute
1:32:04 look that's cute I just made that
1:32:07 I just made that and then you know what
1:32:09 else you can do Leonardo no Kyle what
1:32:11 that's why we're here if if we knew what
1:32:12 to do on Leonardo we wouldn't be asking
1:32:14 you you dumb dumb oh yeah right okay
1:32:16 fine so you can go to training and data
1:32:19 sets and you can upload your own
1:32:21 pictures and train your own data sets
1:32:23 and so if I go to future or fine-tuned
1:32:27 models
1:32:28 and so here's all the models other
1:32:30 people have done that you can go play
1:32:32 with
1:32:32 but then I can go to your models and
1:32:35 I've got two of my giant fat face but I
1:32:37 have one of these cloud pictures that I
1:32:40 took
1:32:42 generate with this model and so now
1:32:45 it's my model
1:32:47 that I'm generating in so I can do
1:32:53 um
1:32:58 skyline of Manhattan
1:33:02 why is it not typing
1:33:04 Skyline
1:33:06 or all right skyline of Manhattan
1:33:14 from the Brooklyn Bridge
1:33:20 let's do Lower Manhattan because I was
1:33:22 just there the lower Manhattan
1:33:26 from the Brooklyn Bridge
1:33:29 during
1:33:32 um
1:33:33 Manhattan
1:33:36 hinge
1:33:38 do you all know what Manhattan hinges
1:33:40 it's so cool it's the one day a year
1:33:43 where the sunset
1:33:45 um sets exactly in line with the with
1:33:49 the cross streets of Manhattan it's
1:33:51 pretty cool
1:33:53 foreign
1:34:01 oh Stevo 42 you were chat gpt4 that's
1:34:03 pretty funny
1:34:11 those are good pictures but it didn't
1:34:13 really
1:34:14 it's sort of used my skylines but not
1:34:16 really it it more used my editing style
1:34:19 it's kind of my editing Style
1:34:21 but it's not quite yeah it's actually
1:34:23 not bad that's my my clouds tend to be a
1:34:26 bit more dramatic than that
1:34:27 but those are good
1:34:29 they're not really the Brooklyn Bridge
1:34:31 at all
1:34:35 well that one is that's well that's one
1:34:39 part of the Brooklyn Bridge but the
1:34:40 middle like the other things right here
1:34:42 like that's just wrong
1:34:44 and that just missed it entirely but
1:34:48 anyway
1:34:49 so you can do that
1:34:51 um what else can you do here I don't
1:34:52 know you can do lots like what's nice is
1:34:54 so Leonardo gives you sort of all of the
1:34:57 bells and whistles of stable diffusion
1:34:59 or kind of the most recent most stable
1:35:02 ones but but just that the whole idea of
1:35:05 being able to start with you know a
1:35:07 pre-existing model or start with
1:35:08 pre-existing images and then render
1:35:10 within them like that's a super cool
1:35:12 style like it would take you a while to
1:35:14 figure out how to do that now you can
1:35:15 just go here
1:35:17 and you know jump in and and do
1:35:20 something in that style right
1:35:22 so you could do
1:35:24 um scary
1:35:26 Kitty
1:35:29 um yawning
1:35:39 we'll see if it
1:35:41 is the right uh model
1:35:46 no it didn't do the drippy paint thing
1:35:49 but it is a scary kitty yawning
1:35:54 so anyway yep I like Leonardo
1:35:59 um
1:36:01 I like it a lot
1:36:03 I got a lot time is it 9 39 I got I got
1:36:06 like 20 more minutes you know I gotta I
1:36:09 could get some shut eye I've been
1:36:10 traveling all day
1:36:13 but I still come here for my Irregulars
1:36:16 we call them the AI learning lab
1:36:17 Irregulars
1:36:19 because we're older
1:36:21 and [ __ ] just don't work anymore
1:36:26 I love that Louis CK routine where he's
1:36:28 like I went into the doctor my ankle
1:36:31 hurt he goes yeah just wore out it's
1:36:34 doesn't really mean it work because yeah
1:36:36 it's just that's how it's gonna be now
1:36:44 it's so true it's so true just ship
1:36:47 breaks it just breaks
1:36:49 Why is my
1:36:51 got my little camera a little too high
1:36:54 there we go
1:36:57 that that
1:36:59 all right
1:37:01 hey oh just jump me to the bottom dang
1:37:04 it dang it
1:37:06 welcome everybody my name is Kyle
1:37:07 Shannon this is the AI learning lab if
1:37:09 you're new here
1:37:10 um if you have not played with chat GPT
1:37:14 go there go to chat.openai.com that's
1:37:17 the official chat gbt website
1:37:20 and if you want to learn how to prompt
1:37:22 and you don't even know what prompting
1:37:24 is and you're like what the hell you've
1:37:25 been talking about go to prompts.chat
1:37:27 and that will teach you how to talk to
1:37:28 these things
1:37:29 and then these other two things are cool
1:37:32 things to
1:37:33 um
1:37:34 to compare
1:37:36 chat GPT too
1:37:43 all right
1:37:45 I lost my place in my comments so I'm
1:37:48 trying to find out where I was
1:37:51 in comments
1:37:54 it is per day the seven hundred thousand
1:37:56 dollars yes
1:37:58 let's just start right there I think I
1:38:00 just skipped some questions but that's
1:38:01 okay bear thank you so much for the
1:38:04 dangly lights I love the dangly lights
1:38:06 they make me happy
1:38:10 Lang chain great to expound on
1:38:12 I've not actually set up Lang chain
1:38:15 um I think I did expound on it but
1:38:17 basically you can with Lang chain you
1:38:19 can
1:38:21 create a gptization of your own data is
1:38:24 the is the sort of short way to say it
1:38:26 there's lots it can do but that's kind
1:38:28 of it in a nutshell you can take your
1:38:30 own data you embed it you put it in a
1:38:32 vector database and then you can search
1:38:34 on it just like you search with other
1:38:36 stuff so it's quite powerful
1:38:38 um
1:38:39 if if you don't want to geek out
1:38:41 that sort of capability is is going to
1:38:44 be coming right
1:38:45 Google's gonna provide that with Google
1:38:48 Drive
1:38:50 um Microsoft will provide that with
1:38:52 their Office 365 suite and their cloud
1:38:55 storage
1:38:56 um AWS will probably provide that Apple
1:38:59 will probably provide that so
1:39:02 right now if you're geeky and and want
1:39:05 to figure this [ __ ] out
1:39:06 go play with Lang chain because because
1:39:08 if nothing else you'll learn a lot about
1:39:10 the different language models and how
1:39:12 you can do that and what embedding is
1:39:14 and what the different Vector databases
1:39:16 do and combining semantic search with
1:39:18 Vector search and there's all sorts of
1:39:21 permutations you can do with that
1:39:24 um but but don't feel like you have to
1:39:27 do that right now that's sort of you
1:39:29 know
1:39:31 Advanced generative AI
1:39:33 but if you're geeky I think it's time
1:39:36 well spent even if other stuff's coming
1:39:38 yep hire the folks that use this to use
1:39:41 that shovel
1:39:42 that seems too high what the seven
1:39:44 hundred thousand dollars a day
1:39:46 is it on Azure it is on Azure yes seven
1:39:49 hundred thousand dollars a day a day
1:39:54 yeah
1:39:55 well they raised 10 billion dollars for
1:39:58 Microsoft that wasn't in cash
1:40:01 that was in compute
1:40:04 so
1:40:06 that might even be the non-real retail
1:40:08 price right
1:40:11 we need solar power for AI wow what a
1:40:14 burn rate yeah it's crazy this stuff's
1:40:16 crazy but here's the thing like I you
1:40:18 know I've got these split on different
1:40:21 um things now but you've got those those
1:40:23 three
1:40:24 you got those four
1:40:27 plus you've got whatever Elon Musk is
1:40:30 going to do plus cohere plus Apple
1:40:34 plus AWS might come out with something
1:40:37 kind of consumer like they might do
1:40:39 something with Alexa
1:40:41 so that's 11.
1:40:45 and then there's probably three or four
1:40:47 waiting in the wing so you so I I kind
1:40:49 of feel like by the end of the year
1:40:51 there's going to be 15
1:40:54 major
1:40:57 viable large language models
1:41:01 trying to figure out how to get people
1:41:03 to use their service so they're going to
1:41:06 keep one-upping each other in terms of
1:41:08 of services
1:41:11 um
1:41:11 so even though that's a ridiculous
1:41:13 amount of money so here's the stat for
1:41:16 you
1:41:17 Apple just said that they have spent the
1:41:20 bulk of their
1:41:22 24.6 billion dollar research budget
1:41:26 for this year
1:41:28 on generative AI
1:41:31 the market cap of
1:41:34 open AI is less than Apple's research
1:41:38 budget
1:41:39 foreign
1:41:42 billion dollar research budget is what
1:41:45 Apple's just toying around with
1:41:48 so I think when Apple comes into this
1:41:51 Market they're going to come in
1:41:54 quite powerfully like they do
1:41:57 and they and they will likely redefine
1:42:02 when Apple comes in it ain't going to be
1:42:04 this geeky
1:42:06 it's gonna be really slick
1:42:09 as as Jobs would say it'll just work
1:42:12 right these things right now kind of
1:42:15 work if you know how to prompt and you
1:42:16 know how to do this and you haven't done
1:42:17 that you don't go over your token rate
1:42:19 and you you got to learn this one does
1:42:21 that and that one does that and that
1:42:22 one's connected to the internet but if
1:42:23 those don't work then you can go to
1:42:24 these three and then you go to the that
1:42:27 won't be apple Apple's gonna be like we
1:42:29 don't talk about this [ __ ] until it's
1:42:31 good
1:42:33 let the caveat that the past decades
1:42:35 kind of sucked for them but you know
1:42:37 other than that
1:42:39 uh how does this dovetail with
1:42:41 blockchain and Bitcoin so
1:42:43 um
1:42:44 I showed earlier
1:42:47 I'll do it again
1:42:49 I did it again I'll just do it I'll just
1:42:51 do it right now
1:42:55 so I let me how do I get rid of that
1:42:58 there you go so this is my this is my
1:43:02 world ID so so it says verified at orb
1:43:06 World ID Anonymous so this is my
1:43:09 verified blockchain
1:43:11 I am a verified human on the blockchain
1:43:14 this is Sam Altman's company called
1:43:17 World coin and you verify your humanness
1:43:20 on
1:43:21 this little orb that scans your irises
1:43:24 which everyone's like you're an [ __ ]
1:43:26 why would you do that I had a couple of
1:43:30 people Taco said well you just earned an
1:43:33 unfollow buddy I'm like okay like I
1:43:37 don't think I did anything to you but
1:43:38 whatever
1:43:40 um
1:43:42 um
1:43:44 so why did I do that why did I get
1:43:47 myself you know scanned and use my
1:43:49 biometric data for that because I think
1:43:52 that
1:43:53 um
1:43:54 moving forward
1:43:57 it's gonna be so easy to
1:44:01 um
1:44:02 fraudulently do everything represent
1:44:05 someone else synthesize their voice
1:44:08 make it seem like the New York Times
1:44:10 wrote a scathing piece on someone that
1:44:12 they did they really didn't write it's
1:44:13 it's just going to be all over the place
1:44:15 and it's going to be really hard to
1:44:17 discern
1:44:18 and so I think you need a foundation of
1:44:20 trust right and so this is a system it's
1:44:22 Sam Altman's company so the CEO of chat
1:44:24 GPT
1:44:26 just launched this two weeks ago and I
1:44:28 was in New York City you can you can get
1:44:29 this done in New York Miami and San
1:44:32 Francisco right now but I assume they'll
1:44:33 have more things coming
1:44:36 um
1:44:38 and
1:44:40 you know the ability to be able to say
1:44:42 like imagine a Twitter where you know
1:44:45 that everyone on Twitter is blockchain
1:44:47 verified as a human
1:44:49 and then every post that they post is
1:44:52 blockchain verified that yeah they wrote
1:44:54 that and it hasn't been modified in any
1:44:56 way and yeah this video that Kyle put
1:44:58 out is of Kyle and it hasn't been duped
1:45:01 um
1:45:02 so I think that AI is going to drive the
1:45:06 adoption of I don't know if it's Bitcoin
1:45:08 and blockchain but you know some sort of
1:45:11 tokenization Technology you know
1:45:12 blockchain maybe makes sense
1:45:16 um and then some sort of payment
1:45:17 infrastructure like like uh this the the
1:45:21 world coin thing that Sam Altman did
1:45:23 this thing
1:45:24 they've got three components they've got
1:45:26 a cryptocurrency so that's the cash
1:45:28 they've got the ID which is the
1:45:30 verification and then they've got a
1:45:32 transaction platform like you know
1:45:34 stripe or or PayPal or [ __ ] like that
1:45:37 but it's it's designed to use with this
1:45:39 so you could imagine building so every
1:45:42 e-commerce
1:45:44 site that's out there right that all
1:45:46 takes cash
1:45:48 you could imagine the eBay version of
1:45:50 this where everyone on eBay is a
1:45:52 verified human or Etsy right
1:45:55 Etsy right now is a cesspool of just
1:45:57 Chinese crap right because no one's
1:45:59 really verified and everyone can just do
1:46:01 these fake accounts imagine having an
1:46:03 infrastructure where you could say yeah
1:46:04 no this is you know the the AI learning
1:46:07 lab Irregulars like we verify that these
1:46:10 are these are people here not Bots we
1:46:12 can do that with this so I think it's
1:46:13 going to drive it um I don't it's not
1:46:16 going to do it anytime soon it'll
1:46:17 probably be three years out but
1:46:20 um but I actually think it's important
1:46:21 and uh yeah so there you have it there
1:46:25 you have it yeah yeah there you have it
1:46:27 yeah see what I'm saying yeah yeah see
1:46:29 yeah yeah see oh who's that that's the
1:46:32 uh that's the uh
1:46:34 that's the brother from Napoleon
1:46:35 Dynamite doesn't he talk like this with
1:46:37 his lower jaw and he just moves his
1:46:39 bottom lip yeah yeah you know it's
1:46:41 really good I threw the football
1:46:43 I see the football
1:46:45 isn't it him
1:46:48 eat the food
1:46:50 dude I'm 57 year old Dev laid off last
1:46:53 week and you've inspired me again thank
1:46:54 you MK thank you for that um 58 here
1:46:58 welcome welcome gen xer let's go [ __ ]
1:47:00 some [ __ ] up
1:47:02 um yeah
1:47:03 here's the deal I I just I I think that
1:47:07 any time spent
1:47:08 in the next 18 months getting your head
1:47:11 around this [ __ ] and getting articulate
1:47:12 with it and the thing that I would
1:47:13 encourage you to do MK is as you're
1:47:15 learning talk about it talk about it on
1:47:18 LinkedIn talk about it with your friends
1:47:20 at dinner parties like let them think
1:47:22 you're a [ __ ] weirdo aren't you too
1:47:24 old to be learning something new yeah go
1:47:25 [ __ ] yourself
1:47:28 learn this stuff because
1:47:31 with your life experience you'll you'll
1:47:34 be able to take the the sort of coding
1:47:36 superpowers the gpt4 does and you'll be
1:47:40 able to use them in a way that that'll
1:47:42 you know obliterate someone who doesn't
1:47:45 know how to do that like myself like I
1:47:47 can make code with this
1:47:49 but I can't look at the code and know if
1:47:51 it's any good
1:47:52 right so if you can look at the code and
1:47:54 go oh that's structurally [ __ ] up
1:47:55 you're going to be able to just quickly
1:47:57 go oh so fix that and it will
1:48:00 um or or anything it doesn't have to be
1:48:02 coding like one of the things that you
1:48:04 can do with these tools is you can kind
1:48:05 of reinvent yourself there's in the AI
1:48:08 salon
1:48:10 there's a guy Roger that was a he's a
1:48:12 retired software engineer
1:48:14 and the the salon originally started out
1:48:17 as the AI artist salon and you know and
1:48:19 he was like I'm not an artist and he was
1:48:21 sort of curmudgeoning about it and then
1:48:23 he ended up with his engineering
1:48:26 mind and his you know developer mind he
1:48:29 got really good at prompting mid-journey
1:48:31 and so he started making amazing art and
1:48:34 now he's like now he's become an artist
1:48:36 right using these tools
1:48:39 um so so be open to wherever it might
1:48:42 lead you but just start getting
1:48:43 articulate with it or uh literate with
1:48:46 it and and go that's that's awesome but
1:48:48 sorry to hear you're laid off that sucks
1:48:50 that sucks McKenzie has introduced an
1:48:53 internal AI chatbot that is the totality
1:48:56 of their knowledge and data yep I think
1:48:58 every company is going to have this it's
1:49:00 going to be quite easy to do
1:49:03 gonna be quite easy to do
1:49:05 so Kyle can you enable SS
1:49:11 I don't know what FF is
1:49:13 but tell me what it is and I'll sure
1:49:15 maybe
1:49:17 single single sign-on
1:49:19 I don't know what you're talking about
1:49:20 no e-commerce for Corporate America
1:49:24 don't know what that means
1:49:26 front end guy
1:49:28 I found an article on it
1:49:30 uh uh
1:49:35 felonious MK now's your chance to Pivot
1:49:37 to helping Services love that some of us
1:49:40 are wondering what Bots are
1:49:42 um okay so I bought
1:49:46 um
1:49:49 so you know in slack when people type
1:49:53 [ __ ]
1:49:56 you could you could create a bot a bot
1:49:59 is basically just a script like a
1:50:01 computer it's computer code
1:50:03 you could
1:50:07 you could write a bot that anytime
1:50:09 someone
1:50:10 asks a question in a certain Channel
1:50:13 it sends that question to open ai's API
1:50:16 and sends you back the answer from
1:50:18 openai
1:50:20 so bot is just kind of a service call to
1:50:22 an API that can that can do some things
1:50:25 so
1:50:27 you could create you could create a bot
1:50:29 if you're doing the teams thing where
1:50:32 um
1:50:35 randomly three times an hour
1:50:39 your Bot
1:50:41 could ask a question of one of your
1:50:43 colleagues
1:50:45 and and you could make it just ask a
1:50:47 name questions like hey do you know
1:50:49 where the report is from last Friday
1:50:51 so it seems like you're working and it
1:50:54 seems like you're acting or you know
1:50:56 you're acting like you're working and
1:50:57 you could be off you know having chicken
1:50:59 nuggets
1:51:01 [Laughter]
1:51:05 um but they can do more sophisticated
1:51:07 things you could have a bot that
1:51:12 every time a new article is written on a
1:51:15 particular RSS feed you summarize that
1:51:18 article and paste the summary into a
1:51:20 slack Channel or paste the summary into
1:51:24 an Excel worksheet or or whatever it
1:51:27 might be so so just think of them like
1:51:29 automations they're they're automations
1:51:31 where something happens it triggers the
1:51:33 bot the bot goes and executes code and
1:51:36 comes back and delivers something
1:51:41 a Bots like an Excel macro exactly
1:51:44 simply recording of things for the
1:51:46 computer to do but Auto yep that's it
1:51:47 and and Firefly zapier is a tool that
1:51:51 allows you to create automation so
1:51:53 they've got what's called a trigger so
1:51:54 the trigger might be someone types
1:51:56 something into a Google form
1:51:58 and then I'm going to take the input of
1:52:00 that Google form put it in a prompt send
1:52:02 it to open AI it's going to generate an
1:52:03 image then I'm going to send another
1:52:05 prompt and it's going to generate a
1:52:07 paragraph
1:52:08 and then I'm going to stick that in a I
1:52:11 don't know Word document and then my
1:52:14 social media person is gonna
1:52:16 check that document once a day for new
1:52:18 ideas
1:52:21 so zapier can do all of that the bot
1:52:24 won't do any good if they monitor the
1:52:26 screen
1:52:29 uh yeah but you'll be a there was a
1:52:32 there was a dude that did this without
1:52:34 um
1:52:36 without AI where he just he just
1:52:38 recorded like a loop of
1:52:40 himself for like 30 SEC no it was longer
1:52:44 than that like 30 minutes like he just
1:52:47 recorded like 30 minutes like
1:52:51 and then he just played that on a loop
1:52:53 in the zoom calls and I think it took
1:52:56 people three weeks to figure out he
1:52:57 wasn't actually on the calls
1:52:59 he was actually watching the zoom calls
1:53:01 but just it was his bot that was on
1:53:03 camera not him
1:53:07 so yeah you'll be able to do that
1:53:09 someone will come up with that
1:53:11 I've used chat CPT to write VBA code yep
1:53:15 I use chat EBT to write python programs
1:53:18 and then run them on Android using
1:53:20 pyroid 3 super cool I then convert them
1:53:22 into Android apps yep super groovy
1:53:27 um no experience in Python but a later
1:53:30 the Chachi BT can generate code and
1:53:32 teach me yeah it really is good at that
1:53:34 you can you can point it at
1:53:36 documentation and have it um
1:53:39 like like using the uh the web pilot
1:53:42 plugin
1:53:43 point it to the documentation of
1:53:45 something that you want to learn and
1:53:47 then have it come up with a custom
1:53:49 lesson for you
1:53:50 and and do it interactively where it it
1:53:54 teaches you something and tell it not to
1:53:55 move on to the next step until you've
1:53:58 completed the first step and you tell it
1:53:59 that and it'll just create a tutorial
1:54:02 for you and it'll just walk you through
1:54:04 it and if you get stuck on something
1:54:05 you're like I'm stuck here
1:54:07 and they'll say oh well you need to
1:54:08 install this Library I don't know how to
1:54:10 install that Library oh you do this I
1:54:12 don't know how to do that well then you
1:54:13 do it like this right
1:54:16 so great way to it's it's a great way to
1:54:18 learn just what these things are capable
1:54:20 of is just pick a project like that
1:54:23 it generates working code it's great yep
1:54:26 yep ion China
1:54:30 I'm gonna dive in MK I was on the
1:54:33 co-pilot beta but had no idea how big
1:54:36 the ecosystem got now yeah it's
1:54:41 it's it's it's big and you know MK one
1:54:45 of the things that I think would
1:54:47 probably be worth you checking out is
1:54:50 um
1:54:51 just go to GitHub or go to go to uh
1:54:53 Reddit and just dig into the whole llama
1:54:56 2 Community because the the whole open
1:54:59 source Community with these large
1:55:00 language models llama2 just came out two
1:55:03 three weeks ago
1:55:05 which is meta's you know base large
1:55:07 language model they've got a 7 a 13 and
1:55:10 a 70 billion parameter model but they're
1:55:13 licensed for commercial use so so
1:55:15 because of that the developer Community
1:55:18 has flocked to that so a lot of the top
1:55:20 trending I don't know what's going on my
1:55:22 throat a lot of the top trending
1:55:24 projects on GitHub are llama 2 projects
1:55:28 so that's worth checking out
1:55:31 Mr Kyle Shannon you're home I am indeed
1:55:34 I am home to General
1:55:37 I recommend Motorola or OnePlus as
1:55:39 affordable Android phone obtuse even
1:55:43 I don't know what that means
1:55:48 someone give me a nice compliment thank
1:55:50 you very much what website is that I
1:55:52 don't know where I was
1:55:54 [Laughter]
1:55:57 all right let me I'm going to scroll to
1:55:59 the bottom here I'm jumping a bunch of
1:56:00 questions but I just want to pop to the
1:56:02 bottom I gotta I gotta take off soon
1:56:03 it's getting late and my my throat is
1:56:06 season up something's going something's
1:56:08 going down I don't know what it is man
1:56:11 hope I didn't catch the crew
1:56:16 all right
1:56:22 SS screenshot oh screenshot
1:56:25 I can't do it either ever since he got
1:56:28 out of tick-tock jail
1:56:30 that's weird wait can I screenshot my
1:56:33 own thing
1:56:34 I can screenshot my own thing
1:56:38 you can't screenshot what do you want to
1:56:40 buy us are you on an iPhone
1:56:43 that's weird
1:56:45 what your reply to explain
1:56:49 of what a bot is is amazing oh great
1:56:51 thank you Firefly I'm glad that helped
1:56:54 I need to know how to code and if so
1:56:56 what language would be helpful python
1:56:58 seems to be the thing of the day and
1:57:01 um G uh code interpreter at chat GPT is
1:57:05 writing and executing python
1:57:08 um
1:57:09 Crossland David I don't necessarily know
1:57:12 that you need to learn how to code
1:57:14 these things are going to do the coding
1:57:16 they're not quite there yet but they're
1:57:18 close so if you want to learn how to
1:57:20 code I'd learn how to code in
1:57:22 conjunction with these tools so go to
1:57:24 GitHub use copilot it's a pair
1:57:27 programming tool
1:57:29 from openai that's designed specifically
1:57:32 to work with a coder
1:57:35 um Firefly is that like your zapier demo
1:57:38 you did last week yep that was that was
1:57:40 exactly what it was that was that
1:57:42 zapier's kind of Bot making the bot just
1:57:45 basically means you're sort of logging
1:57:47 into something like Discord or slack and
1:57:50 interacting is as if you were a person
1:57:55 um just tell it you want to Learn Python
1:57:57 exactly so do all AI tools
1:58:01 make Bots
1:58:05 no I I mean
1:58:08 think of bots as just you know they're
1:58:11 they're just little automations you you
1:58:13 can here's here's what's the big
1:58:15 difference historically if you'd make a
1:58:17 bot
1:58:19 you would you would say Okay I want to
1:58:21 do some sort of
1:58:24 categorization bot and you would say
1:58:26 when
1:58:28 someone writes an article
1:58:30 um I'm gonna I'm gonna pull the metadata
1:58:33 from that article
1:58:34 that's got a category metadata in it and
1:58:37 then I'm going to put the link of the
1:58:39 article the title of the article and the
1:58:41 category of the article in a spreadsheet
1:58:44 and so you would do very kind of manual
1:58:47 logical connections right you know bot
1:58:50 making and automations was very much
1:58:52 like Lego blocks and logic trees and
1:58:54 things like that the difference now with
1:58:56 large language models with the AI is
1:58:59 that you can take unstructured data or
1:59:02 inconsistently structured data and just
1:59:05 ask it [ __ ] and it will do it for you
1:59:08 right so you can just say go read that
1:59:10 article and write me a summary or go
1:59:12 read that article and write me five
1:59:14 tweets or a tweet Thread about that
1:59:17 article
1:59:18 and then you know let's post that tweet
1:59:21 thread to Twitter
1:59:24 um that wasn't possible before I mean
1:59:26 you could you could do it with with gpt3
1:59:29 for a while but but the popularity of
1:59:32 these things the the large language
1:59:33 models make what you can do with those
1:59:36 Bots much more sophisticated with
1:59:39 essentially You Know This Much
1:59:40 programming you're you're just sort of
1:59:44 you go grab a pile of data
1:59:47 you throw it over to one of these large
1:59:49 language models and if your prompt is
1:59:50 clever enough it will throw you back
1:59:53 what you need it might throw you back
1:59:56 structured Json data that goes into your
1:59:59 tool that you know writes a blog post
2:00:02 for your company website and it does
2:00:04 that twice a day automatically so
2:00:07 um and you can do that without a ton of
2:00:08 programming so so the difference is the
2:00:10 sophistication of the data processing
2:00:12 that these tools make possible
2:00:14 that's that's why it's such a big deal
2:00:17 all right 126 people on board this is
2:00:21 crazy it's Wednesday night it's late you
2:00:24 people should be in bed what are you
2:00:26 doing I got to be in bed my throat's
2:00:27 going
2:00:29 all right I'm gonna go I'm gonna go oh
2:00:31 my goodness the education system needs
2:00:33 yeah especially with teacher shortages
2:00:35 and post pandemic I think these things
2:00:36 are going to be if you're interested in
2:00:38 the education system Firefly go check
2:00:40 out
2:00:41 um Khan Academy which you probably know
2:00:43 about it's been around for years the
2:00:45 flip the classroom thing
2:00:46 three four months ago they released this
2:00:48 thing called conmigo okay
2:00:51 h-a-n or k-a-h-n
2:00:55 um
2:00:57 conmigo as in Amigo and it's a gpt4
2:01:02 teacher teaching assistant that will
2:01:04 teach you a student how to learn
2:01:06 something it's really amazing worth
2:01:09 worth checking out
2:01:10 videos on YouTube Real Money strategies
2:01:12 about chat EBT and making money not
2:01:14 showing your face yes or no
2:01:18 that's a joke right
2:01:22 [Laughter]
2:01:25 okay
2:01:27 okay here's the deal
2:01:30 anyone charging you money to teach you
2:01:32 how to make money
2:01:33 or making money on the teaching not on
2:01:36 the chat GPT now they've probably used
2:01:38 chat CPT to write the course
2:01:41 um
2:01:43 you could start a YouTube channel
2:01:46 doing you know text only animations with
2:01:50 music and you could completely automate
2:01:53 that you could do that right now in
2:01:55 Google Sheets with GPT uh for uh gbt for
2:02:00 sheets and you could you could
2:02:02 automatically generate a giant
2:02:03 spreadsheet of copy lines and calls to
2:02:07 action and URLs and pop that into canva
2:02:11 bulk create and create a bunch of videos
2:02:12 so you could absolutely do that
2:02:15 um
2:02:16 here's the challenge so can everyone
2:02:19 else
2:02:20 so it comes back to do you have a good
2:02:23 idea if you can come up with an angle
2:02:26 that people would actually want to pay
2:02:28 attention to that stuff great but just
2:02:30 know you're increasingly competing with
2:02:33 companies and organizations that are
2:02:35 going to be publishing 10 000 videos a
2:02:37 day doing the exact same thing so
2:02:41 what I would say Kathy is
2:02:43 the experience of learning how to do
2:02:45 that with chat GPT and doing it and
2:02:48 putting it out there if you put it in
2:02:50 that context not like I want to make
2:02:52 money but like I want to learn how to do
2:02:54 this then I think it's interesting but
2:02:56 again I think that what what they're
2:02:58 saying what their approach is is
2:03:02 if you automate video production you'll
2:03:05 be rich and that's wrong
2:03:09 that's like saying if you know how to
2:03:10 use Photoshop you're a designer
2:03:13 no
2:03:15 no
2:03:16 photoshop's the tool
2:03:18 you got to go to the other side and say
2:03:20 what are you doing with that automated
2:03:22 video production
2:03:24 okay I know that cat people that like to
2:03:28 crochet
2:03:30 love inspirational poems about yarn and
2:03:34 kitties I'm gonna do a YouTube channel
2:03:36 About Yarn and kitties great okay
2:03:38 perfect
2:03:40 there's your idea
2:03:42 now which tool are you going to use are
2:03:44 you going to go film a bunch of kitties
2:03:46 probably not but could you use
2:03:48 mid-journey to make a bunch of images of
2:03:50 kitties with balls of yarn yeah you
2:03:52 could do that and then could you use
2:03:54 chat to write all sorts of clever little
2:03:56 headlines and then could you use canva
2:03:59 bulk create to assemble them into videos
2:04:00 yeah you can do all that
2:04:04 so so don't mistake the tool for the
2:04:07 strategy the strategy is where is how
2:04:09 you make your money
2:04:11 solving the problem is how you make your
2:04:13 money the tool is just the tool
2:04:15 so yeah that ranch dressing gotcha yeah
2:04:19 exactly it's blue cheese by the way from
2:04:22 Fire on the Mountain and it's delicious
2:04:24 you can't leave now
2:04:27 yeah well actually you said you can't
2:04:29 leave now
2:04:31 it appears that Apple phones are not
2:04:32 able to screenshot and Androids are oh
2:04:36 Apple phones are able to screenshot and
2:04:38 Androids are not that's weird
2:04:41 I use code interpreter with unity 3D I
2:04:44 bet that's amazing so Unity 3D is like a
2:04:47 you know game development environment
2:04:48 you can take code from an existing game
2:04:51 in unity pop it into chat EBT and just
2:04:55 tell it to change all the shapes to
2:04:56 spheres instead of cubes and it'll do it
2:05:01 crazy ass [ __ ] all right I'm gonna go
2:05:04 I'm winding this down I'm winding this
2:05:06 down people I'm doing it I still have a
2:05:08 final paper to write I still have a
2:05:10 bunch of [ __ ] to do it's 12 a.m on the
2:05:11 east coast thank you for hanging out
2:05:13 he show you how to no cost wait he show
2:05:17 you how to know cost
2:05:19 I don't know what that means but I don't
2:05:21 charge
2:05:23 uh I did front end payment for ClickBank
2:05:25 oh interesting that's MK
2:05:29 um
2:05:30 halfway to her
2:05:32 the average person no no Androids can
2:05:36 screenshot
2:05:37 all right so tomorrow all the Irregulars
2:05:41 we're gonna have to figure out what the
2:05:42 screenshot issue is and solve it for
2:05:44 people
2:05:45 because it would be nice if you could
2:05:46 screenshot
2:05:48 I watched The Tick Tock with scuba
2:05:50 divers they hardwire
2:05:51 the Wi-Fi link underwater Oh that's
2:05:54 interesting wow that's interesting
2:05:56 huh that's cool at least I don't have to
2:05:59 do that for the average person in the
2:06:02 office wait for the average person in
2:06:04 the office is to learn it is to learn
2:06:07 about AI
2:06:09 I don't know what you mean there
2:06:12 like what should they learn
2:06:14 for the average person in the office
2:06:15 right now I would say just like personal
2:06:17 productivity [ __ ]
2:06:20 use chat EPT to write emails use chatgpt
2:06:22 to summarize
2:06:24 call notes like
2:06:26 when you do a zoom call or a Google meet
2:06:29 turn on recording and transcriptions and
2:06:31 then take that transcription and pop it
2:06:33 into chat GPT or Claude because Claude
2:06:36 can take bigger documents
2:06:38 and have it summarize your call notes
2:06:40 and create a to-do list for every person
2:06:42 that was on the call
2:06:43 stuff like that just just really start
2:06:46 using these tools to to
2:06:49 extend what you already do make what you
2:06:52 do
2:06:53 better faster smarter than you could
2:06:56 have done it
2:06:57 without it and just start with small
2:06:59 stuff you don't you don't have to like
2:07:01 you don't have to start programming and
2:07:03 doing all this [ __ ] like that's the
2:07:05 whole point of this channel is
2:07:07 we're talking about chat GPT which is
2:07:09 sort of post
2:07:11 when the Geeks when only the Geeks could
2:07:13 play right chat gbt is the first tool
2:07:16 that makes these tools accessible to
2:07:17 real people
2:07:19 so if you're in an office just start
2:07:20 playing
2:07:22 and if you're one of the people in the
2:07:23 office that's good at this don't keep it
2:07:25 secret talk about it
2:07:27 say to people oh I did this with chat
2:07:29 EBT look what I did and just be careful
2:07:32 with you data you gotta if you gotta
2:07:33 anonymize your data if you want to be
2:07:35 safe in a corporate environment but
2:07:37 anyway listen this is fantastic thanks
2:07:39 for hanging out with me everybody I'm
2:07:41 gonna I'm gonna take off I'm gonna go
2:07:44 peace out peace out peace out
2:07:48 thanks everyone
2:07:50 I'll see you tomorrow