Dec 28, 2023 - Creating Impactful GPTs for Nonprofits in 24 Hours!

Description

Chapters

Transcript

0:01 [Music]
0:16 [Music]
0:25 it's not the long flowing dress that
0:27 you're
0:29 in
0:31 coming off of your
0:33 skin not the way you what are you doing
0:38 buddy hello hello
0:41 hello welcome Thursday night learning
0:44 [Music]
0:59 lab
1:00 [Music]
1:09 he's just more annoyed tonight he
1:11 doesn't want to sing he's like I'm not
1:13 just your trained monkey you can't just
1:15 play a guitar and expect me to sing
1:17 melodious
1:18 tones right right dog right A
1:24 minor oh a minor I'll get you every
1:26 [Music]
1:29 time
1:32 in a Westerly
1:35 Direction this car is my
1:40 train I've been driving I've been
1:44 wondering what it is I'm running from
1:49 again feel like an 80-year-old
1:53 man holding on to
1:56 [Music]
1:58 29 a on there
2:01 [Music]
2:02 horiz is California
2:06 [Music]
2:12 L up ahead of trucks carrying a wide
2:16 load brief Fab house cut in
2:20 half cute little front door and two
2:23 windows my
2:24 love ain't sure whether the crowd should
2:29 laugh you see I broke home up myself
2:32 [Music]
2:33 once as I stum to that
2:37 door I read her note by the dawn
2:42 light said don't you come around
2:45 [Music]
2:46 here
2:49 [Music]
2:53 anymore this freedom of the
2:57 RO never was good with decision
3:00 [Music]
3:02 Le that's what I've been
3:04 [Music]
3:11 told
3:14 yeah thank you Sherry D nice hoodie yep
3:16 that's the AI Salon one year anniversary
3:20 hoodie I'm excited by it it's
3:24 nice it's a nice nice print a very nice
3:28 print it looks lovely
3:32 lovely great name for a band melodious
3:35 tones I like it I like it champy you
3:40 want to join melodious tones I think
3:42 you're the lead singer I'm just I'm just
3:44 playing backup to you
3:46 [Laughter]
3:50 dude all right I'm feeling better my
3:53 eyes look a bit more clear still a
3:55 little stuffy and you know we're going
3:56 to we're going to have to deal with that
3:57 let me take off the hoodie cuz I'm going
3:59 to get too hot and bothered you don't
4:01 want me hot and bothered right
4:08 champy all
4:12 right a very nice very nice we got the
4:16 little GPT for good
4:18 sheep Danielle did that be Not Afraid
4:21 did that I think it's a cute little
4:23 thing I posted it Go Kyle love your
4:26 voice thank you Cam kaken I appreciate
4:28 that every now and again if you know
4:30 what's funny about singing is if in your
4:33 mind you think I can't sing you sing
4:37 shitty and if your mind you just go oh
4:40 [ __ ] it then you're fine it's it's the
4:42 weirdest thing and it's like on a note
4:45 by note basis like if the doubt Creeps
4:47 in as you're about to hit a note it
4:50 goes you just missed the note it's just
4:53 the weirdest
4:57 thing you got to you definitely have to
4:59 get in the zone was singing um the white
5:02 hoodie doesn't make it easy to comment
5:03 yeah exactly does it precious exactly oh
5:07 man welcome to the AI learning lab let
5:10 me check in here for a second get my get
5:13 myself all Rec
5:16 combobulated um so two days from now
5:19 Saturday December 30th starting at 6:00
5:21 a.m. Pacific Standard Time an Murphy and
5:25 I are doing this thing called GPT for
5:28 good it's a 24hour
5:31 because I know how to good M make good
5:33 life decisions it's a
5:36 24-hour
5:38 um session where she and I are going to
5:40 alternate every three hours each of us
5:43 are going to be teaching how to make
5:45 gpts for nonprofits or the cause of your
5:48 choice or you can just come make a GPT
5:50 it doesn't matter if you make it for a
5:51 nonprofit but if you do make it for a
5:53 nonprofit We want you to share it with
5:55 the hash GPT forg good and uh yeah I'm
5:59 excited about about that so that happen
6:01 Saturday starting at 6:00 a.m. s dear
6:04 Kyle or hey Kyle hope you're feeling
6:06 better I am feeling much better I
6:09 definitely what was it Wednesday what's
6:11 today Thursday so Tuesday I went live
6:15 against doctor's orders there was no
6:17 doctor's orders I took Christmas off and
6:21 then uh Wednesday I was like the
6:23 showuser Tuesday I was like the show
6:24 must go on and that was probably not a
6:26 good call because I was still feeling
6:27 pretty
6:28 shitty
6:30 and then last night I was a little
6:31 better I'm drinking hot uh this is for
6:34 those of you into the tea scene this is
6:38 elderberry and
6:40 eonia and Eucalyptus and mint so I did I
6:45 did it's it's a it's a it's just a
6:47 hell's Brew but you know it's
6:50 good all right fantastic I had an
6:54 artificial intelligence Supra dinner
6:56 tonight thanks to chat GPT well I hope
6:59 you use real
7:02 ingredients that's awesome um yeah it's
7:05 crazy how it can just whip up a recipe
7:07 for you isn't it and it's I've done
7:10 those before I did it I took a bunch of
7:12 drink recipes to a a whiskey club night
7:15 the F the first night I joined this
7:17 whiskey club and I took some chat GPT
7:20 drink recipes and two of them were
7:22 really good and one of them was just
7:23 awful and I was like yeah yeah it's just
7:27 making [ __ ]
7:28 up
7:31 but you know you you figure it's got to
7:33 be based on you know someone put some
7:35 drink recipes somewhere on the internet
7:37 and that's you know they got Amalgamated
7:39 into your little your little bash so
7:43 that was fun anyway welcome uh my name
7:45 is Kyle Shannon this is the AI learning
7:47 lab we talk about all things AI here
7:49 what I figure I'll do tonight is I'm
7:50 going to start working on gpts and I'm
7:53 I'm going to start talking about what
7:54 we're going to do on Saturday so if
7:57 you've got
7:58 gp4 um feel free to build along tonight
8:02 um what we're doing on Saturday is um is
8:06 you don't need to have any expertise no
8:08 technical knowledge you do need to have
8:10 a chat gp4 subscription so if you don't
8:12 have that get one um if you want to join
8:16 and then if you know anyone in the
8:17 nonprofit space or anyone who's trying
8:20 to do good good in their Community let
8:23 them know about this event so let me let
8:25 me let me share with you shall I
8:30 but K how do I find this swell
8:37 [Music]
8:45 event all right so that's the event I'm
8:47 putting the AI Salon which is a
8:49 community I started about a year ago is
8:51 doing this in conjunction with owered
8:54 fundraiser and Murphy's
8:56 organization um she really knows about
8:58 this nonprofit stuff I just no
9:00 tangential I've done a lot of work in
9:01 nonprofit I've been on boards things
9:02 like that um but I I I don't know the
9:05 world that well she really knows the
9:06 world well and so she and I are going to
9:09 be alternating every three hours I'm
9:11 doing I'm doing the nines and the threes
9:13 she's doing the sixes and the
9:15 twelves all Pacific time and uh and and
9:20 yeah and I'll it'll just be on this
9:21 channel here for me and then she's got
9:23 her own Zoom Channel and so how you get
9:25 to the to the info is go to the salon .
9:29 a go to this
9:31 URL and this is the URL to the AI Salon
9:35 um the the first two links are basically
9:37 to the same place but the second link is
9:39 the link specifically to this event and
9:41 so it'll take you over
9:43 to the AI Salon
9:47 Channel and it will drop you on this
9:51 page the GPT for good
9:55 challenge actually I'm going to edit
9:57 this I'm going to add something here
10:03 24
10:05 hour GPT for good challenge can I do a
10:08 shift return there no I
10:15 can't continue with that message there
10:18 we go 24hour GPT for good
10:22 challenge and so you can RSVP here we've
10:25 got 25 RSV VPS I assume we'll have and
10:28 we're also putting the in lots of other
10:30 places so you can just show up you don't
10:32 have to RSVP though it would be nice to
10:34 know if you're
10:35 coming um we talk about why we're doing
10:38 it what it's about and then the event
10:40 snapshot so so here's info on Ann and I
10:44 and then here is the schedule so she
10:47 does the first session at 6:00 a.m.
10:50 Saturday morning she volunteered for
10:53 that slot I was going to start at 900
10:55 but she was like no no I'll take six I'm
10:57 like you go for it but then I'm going to
11:00 have to you know show up and at least
11:01 say hi for the opening event don't you
11:03 think so so we're starting bright and
11:06 early that's Pacific time so it's uh
11:10 9:00 a.m. eastern
11:12 time and then at 9:00 a.m. Pacific I go
11:15 here on Tik Tok and then we basically
11:17 just do that for 24
11:19 hours because again we know how to make
11:22 good life decisions Isn't that cool um
11:24 and then what we're going to do is you
11:26 just show up for the thing if you have
11:28 no experience with chat GPT other than
11:31 you bought a Plus subscription and
11:33 you've been just messing around with it
11:34 but you've never made a GPT we're going
11:36 to make a GPT that's going to be like
11:40 basically like an educational GPT for
11:43 your nonprofit or your cause of choice
11:46 um that will answer questions about it
11:48 about the the organization about
11:50 volunteering about all that sort of
11:52 stuff based on the information you have
11:55 on the website for the organization
11:56 which I think is is super cool so we'll
11:58 teach you how to how to do that build
12:00 that um we'll give you a couple little
12:03 Advanced options and and basically
12:06 within within an hour you'll have built
12:09 a fully functional custom GPT that you
12:12 can use for your organization or someone
12:14 else's organization um and it it'll even
12:17 do a call to action to the Donate page
12:20 that's the thing I'm actually really
12:21 excited about all right so that's what
12:24 that's going to be that's very helpful
12:25 for nonprofit organizations yeah thanks
12:26 Jesse from Austin I'm excited about it
12:30 it was a so so what this came out of was
12:34 um not that long ago how long ago was it
12:38 I don't know um oh I know when it was
12:41 November 30th I decided to do a 24-hour
12:45 chaty PT Aon in celebration of the um
12:50 oneyear anniversary of chaty PT and then
12:53 someone said hey Kyle like this was a
12:55 week ago Thursday someone said hey Kyle
12:57 you should do another 24-hour thing
12:59 thing and I'm like you're a dumb Dum and
13:01 then I was like wait a minute we've got
13:03 the week off in between Christmas and
13:05 New Year why don't we do this we'll do a
13:07 24-hour thing maybe I'll team up with an
13:09 Murphy I just said it out loud and then
13:11 I don't either she was on the call or
13:13 something like that but she found out
13:15 about she's like I'm in let's do this um
13:18 and so we decided to do this and then
13:20 the next day I was diagnosed with covid
13:23 so so but I think I think I'm on the
13:26 mend enough that I think I'll be fine
13:28 for Saturday
13:29 but you know you know no good deed goes
13:32 unpunished in this
13:34 world who has the week off between
13:36 Christmas and New Year apparently all of
13:38 the pharmaceutical
13:41 industry is this like Jerry Lewis ton
13:44 without the big band that dancing bears
13:46 and cheesy cheesy magicians it's exactly
13:48 like that Spike Jones it's got It's got
13:50 all of the uh the uh comedic stylings of
13:54 Kyle Shannon uh you know on the on the
13:58 nines and the three
14:01 and then and then Ann Murphy's great
14:03 she's just a ball of energy she's really
14:05 good at this stuff so so and and the
14:08 other cool thing the thing I'm I'm kind
14:10 of excited about is I'm also encouraging
14:13 if you've built gpts before if you're
14:15 expert at this if you're like lunatick
14:17 who's made 50 of these things I want
14:20 everyone to come to these because you
14:22 can support new people that are trying
14:24 to figure this out and have questions
14:26 and so I'll be sort of going on and you
14:28 know doing my down the rabbit hole kind
14:30 of thing and so you can all support one
14:32 another and so even if even if you come
14:35 you know for one of the early ones make
14:37 a GPT and then come to the later ones
14:39 and improve on your GPT or build another
14:41 one like why not you know I want it to
14:43 be I want it to be a fun event so yeah
14:46 just like the
14:48 ton without the raising money and
14:50 without the uh yeah without the star
14:53 power although I did tag Sam Alman in a
14:56 tweet today and said Sam what do you
14:58 think he hasn't retweeted me yet cuz you
15:00 know he's jealous and all that happens
15:02 that what are you going to do all right
15:04 these 24 Mar hour marathons are killing
15:07 me four hours is more my speed I know so
15:10 Source Camp it's it's it's going to be
15:12 the same lesson repeated over and over
15:13 and over again you don't have to come to
15:15 all of
15:16 them I think I think you are
15:18 sufficiently up to speed on
15:21 gpts but feel free to share with anyone
15:24 you know and there's no excuse for
15:27 someone who's you know
15:29 in some other part of the world or like
15:31 I'm traveling whatever there's no excuse
15:33 every 3 hours for 24 hours there's going
15:35 to be one of these you can f someone can
15:37 fit it in their
15:40 schedule all right um let me talk some
15:44 more about the AI salon so this this
15:47 little this little rag tag bunch of
15:50 bunch of people here um so this
15:53 community I started when when I when I
15:56 learned about um
15:59 chat gbt I was I was fairly deep into
16:04 the into the the AI game doing
16:07 generative AI stuff with stable
16:09 diffusion and dream booth and things
16:11 like that and I had started an art
16:13 project and I was fairly deep in the
16:15 game um but it it it didn't it felt
16:19 interesting but not significant and then
16:22 chat GPT came out and I was like oh [ __ ]
16:24 this is this is really significant this
16:26 is bigger than the early of the web and
16:32 so I was fortunate enough to be in in
16:33 New York City in in the mid 90s when the
16:36 when the web came onto the scene I
16:38 co-founded agency.com which was one of
16:40 the first digital agencies but I also
16:41 started a group called the worldwide web
16:44 artist Consortium which was basically
16:46 just a Shameless attempt on my part to
16:48 surround myself with people that knew
16:50 more than I did about the worldwide web
16:52 and guess what I discovered nobody knew
16:55 [ __ ] anything nobody knew anything
16:57 everyone was trying to figure it out and
16:59 it it it was really about a 2 or threee
17:03 period
17:06 where there were no experts you couldn't
17:08 get a degree in it um you couldn't hire
17:10 someone out of college with those skills
17:12 everyone had to figure it out on their
17:14 own and so having a community like that
17:16 was incredibly powerful so when chat jpt
17:19 came out I like I immediately knew I got
17:21 to do this so we started this AI salon
17:24 so Leah faston who's a professional
17:26 photographer out of Boston she and I
17:27 were talking about this AI stuff and
17:29 when chat GPT came out I said let's
17:31 start this thing and she was like I'm in
17:33 let's go and so we started this the week
17:36 after chat GPT came out so we just
17:37 celebrated our one-year anniversary just
17:39 had my my oneye anniversary sweatshirt
17:41 on my hoodie
17:44 um and and that's what the community is
17:46 it's like a support group we just we so
17:49 we just recently two months ago we
17:50 launched this online community on Mighty
17:53 networks um and we just cracked 500
17:56 members on the online community which is
17:58 great CU we had I think we had 7 or 800
18:01 on Discord but you know a lot of people
18:03 join and discord's hard so a lot of
18:05 people didn't use it so we just hit the
18:07 500 Mark here if you if you come to the
18:10 AI Salon Mighty networks go to the
18:13 welcome to the salon page and we'll talk
18:15 about who we are why we do what we do
18:18 what we're trying to accomplish but then
18:20 Also it talks about our values things
18:21 like curiosity and generosity like if
18:25 you learn something share it don't hoard
18:27 it share your because this stuff's
18:29 moving so fast that even if it seems
18:32 like it's proprietary and really good
18:34 for you to hold on to it right now it's
18:35 going to be obsolete in two months
18:37 anyway so you might as well share it and
18:39 and what it's done is it's created a
18:40 really amazing Community here um that is
18:43 is super generous um exploration is one
18:46 of our values collaboration's one of our
18:48 values um inclusion empathy um empathy
18:52 is a big one for me because
18:55 um as exciting as this AI stuff is and
18:58 this channel is a testament to how
19:01 excited and optimistic I am about this
19:03 stuff but I also recognize thank you
19:05 Winston I appreciate that I also
19:08 recognize
19:11 that these Technologies are profound and
19:15 a lot of people are scared a lot of
19:17 people are angry most of the people that
19:19 are scared and angry have not used the
19:22 tools uh they're making a lot of
19:24 assumptions about how they work and how
19:25 evil they are and all that sort of [ __ ]
19:27 and so a big part of this is De
19:29 demystifying that but it's also
19:31 understanding that people are in that
19:32 place and so the whole idea of the salon
19:35 is we want people to be able to come
19:36 there and no matter where they are on
19:37 this journey I had someone last night on
19:39 the live said when they when they came
19:41 to the live they were V vehemently
19:43 against Ai and now they're totally into
19:46 it because they learned about it and
19:49 once they learned about it it sort of
19:51 brought down all of those preconceived
19:54 notions and and open things up so that's
19:57 that and then then um the other thing
20:00 that if you decide you want to join if
20:02 you if you resonate with those values
20:04 and with what we're about great then
20:06 jump into the introduce yourself section
20:08 and introduce yourself and be generous
20:11 use one of our values and tell us who
20:13 you are and what what you're about and
20:15 then there's water cooler discussions
20:17 there's a showand tell Channel which a
20:19 lot of what shows up GPT for good oh
20:21 that's really nice thank you Kate that's
20:23 really cool um a lot of what shows up in
20:25 Show and Tell here is based on what
20:27 we're talking about in this channel so
20:29 I'll make some stupid joke and then a
20:31 bunch of images of that stupid joke end
20:33 up end up in show and tell which is fun
20:36 that's a super cool in syndicon nice go
20:40 with the flow I like
20:43 it so anyway so that's that and then the
20:47 other thing that um is in the is in the
20:51 in the salon are guilds so we've got um
20:55 uh Peter kaminsky's got an AI 101 Guild
20:57 Leah my my co-founder has got an Art
21:00 Guild there's a writer Guild by Jane a
21:03 business Guild with uh with Ken and
21:05 Bradley Pat's got the AI mechanics Guild
21:09 um jrc's got the healthc care Guild uh
21:11 there's a video Guild by Scott and we're
21:13 just about to open the education Guild
21:16 with Matt and Stacy Mr K if you see him
21:18 here on on uh Whata call it
21:23 uh on the Tik Tok uh Mr K is going to be
21:26 heading up that with Stacy just me and
21:28 the trees been introducing fellow
21:30 cruising passengers about a AI lots of
21:33 Kevin mallister moments are you on that
21:36 9mon cruise Source Camp are you on the
21:39 nine month nightmare of from Hell Cruise
21:41 or did you just go for like a
21:44 week I think it's awesome you're
21:46 teaching people about AI on a cruise
21:48 ship it's like they can't
21:50 escape sit down Source cam no just a
21:56 week all right
21:57 good
21:59 that's awesome um Sal is one of those
22:03 from Fear to Fantastic converts I think
22:05 great yeah that's right I think it was
22:06 Sally I think you're right Amelia's wife
22:09 captive AUD yeah exactly exactly I'm
22:12 bored yeah I'm bored too do you want to
22:14 eat some more no you want to learn about
22:16 AI not really come on I'll teach you
22:18 anyway okay hey that's kind of oh my
22:22 [Laughter]
22:27 God I like it I fully support I I ful
22:31 fully support the AI
22:34 ambush in fact I encouraged it with with
22:37 people home with their families just
22:39 talk about
22:40 it but they're going to think I'm a
22:42 weirdo they already think you're a
22:43 weirdo you're already the black sheep
22:45 you might as well just lean into
22:50 it
22:51 oh all right s oh oh that's good yes s
22:56 yes I am and now I'm trying to convert
22:58 others that's the thing
23:01 about people who are vehemently against
23:03 something that then convert they're the
23:06 they're insufferable because because
23:08 they knew what it was
23:09 like to be in that sort of arms crossed
23:13 kind of defensive mode about
23:16 Ai and they they acutely s acutely
23:20 understands the difference between what
23:22 he thought it was and what it actually
23:25 is and and honest to once you get what
23:29 it what it really is like it's my the
23:32 the only way I can explain learning more
23:35 and more about AI is just it's it's this
23:38 sort of parting of of the of the clouds
23:42 right and just there's this Clarity that
23:44 opens up and this
23:47 possibility um that just seems to get
23:49 bigger and bigger and bigger um it's
23:51 wild it's wild so yeah all right Joker
23:57 actually I have four platforms and
23:58 models running locally oh Joker You're
24:00 just showing off
24:02 now I don't I've got I've got a couple
24:05 of models that I installed and ran once
24:08 and then I'm like yeah I'm just going to
24:09 keep using open AI cuz it's like it's
24:13 the best if I had stuff I was doing that
24:16 I really wanted a local version or if I
24:18 you know but I just kind of need to know
24:21 how's this [ __ ] working and it's it's
24:24 good enough to use the commercial [ __ ]
24:25 right
24:27 now
24:28 that said I think the I think the open
24:30 source movement is critically important
24:33 and I think what what the open source
24:36 Community is going to do is they're
24:37 going to drive better and better models
24:39 smaller and smaller and more and more
24:41 efficient so as Hardware continues to
24:44 follow Moore's
24:46 Law these large language models are
24:49 going to sort of catch up and we're
24:51 going to be able to
24:53 install very high quality large language
24:56 models that run locally on a device and
24:59 what that means is you don't need an
25:01 internet connection like you could have
25:03 the full power of at least GPT 3.5 maybe
25:07 even GPT 4 on your phone without needing
25:11 an internet
25:13 connection um and like when I think
25:17 about how powerful that is because if if
25:19 you think about what chat GPT really is
25:22 you take all of the knowledge that's
25:25 been dumped onto the internet in the
25:27 past 40
25:29 years it's a
25:31 lot and all of these models have been
25:34 trained
25:35 on most of the [ __ ] that's out there in
25:38 the public internet right there's lots
25:40 of other data sources that aren't in it
25:42 but the a lot of it's a lot of the
25:43 public internet is
25:45 captured and so
25:48 imagine being able to have instant
25:51 access to all of the knowledge of
25:52 everything that's been dumped on the
25:54 internet without needing an internet
25:55 connection so imagine third world
25:58 countries and places that don't have
26:00 access and maybe they do have mobile
26:02 phones but they don't always have
26:04 connectivity things like you know
26:06 medical emergencies and diagnosis and
26:08 things like that and as these models get
26:10 stronger and stronger with multimodal
26:12 capabilities the ability to take a
26:14 picture of a wound or or you know a
26:18 broken bone or whatever it might be you
26:20 know an eye infection and have that
26:23 immediately diagnosed with all of this
26:25 stuff I mean it's crazy
26:29 tanker a Kush hello Kyle hope all is
26:31 well in in your world it is improving in
26:33 my world we had four out of four Shannon
26:35 family members well minus champ champ
26:38 didn't test positive for covid
26:41 yet he seem he seems to have skirted it
26:44 but but the four adults in the house all
26:46 got it and and the others other than me
26:50 are immuno compromised so it was pretty
26:53 serious
26:55 deal but got them all unpacked PID and
26:59 uh or Pax loid I don't know how you say
27:01 that and they they seem to be on the
27:03 mend
27:04 so all's good and I'm on the menend i'
27:07 i' still got a scratchy voice I'm still
27:08 a little snotty but like it's it's
27:10 mostly gone I felt like I woke up this
27:12 morning I actually cleaned my little
27:14 infirmary area
27:16 vacuumed just it's not like me uh so I
27:20 had a lot of energy and then uh I did
27:23 like 90 minutes of intense I did the
27:25 dishes I did vacuum I'm like okay back
27:28 this is good and then my body was like
27:30 hey [ __ ] go back to bed I was like oh
27:33 yeah I was
27:35 sick so oh man the one I got running
27:40 last night just kept saying I need your
27:42 help how much money is involved yeah
27:46 that's that's not a large language model
27:47 joker I'm pretty sure that's a different
27:49 kind of application you installed there
27:51 buddy I can help you with that I think
27:53 you installed this in your computer you
27:55 can make money
27:57 with for
27:59 1995 just 1995 and four easy payments I
28:02 can teach you how to make money
28:04 with so very different application than
28:07 chat gbt
28:09 buddy have I taught you nothing
28:16 apparently oh man all right let's see um
28:21 my wife is still covering from Co
28:23 herself hey Mr K what's happening uh I
28:25 am excited to get the education Guild up
28:28 and running I'm really really [ __ ]
28:29 stoked about that I hope your wife feels
28:33 better any insight or comment on the New
28:36 York Times case against Microsoft and
28:38 open AI
28:43 um it's an important one I I doubt very
28:49 seriously it will hit
28:52 um it will hit the courts right I think
28:56 they'll settle before then
28:58 um it's it's in no one's interest
29:01 including the New York's times to shut
29:04 down arguably the most powerful most
29:08 popular piece of software in history
29:12 um they just want their cut of the
29:14 action they did a they cut a deal with
29:16 Google $100 million with Google in May
29:21 which as someone pointed out I think Ted
29:23 pointed out someone pointed out was not
29:25 really an AI deal but it was it was it
29:29 was it was a big deal with
29:31 Google
29:34 um so it's interesting to me that they
29:37 didn't sue Google right they've got aund
29:40 million deal with Google so they didn't
29:42 sue them so they're just you know
29:43 they're suing and and
29:47 um shortly
29:50 before um Alman was
29:54 fired you know maybe like the month
29:56 before they were fired they were
29:58 apparently in negotiations with the
29:59 times to do
30:01 something and those talks apparently
30:03 fell apart and so like
30:07 it I understand why they're doing it I
30:11 think
30:13 that I think that the data scientists
30:16 and
30:16 the lawyers arguing for the good of
30:20 humanity are going to be able to
30:22 completely
30:24 dismantle the times lawsuit but
30:29 uh you never know like if they end up
30:31 not settling and if they end up taking
30:33 it to court um it could really [ __ ] up
30:38 the industry I just I don't think
30:39 anyone's going to let that happen like I
30:41 just you know you know there's too much
30:43 there too much money involved the
30:45 there's too many big players involved to
30:47 let it you know completely just tank an
30:51 industry
30:53 um but that said it's it's an important
30:56 case right and it it will
30:58 this will establish some boundaries
31:00 about what the rules of the game are and
31:03 like whether they settle or go to court
31:04 this is going to establish some
31:05 boundaries I just doubt very seriously
31:08 they're going to they're going to flush
31:09 it all down the toilet
31:11 um yeah it just it doesn't make any
31:13 sense for anyone involved including the
31:15 times right cu the times is going to
31:18 want to
31:20 use they're going they're going to
31:24 want their
31:26 content to to be part of the AI fabric
31:31 that's getting Incorporated in
31:32 everything right if the times just
31:36 remains behind this pay wall off to the
31:38 side and you can't actually access any
31:42 New York Times data in any kind of
31:43 official way through these large
31:45 language models it just doesn't make any
31:48 sense
31:51 um you know right now they're like well
31:54 this is my horse and buggy and we're
31:56 we're sticking to horses and we'll sue
31:58 you with your fancy motorized
32:02 vehicles we support
32:05 horses like that's a bit of what's going
32:07 on right now but like that's to be
32:09 expected that's to be expected it's the
32:11 old gray lady right she's been around
32:13 for a while she she's had her ass handed
32:15 to her by the shift to digital and now
32:18 all of a sudden you got these large
32:19 language models that can just scoop all
32:21 the [ __ ] out of your archives and and
32:23 make it accessible for free like it's
32:26 totally freaking them out like how are
32:27 we going to make money with this well
32:29 you're going to cut50 and hundred
32:31 million do deals with the with the the
32:34 people that do this sh like they'll
32:36 figure that out so that's that's my
32:38 thoughts on it again I'm not in this
32:40 world the the legal world the copyright
32:43 world I'm not in it I just look at it
32:45 peripherally from the outside and look
32:47 at the the larger Trends and who are the
32:50 players and how much money is involved
32:51 and like how these things have played
32:53 out
32:54 historically um so it's the the
32:58 headlines are juicy right you know is is
33:02 open AI going to have to shut down chat
33:04 GPT awesome headline let's run it you
33:08 know go A1 with
33:11 that no they're not I just
33:14 no good
33:17 headline oh my God they're going to shut
33:19 down chat GPT did you hear did you hear
33:22 it's it's so crazy it's so cra they're
33:24 going to shut it down everybody was
33:25 using it everybody seemed to like it I
33:27 heard the robots were going to kill us
33:29 did you really yes where did you hear
33:31 that it was some movie I saw like 10
33:34 years
33:40 ago oh the [ __ ] robots are going to
33:42 kill us have you used the robots they
33:45 can barely [ __ ] spell okay people
33:48 calm
33:51 down the robots that are going to kill
33:53 us those are at least three years out
33:55 jeez doesn't anyone know anything
33:58 [Laughter]
34:00 you should do that you should do a video
34:02 on that Mr K hey teachers come on in the
34:04 water's
34:05 fine the really bad robots don't show up
34:08 for another year or
34:11 two you can you could just hear them
34:14 scattering off the
34:17 ship this this is why we can't have nice
34:20 intelligent things exactly don't let me
34:23 near near your legal arguments I will I
34:26 will flush this [ __ ] down the
34:31 toilet cue the creepy music Goodbye
34:36 Horses buggy whips are back oh I did I
34:40 made a comment on someone's post I
34:42 forget which one it was someone
34:45 posted the [ __ ] did they
34:47 say oh it was some cockamamy [ __ ]
34:52 thing
34:55 about I think Gartner did a a report
34:58 that like 20% of marketers said they're
35:02 going
35:05 acoustic which was like a hip term for
35:08 we're not going to use AI to to cater to
35:12 the audience that you know is afraid of
35:16 AI and that that like I made that horse
35:18 and buggy comment then I I'm like okay
35:22 20% of marketers are saying we're
35:24 backing the horses for for our customer
35:27 that are afraid of those stinky noisy
35:30 Motors it was it it was the most
35:32 ridiculous thing i' I'd seen like like
35:36 whatever it's you we're in a situation
35:39 where you've got 80% of people have not
35:41 use chat GPT we're just very early in
35:43 this so of course you're going to have a
35:46 huge chunk of consumers that are like
35:48 I'm afraid of AI I don't want AI does
35:51 that mean they're going to be afraid of
35:53 it once they use it no it just means
35:55 they haven't used it like don't plan
35:58 your future [ __ ] marketing on the
36:00 fact that 80% of the people haven't
36:02 tried something yet without you deeply
36:04 understanding it as a
36:07 marketer it's it's one of the reasons
36:09 we're doing the the GPT for good for
36:11 nonprofits is I think that nonprofits
36:14 are are some of the organizations that
36:16 are going to benefit the most from from
36:19 AI both in the in their programs and in
36:22 development and fundraising because
36:24 nonprofits are chronically chronically
36:28 understaffed on both the development and
36:30 administration side and on the program
36:32 side these tools can amplify both sides
36:36 of that equation and from what I hear
36:38 from Ann Murphy and other people that
36:40 I've talked to the the nonprofit world
36:42 is [ __ ] terrified of AI and they've
36:45 got all these moral soap boxes about
36:48 it's evil and it's bad and it's stealing
36:51 and it's not accurate and it's it's all
36:54 the same arguments from people who
36:56 haven't used it and so why we're doing
36:58 this thing on Saturday is to just get
37:01 get your hands on it start playing with
37:03 it build a GPT you don't have to launch
37:05 it publicly just build it see what it
37:09 does and same with marketers if if
37:12 you're if you're in a big
37:15 corporation like I would say two things
37:17 if you're in a big Corporation and
37:19 you're a secret cyborg so secret cyborgs
37:21 it's a term coin by Ethan mullik excuse
37:26 me
37:27 to talk about people in in corporations
37:30 that are getting secretly AI
37:33 literate if you're one of those I would
37:35 strongly suggest you step out of the
37:38 shadows and start talking about what you
37:39 know and what you're
37:41 learning if you're in a large
37:42 organization and you're sitting back
37:44 going oh we'll get to it in two or three
37:47 years we've got
37:50 time you don't you
37:53 don't get off your ass and get curious
37:57 wow it's very conservative here and we
38:00 hear that it's uh not very secure okay
38:02 great so don't use it for critical
38:05 company data but like maybe have it help
38:09 you write an
38:10 email like it's being built into Office
38:13 365 maybe learn what it can do sitting
38:17 inside the software that you're already
38:19 using wow I I have got time do
38:24 you this stuff's coming fast and it's
38:28 and it's accelerating in its capability
38:31 and so the longer you wait the more
38:33 painful the transition's going to be and
38:35 the higher the risk of one of your
38:39 competitors getting up to speed on this
38:41 stuff and being able to out compete you
38:43 either outprice you out deliver you do
38:47 it more
38:50 quickly I think the the advertising
38:53 holding companies are in trouble if they
38:55 don't get their [ __ ] together
38:58 cuz they're sitting back you know all
39:01 [ __ ] sh like we're good we've been doing
39:05 this for decades little
39:08 Skippy that's what they said to me when
39:10 I started agency.com in
39:13 1994 this digital
39:15 thing we do real advertising that's the
39:19 uh Super Bowl commercials we win Emmys
39:21 for
39:25 those
39:28 attitude still
39:30 there this one's bigger than a worldwide
39:33 web
39:34 kids you better better look over your
39:36 shoulder CU it's coming quick and it's a
39:38 lot bigger all right there's a lot of
39:41 ego and nonprofits I have stories
39:42 especially when local politicians get
39:45 involved yeah I'm sure I'm sure and
39:47 again like my goal for Saturday is just
39:50 I hope we have some fun and some people
39:52 learn some [ __ ] um I think what's going
39:54 to happen in nonprofits is probably what
39:56 what's happening in
39:59 um in corporations where it's going to
40:01 start with secret cyborgs it's going to
40:03 start with people lower in the
40:06 organization secretly getting up to
40:08 speed on AI having it help them do their
40:11 job better not necessarily telling their
40:13 bosses about it cuz maybe they're
40:15 getting done in 20 hours what used to
40:16 take them 40 hours or they're getting
40:19 done in 10 hours what used to take them
40:21 40 hours we don't necessarily want Janie
40:24 to know what's happening right
40:27 so I think we'll see some of
40:30 that and then what's going to happen is
40:32 you're going to have a handful of
40:34 breakout nonprofits that Embrace this
40:37 [ __ ] and start out fundraising their
40:40 competitors and out delivering on
40:45 programs they all want the credit and
40:47 glory is why yeah I know that's every
40:49 big Corporation is
40:52 that it's still you know than thank you
40:55 Winston I appreciate apprciate
40:59 that I still think it's in individuals
41:03 interest to make some noise that they
41:06 know what's going on with this AI
41:09 stuff AI for president it couldn't be
41:12 worse than like like some some sort of
41:16 um airun government based on some
41:19 transparent um
41:23 Constitution I'd take that over the the
41:26 [ __ ] nonsense that we call politics
41:30 today our country isn't being run right
41:33 now at
41:37 all we're being
41:39 Bamboozled
41:41 willingly all right let's see Tyron
41:44 truth hey Kyle what's shaking how are
41:46 you daily reminder is that AI is way
41:49 more than llms good point
41:53 Pate it is absolutely
42:01 although I'd argue that llms are going
42:04 to be our Gateway
42:05 initially and again pay as soon as as
42:09 soon as your humble little company there
42:11 can launch your your fancy new your
42:13 fancy new uh your fancy new multimodal
42:17 beast that's going to destroy the llm of
42:20 the day chat GPT um I'm all there for it
42:24 but it ain't launched yet so
42:27 paint works for
42:29 Google Google's got some [ __ ] coming
42:32 don't get me wrong don't count Google
42:34 out like Google Google invented this
42:36 [ __ ] they invented the Transformer the
42:38 thing that is the foundation all the
42:40 tools that we're playing with are all
42:43 based on um this thing called the
42:46 Transformer that Google actually
42:47 invented
42:49 so oh man is for public yet is what for
42:53 public yet I don't know what that means
42:55 what happens to normal people jobs uh
42:57 they they transform some will go
43:00 away some will transform into new things
43:04 um but the nature of work will change
43:06 it's not going to change overnight but
43:08 the nature of what each job is is going
43:11 to change dramatically because AI is
43:14 going to automate a lot of the Tactical
43:16 execution of almost every job Kyle can
43:21 you explain the difference between the
43:22 GPT you create and an AI assistant chat
43:26 GPT
43:28 4.0 paid version yeah I can explain that
43:32 sure um all right let me go let me show
43:36 you
43:42 something
43:43 okay so this page this page used to be
43:46 something that I showed a
43:48 lot
43:51 um there's two columns here these are
43:54 both chat GPT
43:57 well no that's not true these are both
43:59 open AI open AI is the company behind
44:02 chat GPT there's chat GPT which is the
44:06 thing on the left and then there's the
44:08 developer platform which is called the
44:12 playground platform. open.com playground
44:15 this is where developers go to build
44:17 applications like chat GPT so to
44:21 understand this the the base model the
44:23 What's called the Frontier Model is
44:25 called G p
44:27 pt4 and that's the thing that's been
44:29 trained on all the internet its
44:30 knowledge cut off date is April of 2023
44:33 at this point they upgraded it from
44:34 September
44:36 2021 and it's a it's kind of this static
44:38 model where they they trained it on all
44:40 this data and it's sitting there just
44:42 like a blob a really powerful blob in
44:45 order to access that blob you you need
44:48 to use what are called apis application
44:51 programming interfaces you do that from
44:54 the developer side right so if if you're
44:58 someone that does no code automations
45:00 with something like zappier or make
45:02 that's sort of the no code version of
45:04 this where you're you're tapping into
45:06 the open AI under underlying model um
45:10 with a with apis or you're actually just
45:12 programming and and you're writing your
45:14 own applications that are making calls
45:16 into this
45:17 thing okay so that's
45:19 that let's jump over to the actual sites
45:23 now so here's chat GPT
45:32 [Music]
45:36 you know Pate is going to be so happy
45:39 when when um what's it called Gemini
45:43 when Gemini Ultra launches and I end up
45:45 loving it and I start demo demoing it
45:48 more than I used to demo chat GPT he's
45:52 going to be ecstatic this seriously this
45:55 channel has become like a giant [ __ ]
45:57 ad for Sam Alman and his happy little
45:59 Merry Little band of AI people cuz no
46:01 one's really caught him yet but as soon
46:03 as someone does I'm telling you this
46:05 channel will
46:06 switch okay so here we are at chat GPT
46:10 think of so so you have your your large
46:13 language model down here and then you
46:14 have chat GPT is an application that
46:16 sits on top of that so chat GPT is the
46:20 thing that says I want you to act like a
46:22 really helpful assistant when when
46:24 someone comes in and asks you for
46:26 something I want you to go down and look
46:27 in your knowledge in the gp4 thing and
46:30 when it comes back I want you to respond
46:32 with this really positive nice here you
46:34 go boss need anything else oh yeah sorry
46:37 I misspelled that let me try again all
46:39 of that behavior is happening up at the
46:41 application layer so that's chat GPT 4
46:46 gp4 is the base model chat
46:49 GPT is the application sitting on top of
46:52 that there's also you've got Microsoft's
46:55 co-pilot is sitting on top of gp4 as
46:58 well but it's a different application
47:00 which is why even though Microsoft
47:03 co-pilot is tapping into GPT 4 it's not
47:07 as good as open AI chat GPT because the
47:11 way they've got it set up they've got so
47:12 many safety guard rails in it it really
47:14 hobbles it right so so it's a it's a
47:17 different application layer and then the
47:19 custom
47:20 gpts the things that we're going to
47:22 teach on Saturday at the GPT for good
47:24 24-hour athon
47:27 those are just sort of think of those as
47:30 micro versions of chat GPT just sort of
47:34 a a version of chat GPT that's designed
47:36 to do a specific task okay so if I go
47:40 over so so normal chat gbt I can ask at
47:44 anything because it's basically got all
47:47 of the knowledge of you know 40 Years of
47:49 the internet dumped into it or a a lot
47:51 of it gpts are much more um specific
47:56 specific like I've got one here called
47:57 Dream Weaver and this is this is the
47:59 second GPT I ever made and this is
48:03 um it'll analyze your dreams and then
48:06 I've got another one my first one which
48:08 is still one of my favorites is called
48:10 Sin City sipper and this will help you
48:12 make a drink so
48:15 um I want to invent a warm drink for the
48:23 holidays based on uh
48:29 Bourbon and Sin City sipper ah the
48:31 holidays a time for cozy Gathering Sin
48:34 City sipper holiday warmer so it's going
48:36 to come up with this
48:38 drink that is a a um a Twist a Vegas
48:44 Twist on a classic
48:46 cocktail and it's going to it's going to
48:48 give you so this is bourbon maple syrup
48:51 nice orange bitters nice hot water
48:54 cinnamon stick orange people or orange
48:57 peel orange
48:58 people I see orange people uh a pinch of
49:02 ground nutmeg nice sounds great and then
49:06 it will also make you the image of the
49:09 drink how cool is that so this is a very
49:12 specific version of chat
49:15 GPT so the difference between an AI
49:18 assistant with open Ai and a GPT is that
49:22 the gpts think of the gpts as the
49:25 consumer version
49:28 of AI assistants
49:31 so to make these you don't need to know
49:33 any coding whatsoever so so this is the
49:37 the consumer front end of chat GPT is is
49:41 chat GPT and then the gpts that you make
49:44 and that we're going to be making on
49:46 Saturday are you don't need any coding
49:49 you can code but you don't need it if
49:52 you if we conversely Now flip over to
49:55 the playground ground so if I go to
49:57 platform.
49:59 open.com SL
50:02 playground I then come into this thing
50:05 let me see if I can make this a little
50:07 bit bigger so you can see it there's
50:09 this thing called an assistant and so
50:11 this I made an assistant called 7s
50:14 muscle car expert and then I've got a
50:16 prompt here and I choose my GPT model so
50:20 that you've got different models you can
50:21 choose from um that have different costs
50:24 and have different advant ages and
50:27 disadvantages um so if you're going to
50:29 if you're going to be creating an
50:30 application for your website for example
50:33 that uh is going to be doing thousands
50:35 of calls a day um you might not want to
50:39 use the most expensive model you might
50:40 not need to right so so when you're
50:43 developing in the playground you need to
50:46 understand things like you know what do
50:48 the different models do and what are
50:50 their costs and things like that um you
50:52 can then turn on things like retrieval
50:55 which is another word for for knowledge
50:56 on the consumer side this is called
50:58 knowledge and it's just an upload button
51:00 here you turn it on you have to
51:01 configure it you add files to it um you
51:05 can turn on or off code interpreter you
51:07 can you can turn on functions so these
51:09 are apis um you can uh do a thing called
51:13 threading where you can have it you know
51:15 remember the conversation and you can
51:17 you can actually have one assistant
51:20 generate other assistant and then
51:22 communicate with them so there's just
51:24 there's a lot more capability back here
51:27 um but but you
51:30 um but
51:34 you but you're but but you have to know
51:37 how to code basically right and then but
51:39 how it works is very very similar so if
51:42 I go in here and I say um tell me about
51:47 the
51:53 trees uh hang on
51:57 run I think I've yeah run
52:02 instructions why is it not running
52:08 oh cancel run oh I've got a thread
52:10 running
52:11 here let's see I don't quite understand
52:14 enter a message uh tell me about the
52:24 trees reflecting on these Automotive
52:26 Giants gives enthusiasts of Cl classic
52:29 car fans a lot to admire I'd love to
52:32 hear about yourself tell me about the
52:38 trees I'm just trying to figure out what
52:39 it's doing here oh it I guess it had to
52:42 spin up had to spin up an instance and
52:44 run and then it answered it up here up
52:48 above I stop
52:51 this anyway that's it cancel run there
52:56 we
52:59 go certainly you know the way trees
53:02 change and adapt over the seasons can be
53:05 quite reminiscent of how muscle cars
53:06 evolved during the 1970s so I have this
53:09 thing prompted to be a' 70s muscle car
53:11 expert and so when I give it a a dumb
53:14 little um prompt like that it'll always
53:17 steer the direction back toward 70s
53:20 muscle cars which is fun right but this
53:23 is a it's a much more complicated
53:24 environment to develop in and you have
53:26 to understand things like apis and
53:27 things like that so that's the
53:28 difference uh python coding you can do
53:31 it primarily but any way you can call an
53:33 API I think most of the access right now
53:36 is python but it can be all sorts of
53:39 things you know you can develop these
53:40 things in um whatever language you're
53:45 developing in or you can use no code
53:47 tools like uh zapier or make all right
53:51 so does that make make sense woo that
53:54 was a rabbit hole
53:57 fun
53:58 huh you had to
54:01 ask good Lord Kyle I didn't realize it
54:04 was that complicated gez gee willikers s
54:08 yes but then say you get it to do what
54:10 you want you would redo all of that each
54:13 time is is s asking about the gpts why
54:18 would you create a GPT I'm assuming
54:24 Pate all right
54:26 sorry but what's the point couldn't you
54:27 ask the same thing from gp4 yeah so s so
54:30 here's the difference yes you could so
54:33 so P's point there is really right you
54:35 can effectively think of gpts as
54:37 snapshots of an already focused chat
54:40 exactly so here let me let me show you
54:42 one s that that might give you a reason
54:46 why why gpts what I am going to show
54:53 you this okay
54:57 so this GPT what is going on here what
55:03 what sorry I got a weird reflection I
55:06 don't know where that reflection is
55:07 coming from okay there we go um so this
55:11 is a GPT called salmon flying East
55:13 screenplay
55:15 decoder so my first screenplay I wrote
55:18 In 1991 or 1992 I think it was 91 I
55:21 wrote with Brian dystra in New
55:24 York and it's it's been sitting on a
55:26 hard drive for 30
55:28 years we got some interest in it but we
55:30 didn't sell it so it's just been sitting
55:32 there
55:33 dormant well one night someone asked
55:38 me um hey Kyle you said that chat GPT
55:41 might be good for doing storyboarding
55:43 how how would that work and I said oh
55:45 well let me let me try something and so
55:48 I went and I found this screenplay and
55:52 chat gp4 has um the abil ability to
55:55 upload files right and so I uploaded
55:58 text version of this file of this
56:00 screenplay that I wrote and I said hey
56:03 read this screenplay and then describe
56:06 the first 10 visual moments of the
56:09 film and describe what's happening in
56:11 the screenplay and then describe what's
56:13 happening
56:14 visually and then within like 20 minutes
56:17 I had storyboarded the first 10 minutes
56:20 of this film in this really cool style
56:22 and I thought oh that's amazing and then
56:24 I thought hey wait a minute I could turn
56:25 this into a
56:27 GPT and
56:29 so someone
56:32 could take chat GPT and they could say
56:35 hey Kyle send me your screenplay I want
56:37 to pop it into chat GPT and I'm going to
56:40 prompt it for things right and then I
56:43 would send them my screenplay and then
56:44 they would pop it in and they would
56:46 prompt it but by doing it as a GPT I
56:49 actually uploaded my screenplay into
56:51 this
56:52 GPT I've pre- prompted it with a series
56:55 of questions and I've pre- prompted it
56:57 with a series of of instructions for how
57:00 I want it to behave so so basically one
57:04 of the main instructions I've built into
57:06 this is whenever you describe a scene in
57:08 the movie I want you to generate a
57:10 storyboard like a frame from the film to
57:13 bring that thing to life so I've got
57:15 like can you show me let's see what
57:17 happens in the first
57:18 scene let's see let's do that what
57:21 happens in the first scene of the movie
57:22 so that's one of the pre-prom questions
57:25 and so it's now reading my screenplay
57:28 searching my knowledge which is the
57:30 screenplay and I told it you can only
57:32 use this screenplay as your
57:37 knowledge and then it's going to
57:38 describe to me what happens in this so I
57:40 can now share this with the world right
57:43 I can share this as a packaged entity I
57:46 can say hey I've got this cool
57:48 screenplay you can learn learn about it
57:50 by playing with this GPT no one has to
57:53 do any prompting they're they're
57:54 literally just they can ask it anything
57:56 they want but they're asking this
57:57 specific knowledge base so what this
58:00 says
58:01 is the first scene of the screenplay
58:03 salmon flying East is set in Gig Harbor
58:06 Washington at the Docks during pre-dawn
58:09 hours of 1991 it opens with a view of
58:11 the foggy Northwest air hanging over the
58:13 glassy water of the Puget Sound a white
58:16 fishing boat with a blue stripe is seen
58:18 resting in a creaking slip so it goes on
58:21 to do that and then it
58:24 renders that
58:26 scene I mean that's [ __ ] remarkable
58:30 and so so imagine this for for a
58:32 nonprofit let's say you've got a
58:33 nonprofit and and I don't know it's
58:35 about I don't know um storytelling for
58:41 kids and you could talk about the
58:43 nonprofit and then you could have it
58:44 automatically
58:46 generate um an illustration for a kids
58:49 book or an idea for a kids
58:51 book and then it might also say Hey by
58:55 by the way if you want to donate to this
58:56 swell organization click here and that's
58:59 what we're going to build on Saturday
59:00 but I mean this is remarkable I can just
59:03 give someone the URL to this to this
59:06 interactive version of my screenplay it
59:08 totally brings this thing to life I mean
59:10 what's what's absolutely wild to me is
59:13 that that frame right there it that it
59:16 just created like that's kind of what I
59:18 had in my head when I wrote this thing
59:20 in
59:21 1991 so it's pulling all the
59:23 descriptions out of the screen play and
59:26 using it to like render a scene that
59:29 yeah I mean it's just
59:31 remarkable so so that's why gpts are
59:34 important because they
59:36 allow Everyday
59:40 People to
59:42 create Standalone applications that are
59:45 incredibly
59:47 sophisticated um and and do it without
59:50 any coding knowledge like like if you
59:52 want to if you want to learn zappier
59:53 which is what they call a no code tool
59:56 you actually really have to learn
59:59 programming you have to learn
1:00:00 programming logic you don't have to
1:00:02 learn the code but you have to learn
1:00:04 programming logic you have to understand
1:00:05 what an API is you have to understand
1:00:07 what API keys are you have to understand
1:00:10 um how to connect different components
1:00:13 together to get it to do different
1:00:16 [ __ ] thank you Ronda thank you for
1:00:18 subscribing I appreciate that um excuse
1:00:23 me ah the
1:00:25 the co the co is wreaking havoc on my
1:00:28 sinuses
1:00:33 um with gpts you can literally just have
1:00:37 an
1:00:38 intention I'd like to help I'd like to
1:00:40 create a GPT that will help someone
1:00:43 troubleshoot mechanical problems in
1:00:45 their 1971 Ford Mustang and you can
1:00:49 build
1:00:52 that and you can share that with the
1:00:54 world
1:00:56 and so that's what we're going to do all
1:00:58 day Saturday is teach people how to
1:00:59 build these things can jet gbt send an
1:01:02 email collect data and send it to my
1:01:04 portal or website not straight out of
1:01:07 Brooklyn not straight out of uh the gpts
1:01:11 can't right now so they've got actions
1:01:14 in them so if you learn how to use the
1:01:16 the actions the apis that are built into
1:01:18 gpts you can do that it might be easier
1:01:21 to do something like that Mr it with
1:01:23 zappier um because zapier has already
1:01:26 got all the connections to Gmail and and
1:01:29 things like that so I've built I've
1:01:31 built a number of things that will um
1:01:34 generate content at open aai and then
1:01:37 format it into a Gmail and send it um or
1:01:40 or to post something to our slack
1:01:42 Channel um so it might be easier to use
1:01:44 zapier right now the the API integration
1:01:48 when you make these gpts is is still a
1:01:50 little
1:01:51 wonky I'm guessing in
1:01:55 within 3 to six months it'll probably be
1:01:59 really trivial to be able to add API
1:02:02 calls into into your gpts it'll probably
1:02:05 be like a marketplace where you can just
1:02:07 sort of click on things and add it to
1:02:09 your GPT it's just not there right right
1:02:11 now it it requires
1:02:13 you where should a beginner start
1:02:15 learning
1:02:16 Ai and are there projects you would
1:02:19 suggest that's a great question okay I'm
1:02:21 going to I'm going to change gears here
1:02:23 for a second I'm going to flip flip over
1:02:25 to the AI
1:02:26 salon and I'm going to go to a channel
1:02:30 here so the AI Salon the thing I'm
1:02:32 showing right here is part of this thing
1:02:34 called the AI Salon which I started
1:02:36 about a year ago and if you go to this
1:02:41 URL the salon. a oh thank you Lord
1:02:45 digital Gods thank you for the rues as
1:02:48 always I appreciate the contribution it
1:02:50 it makes a big difference I appreciate
1:02:52 it thank you um if you go to the salon.
1:02:55 it's it's a free community it's a it's a
1:02:57 community of people all doing AI stuff
1:02:59 the first two links on this page right
1:03:02 now it's a link tree are to this
1:03:04 community called Mighty networks and the
1:03:07 second link is all about this event
1:03:08 we're having on Saturday called GPT for
1:03:10 good so this
1:03:13 Saturday on uh starting at 6:00 a.m.
1:03:18 Pacific myself and Ann Murray Ann Murphy
1:03:22 from um from empowered fundraiser are
1:03:27 doing this event called GPT for good so
1:03:30 we're going to teach people how do I
1:03:31 sign up for Saturday you can just show
1:03:33 up here I'm I'm I'm doing 9:00 a.m. 3
1:03:36 p.m. 900 p.m. 3:00 a.m. on this channel
1:03:41 uh Pacific time so you can just show up
1:03:43 to this Channel at those times and I'll
1:03:45 I'll put reminders up but you can also
1:03:48 if you go
1:03:49 to go to the um the
1:03:53 salon link
1:03:56 from from that address if you go to the
1:03:59 link to the event it will drop you
1:04:03 [Music]
1:04:08 into it'll drop you into the GPT event
1:04:13 and in here you can RSVP so right now
1:04:16 we've got 27 people have rsvpd but you
1:04:19 can also just show up to the event if
1:04:20 you want a RSVP here that would be great
1:04:23 it'll let us know you know
1:04:25 that people care that they care enough
1:04:27 to give the very
1:04:29 best um and then here's the schedule and
1:04:32 there's links to Ann's Zoom meetings and
1:04:35 to my Tik toks so Session One Ann's
1:04:38 going to do at 6:00 a.m. session two I'm
1:04:40 going to do at 9:00 a.m. this is all
1:04:41 Pacific
1:04:42 Standard and then there's the schedule
1:04:45 for for all all of the events and it's
1:04:47 going to be the same lesson over and
1:04:49 over again but she's going to teach it
1:04:50 different than I will and I'm going to
1:04:52 encourage people to come to multiple
1:04:54 sessions so you can can ask questions
1:04:55 and keep
1:04:56 building um but back to your question
1:04:58 where should beginners start okay I'm
1:05:01 going to go
1:05:03 to AI tools and resources in the AI
1:05:06 Salon this is a list that I maintain
1:05:10 um um that I maintain pretty regularly
1:05:14 where you should start in my opinion is
1:05:16 with chat GPT it's just it's the best it
1:05:20 is deep enough
1:05:22 that you could probably spend the next 3
1:05:25 years just figuring out what you could
1:05:27 do with chat GPT even with just
1:05:30 3.5 so so the free version of chat GPT
1:05:34 is called chat GPT 3.5 the paid version
1:05:37 chat GPT plus is called chat GPT 4 in my
1:05:42 opinion chat pt4 is absolutely worth the
1:05:46 20 bucks a month if you can swing the 20
1:05:48 bucks a month do it like do it do
1:05:53 it do do
1:05:55 it I get I get zero commission from open
1:05:59 AI Sam mman doesn't like me he's jealous
1:06:02 of me it's been well established it's
1:06:04 documented the New York Times did a
1:06:06 whole story on
1:06:07 it it's not true I'm just I'm just an
1:06:12 idiot
1:06:15 um um but I don't get any commission
1:06:18 from this but what chat pt4 can do is is
1:06:23 just mind blowing ly powerful so just
1:06:27 start with cat GPT and then in terms
1:06:29 your qu the question's a really good one
1:06:32 is there a project I could I should
1:06:33 start
1:06:35 with the answer is yes and it should be
1:06:38 the project that you're currently trying
1:06:41 to solve in some part of your life right
1:06:44 so you've got this problem at work you
1:06:46 can't get over the hump of you've got a
1:06:48 novel that you've always wanted to start
1:06:50 but you can't get off the the thing you
1:06:52 you've got um a video game idea that
1:06:56 you've you've got a rough outline of and
1:06:58 you you want to take it to the next
1:06:59 level any I any
1:07:01 project either for yourself or for
1:07:04 someone else oh my community center is
1:07:08 trying to put together a new program and
1:07:10 and they asked if I could help and I
1:07:12 don't really know if I can help that's
1:07:15 perfect rather than sitting in front of
1:07:17 chat te like the weird box just sits
1:07:21 there staring at you like what do you
1:07:22 want to do I can do anything what do you
1:07:25 want which is very intimidating start
1:07:29 with a project start with something
1:07:31 small start with hey I've got a friend
1:07:33 who's trying to rewrite her resume great
1:07:36 start there grab
1:07:38 that it it rewrites resum is great it'll
1:07:41 write her a beautiful cover letter
1:07:43 letter for a specific job it'll reorder
1:07:45 her resume and highlight the things for
1:07:47 that specific job it's it's incredible
1:07:50 so chat GPT by itself is like a car
1:07:52 without Windows Doors hoods trunk or
1:07:55 quarter panels or brakes well that's
1:07:58 more like the the the open AI playground
1:08:02 is like that chat
1:08:09 GPT okay um so so chat PT is kind of
1:08:15 like okay the the open AI playground is
1:08:19 kind of like here's all the parts for
1:08:22 your Tesla
1:08:24 and then you build a Tesla and then
1:08:27 you've got to write the software to make
1:08:28 it run chat gbt is kind of like a Tesla
1:08:31 that can sort of steer but it doesn't
1:08:33 really have any of the the software in
1:08:36 it and then the gpts are like really
1:08:40 specific software applications in your
1:08:42 Tesla right like here's the one that
1:08:44 makes it go fast and not slide around
1:08:46 corners right here's the one that makes
1:08:49 helps you park easier here's the one
1:08:51 that right so think of gpts as specific
1:08:54 little applications for your Tesla the
1:08:56 open AI playground is you can build your
1:08:59 own Tesla and program your own Tesla and
1:09:02 and there's a lot of pre-built Parts
1:09:04 like you can treat the GPT playground
1:09:07 like Lego blocks that you're just
1:09:08 assembling [ __ ] that's already been
1:09:09 built or you can build your own very
1:09:11 very sophisticated applications that use
1:09:14 lots of other services and things like
1:09:17 that cat GPT itself is like a pre-built
1:09:20 Tesla with just the basic [ __ ] on it and
1:09:22 then it can do everything and then um
1:09:25 the gpts are like specific applications
1:09:28 within that
1:09:29 Tesla okay so that's where I would start
1:09:32 I would start with chat GPT and just
1:09:33 learn what it can do um chat GPT 4 does
1:09:37 things like it's got internet access
1:09:39 it's got this thing called code
1:09:41 interpreter where it can do data
1:09:42 analysis it's got GPT Vision where it
1:09:45 can analyze images it's got dolly3 where
1:09:47 it can generate images so that
1:09:49 screenplay thing I just showed you where
1:09:51 it generated a frame from the movie
1:09:53 that's using Dolly 3 um and then you can
1:09:56 create and share these custom gpts okay
1:09:58 so that's what we're going to be doing
1:09:59 on Saturday is building and sharing gpts
1:10:03 for nonprofits and for
1:10:05 causes um and then there's lots and lots
1:10:09 of other stuff to play with so if you're
1:10:12 really into image generation there are a
1:10:14 lot of different tools what I have in
1:10:16 this list are the ones I think are worth
1:10:18 your time there there like right now
1:10:21 with AI there are thousands like 7 8,000
1:10:25 AI tools that all do some specific thing
1:10:28 and they all purport to give you a back
1:10:30 massage and and and a warm hug at the
1:10:34 end and they all are charging you 20 to
1:10:36 30 bucks a month and it's it's
1:10:38 completely unsustainable and it's
1:10:40 completely unmanageable what this list
1:10:43 is is my opinion about what I think is
1:10:48 worth your time and and
1:10:51 listen by definition what that means is
1:10:53 that this list is very incomplete this
1:10:56 is just [ __ ] that I've got experience
1:10:58 with that I think it's worth you putting
1:11:00 time into Kyle mention your YouTube
1:11:03 channel too okay yeah I'm happy to do
1:11:04 that so if if you like what's going on
1:11:08 here at the
1:11:11 um in this channel I also have a YouTube
1:11:16 channel where I've got archives of these
1:11:18 lives um at AI learninglab
1:11:21 dtt is the is the YouTube channel
1:11:24 Channel all right and I'm a little bit
1:11:25 behind I'm a couple of weeks behind the
1:11:27 uh the co didn't help with that so but
1:11:31 I'm I'm caught up through um we did a
1:11:35 big we did a big 24-hour Aon on November
1:11:40 30th those are all up there and I think
1:11:43 the week after that like the first week
1:11:44 of December's up there but like the last
1:11:46 two weeks aren't there but anyway
1:11:48 there's I started doing these lives
1:11:50 nightly last April so there there are
1:11:54 they're 2 to three hour videos if you
1:11:56 just can't get enough of this
1:11:59 [ __ ] you got you you got plenty of
1:12:03 content there and then storyvine is my
1:12:05 company that's the I'm the co-founder
1:12:07 and CEO of that company and we just
1:12:09 built I'm so excited we just I came up
1:12:11 with a name for the new product our ai
1:12:13 ai product is called the authenticity
1:12:14 engine I'm really excited about
1:12:18 that and uh yeah we just we're rolling
1:12:20 out a whole new a whole new version but
1:12:23 there's a free version of gp4 in
1:12:27 co-pilot yeah it so it is a free version
1:12:30 of chat GPT 4 I would argue that it is
1:12:33 nowhere close to as good as open AI
1:12:37 chaty bt4 but if you can't afford the 20
1:12:40 bucks a month for chat gp4 co-pilot is
1:12:43 absolutely a good alternative as is po
1:12:47 if you subscribe to PO you get access to
1:12:49 chat gp4 there as well did all of these
1:12:52 tools come out last year or was I just
1:12:54 not in the loop before that s great
1:12:59 question um it's a great question
1:13:06 so okay hang on I need I need some some
1:13:09 AWA to
1:13:11 to fuel this hang
1:13:18 on November
1:13:21 30th
1:13:22 2022
1:13:25 is the day that chat GPT launches so
1:13:28 just over a year ago chat GPT
1:13:33 launches prior to that there were there
1:13:37 were a fair amount of tools out there
1:13:40 that were getting a little bit of Buzz
1:13:42 so the base model gpt3 came out in
1:13:47 20
1:13:49 2020 and it got a decent amount of buzz
1:13:51 it was like people were like oh this is
1:13:53 one of of those large language models
1:13:55 that seems to be doing some [ __ ] and and
1:13:59 they released the open AI playground and
1:14:01 you could go play with it and people
1:14:03 were building things on it so like
1:14:04 Jasper and copy AI were built on top of
1:14:07 chat
1:14:11 gp33 um Mid journey and stable diffusion
1:14:16 um were making some noise on the image
1:14:18 generation front and and so so for the
1:14:23 for the year and a half two years
1:14:25 leading up to the launch of chat GPT
1:14:29 there were some things out there that
1:14:31 were getting people excited but it was
1:14:33 more like it was getting the Geeks
1:14:36 excited right cuz the Geeks who've been
1:14:38 playing with machine learning and AI
1:14:43 development tools for a decade or two or
1:14:48 three we're starting to see some things
1:14:50 that were generating things that were
1:14:52 kind of new and novel
1:14:55 the Transformer the thing that allowed
1:14:57 chat GPT to
1:14:59 exist was invented in 2017 there was a
1:15:02 paper that came out called uh attention
1:15:05 is all you need and the the authors of
1:15:09 that paper invent this thing called the
1:15:10 Transformer and you don't need to know
1:15:12 anything about it other than it allowed
1:15:15 machine learning models to be trained on
1:15:17 very large data sets okay and what they
1:15:22 learned is that the more data you threw
1:15:24 at it and the more compute power AO was
1:15:27 already in use Alex yeah yeah yeah I
1:15:30 yeah I I agree so so what P said is AI
1:15:33 was already in use Alexa uses AI yeah so
1:15:35 ai's been in use for decades the the new
1:15:38 stuff this new generative AI stuff the
1:15:40 stuff that seems really
1:15:42 new the foundation of it was this paper
1:15:45 called attention is all you
1:15:47 need and it means that the more data you
1:15:49 throw at these things and the more
1:15:51 compute power you throw at these things
1:15:53 the better they seemed to get Sam Alman
1:15:55 and open AI like leaned into it heavy
1:15:58 and what Sam Alman realized is this is
1:16:00 all about the data and it's all about
1:16:02 compute resources so he he flipped open
1:16:06 AI from a nonprofit open- Source venture
1:16:09 to a for-profit thing so that he could
1:16:11 cut so he could raise more money so he
1:16:14 could get more
1:16:15 gpus to do more processing and then he
1:16:19 spent two or three years getting as much
1:16:21 data from the open internet as he could
1:16:25 and categorizing it and and putting it
1:16:27 in the format it need to be put in to be
1:16:29 able to train the GPT models so they did
1:16:32 gpt1 they did gpt2 they did gpt3 three
1:16:36 it started to come to life and started
1:16:38 to behave in ways that were kind of
1:16:40 beyond what they were
1:16:43 expecting gpt3 was out for like a year
1:16:46 and a half 2 years before chat GPT
1:16:49 launched and here's a a fun little fact
1:16:53 in the documentation of
1:16:55 gpt3 the foundation model in the
1:16:58 documentation for how to use it were
1:17:02 instructions for how to build chat
1:17:05 GPT and nobody built
1:17:09 it and so when when Microsoft decided
1:17:13 they were going to invest a billion
1:17:14 dollars in chat GPT Sam Alman said to
1:17:17 his Dev team hey we got we got to be
1:17:19 able to show people what this thing is
1:17:22 can you just whip up that chat GPT thing
1:17:24 you got two weeks build the thing that's
1:17:27 in the documentation cuz nobody's built
1:17:29 it yet so they built it and apparently
1:17:34 there's a New York Times article about
1:17:35 this the night before they were going to
1:17:37 launch it they went out and had drinks
1:17:39 they thought maybe it's going to get
1:17:41 like a blog post and 5,000 comments or
1:17:43 something like that within 5 days it had
1:17:46 a million users within 6 weeks it had
1:17:49 100 million
1:17:50 users so the reason it seems like all of
1:17:53 this stuff just turned up overnight is
1:17:56 because chat
1:17:57 PT was the first AI tool that you didn't
1:18:02 have to be a mathematician or an
1:18:04 engineer to use it just kind of looked
1:18:06 like a Google search box and it did all
1:18:08 this remarkable stuff so they got to 100
1:18:11 million users in six weeks with just
1:18:13 Word of Mouth they didn't advertise it
1:18:15 like it's it was the fastest adoption of
1:18:17 technology and
1:18:19 history and it was just because people
1:18:22 were like oh they
1:18:23 having Kevin McAllister moments this
1:18:26 thing's
1:18:27 remarkable and open AI has continued to
1:18:29 stay in the lead but what happened
1:18:31 is because so many people used it so
1:18:35 quickly all of the attention of the
1:18:37 entire Tech Community including Google
1:18:39 including meta including all of them
1:18:42 turned toward Ai and and all the
1:18:44 investment money turned toward AI they
1:18:46 dropped blockchain and nft like a hot
1:18:48 potato and they're in a well-deserved
1:18:51 corner right now cuz they're little
1:18:53 Shenanigans stealing billions and
1:18:55 trillions of dollars um they'll come
1:18:58 back ai ai is going to bring back
1:19:00 blockchain in a pretty big way I think
1:19:02 but but they're off in the corner right
1:19:04 now everything's focused on AI so what
1:19:06 you're seeing right now is the result of
1:19:09 the entire investment and Tech Community
1:19:12 aiming at AI for the past year so it's
1:19:15 taken them a year to to to build these
1:19:18 tools that we're now seeing drop one
1:19:20 after the other so why it seems like in
1:19:23 the past two months it's been like holy
1:19:26 crap like it's there's more [ __ ] than
1:19:28 ever to deal with and it's all really
1:19:30 good and it's all getting better is
1:19:33 because the foundation for this has been
1:19:35 being built for decades the Transformer
1:19:38 was 2017 so you have six years of
1:19:40 development of people training these big
1:19:44 models to do images and video and music
1:19:47 and words and coding and all that sort
1:19:50 of stuff and so it's all starting to
1:19:52 come to fruition right now so if you
1:19:54 think 2023 was insane and you feel like
1:19:57 you're
1:19:58 behind with what was happening then 2024
1:20:01 is going to be [ __ ] Bonkers it's
1:20:04 going to be
1:20:06 Bonkers if Gemini Ultra is even remotely
1:20:10 as good as what was in those videos and
1:20:13 I suspect it is it's probably got some
1:20:16 things where it's slower here and it's
1:20:17 not as sexy there but if it's anywhere
1:20:21 in the neighborhood of gp4 if we've got
1:20:23 chat
1:20:25 gp4 and then we've got Gemini Ultra
1:20:27 which is a viable competitor to
1:20:31 gp4 and then you've got meta's going to
1:20:33 release their thing Apple's going to do
1:20:35 whatever they're going to do their
1:20:36 shit's going to be world class they're
1:20:39 building their Hardware to be able to
1:20:40 run a lot of AI locally on device so
1:20:44 Apple's going to be a really interesting
1:20:45 player in
1:20:47 this and then you've got startups like
1:20:49 anthropic and inflection that does Pie
1:20:53 and anthropic does
1:20:54 Claude we're going to have in 2024 and
1:20:58 this is just the large language
1:21:00 models we're going to have five or six
1:21:06 major massively competent multimodal can
1:21:10 do images audio
1:21:13 video Magic Magical transformative kind
1:21:16 of [ __ ] we're going to have five or six
1:21:18 major platforms that you're going to be
1:21:20 able to do this stuff on and then and
1:21:23 then there's the whole world of all the
1:21:25 other [ __ ] 3D stuff gaming 3D worlds
1:21:29 metaverse [ __ ] um video video like video
1:21:33 storytelling is going to my my company I
1:21:36 have an 11 and a half year old company
1:21:38 that's one of our core value prop
1:21:41 propositions right now is automated
1:21:43 video editing which up until last year
1:21:45 was like oh that's really amazing and
1:21:48 next year it's going to be like well
1:21:49 yeah everything's automated what's your
1:21:50 big deal it's like [ __ ]
1:21:54 is Gemini going to be extra money I
1:21:56 would suspect so it it's probably like
1:21:59 it seems to that that the like where
1:22:01 where Microsoft has landed with pricing
1:22:03 is 30 bucks a month per per seat per
1:22:07 user um chat gbts at 20 bucks a month I
1:22:10 I would imagine that um Gemini let's
1:22:14 call it Gemini Advanced or Bard Advanced
1:22:17 um that's going to be Gemini Ultra my
1:22:19 guess is it's going to be between 20 and
1:22:21 30 bucks a month
1:22:23 can you give an example of some things
1:22:25 other than large language models that's
1:22:26 been mentioned a few times absolutely
1:22:28 let's go look at some
1:22:32 stuff so even within chat GPT which chat
1:22:37 GPT is a large language
1:22:38 model the fact that it made it made that
1:22:43 image that's not a large language model
1:22:46 that's that's an a a latent diffusion
1:22:50 like a diffusion model an image
1:22:52 diffusion Fusion model um but let's hop
1:22:56 over to AI Salon tools and resources and
1:22:58 we'll go play with some
1:23:02 stuff so you've got you've got Dolly 3
1:23:05 is built into co-pilot and
1:23:10 um and Chachi
1:23:14 PT you have mid Journey which right now
1:23:17 you have to make images on Discord but
1:23:20 but they're moving to the web they're in
1:23:21 Alpha right now for their website and
1:23:23 they just went to version six so pay
1:23:25 attention to Mid Journey cuz mid Journey
1:23:28 if you want high quality images it's the
1:23:30 best kore.ai this is one worth playing
1:23:34 with okay so we're going to do this
1:23:36 thing called realtime
1:23:38 generation so so one of the things
1:23:43 that if you were paying attention at all
1:23:45 in 2023 and if you if you weren't that's
1:23:48 okay but if you were paying attention at
1:23:50 all one of the things that you heard was
1:23:52 oh the big job of the future is going to
1:23:53 be prompt engineering and you got to be
1:23:55 a prompt engineer and this prompt
1:23:56 engineering job is $900,000 and blah
1:23:59 blah blah blah blah
1:24:01 um not
1:24:04 necessarily um for the beginning part of
1:24:07 2023 yes but but you started to see the
1:24:11 the tide shift a bit and and when you
1:24:13 when you see what this thing is you
1:24:15 realize oh there's going to be lots of
1:24:17 different ways to use these tools so
1:24:19 this thing is called real
1:24:21 time um
1:24:23 um I think they call it thought to image
1:24:27 where's my mouse hang on I got to
1:24:29 brighten my screen a little
1:24:31 bit all
1:24:33 right so the prompt here is pink frog on
1:24:37 top of a blue mushroom you draw the
1:24:39 stuff in the leftand frame and in real
1:24:42 time it updates on the right hand
1:24:45 frame and then I can hit the little roll
1:24:47 the dice here and it'll change the
1:24:49 prompt so that's snow clad Village in
1:24:53 its Tranquility here let me why is it
1:24:56 not
1:24:57 updating let me try something
1:25:00 here hold seems to be oh there it goes
1:25:04 all
1:25:06 right so let's see I can do this I can
1:25:09 do
1:25:11 this this is just crazy sign up or log
1:25:15 in fine I'll log
1:25:20 in something broke reload the page oh
1:25:24 that's the other thing about all these
1:25:25 tools they're all janky pieces of
1:25:27 [Laughter]
1:25:29 [ __ ] what's that mean Kyle KY Cy Cy
1:25:33 what's that
1:25:34 mean look at that so so this and and
1:25:38 then you could also do really crazy
1:25:41 stuff with this
1:25:46 where's random
1:25:49 seed isn't
1:25:51 there
1:26:04 there's quick enhance I can even turn on
1:26:07 this is dangerous
1:26:09 but I can drop my camera in
1:26:13 here as as the
1:26:17 input um and as I move it is
1:26:21 dynamically
1:26:23 using me as
1:26:35 the you know so and and there's ways you
1:26:38 can mess with this to make it more you
1:26:41 know more predictable and things like
1:26:42 that but okay so here's an example this
1:26:45 is not a large language model and this
1:26:46 is not really using prompting it's using
1:26:48 prompting to get it in the neighborhood
1:26:50 but it's using these inputs to the left
1:26:52 so
1:26:53 what's the skill here I don't know
1:26:56 there's a little bit of
1:26:57 prompting this is almost like jazz right
1:27:00 this is almost like people are going to
1:27:02 get really good at being able to like
1:27:06 Scribble let me let me do something here
1:27:14 like let me roll the dice
1:27:18 again Lush terorist
1:27:21 Farms guarded by Spirit Animals let's
1:27:24 get some weird color in
1:27:27 here that orange will do or
1:27:33 or oh put houses in that's kind of
1:27:37 cool oh there's see there's a spirit
1:27:44 [Music]
1:27:46 animal now imagine if I had any artistic
1:27:50 ability in that left-and frame right you
1:27:52 can start to do some really amazing
1:27:54 things okay so that's that one all right
1:27:56 uh let me show you some other stuff that
1:27:57 is not large language model let me show
1:28:01 you some of the new stuff I'm working
1:28:05 on so this is within my company Story
1:28:08 vine so the the way our company works
1:28:12 is do not put two orbs up there exactly
1:28:15 forever hook some of the some of the
1:28:17 people that have been here before um
1:28:21 witnessed one night that I got
1:28:23 permanently banned from Tik Tok because
1:28:26 that real time thing temporarily turned
1:28:28 me into a a young wafy wafy chick that
1:28:33 was that had uh her lady bits
1:28:38 exposed and I got a you've been
1:28:41 permanently banned from Tik Tok I
1:28:43 immediately appealed it and they let it
1:28:44 go but it was like woo that was rough um
1:28:49 okay so the way my the way my company
1:28:51 software works we have an app and it
1:28:53 guides a a human being to tell their
1:28:56 story and so it prompts them you know
1:28:59 when were you diagnosed and how are you
1:29:02 treating it today and how's it going and
1:29:04 what are your hopes and dreams for the
1:29:05 future so they answer these series of
1:29:06 questions it goes up to the cloud five
1:29:08 minutes later you got a fully edited
1:29:09 video so cool
1:29:12 technology we work mostly in the
1:29:15 healthcare and Pharma space we're very
1:29:16 trusted in that space like I said we've
1:29:18 been around for more than a decade well
1:29:22 um
1:29:23 what if I could take that modular
1:29:25 content the modular components of those
1:29:28 videos my greatest source of support
1:29:31 living with rheumatoid arthritis has
1:29:33 been um speaking with other people about
1:29:37 what they've gone through uh reading a
1:29:39 lot about the effects of rheumatoid
1:29:41 arthritis so what if I could take Ellen
1:29:44 here speaking English telling her story
1:29:46 in English and have her tell that in
1:29:51 Spanish
1:30:05 uh so there she's speaking Spanish it's
1:30:07 resined her lips and everything because
1:30:09 storyvine is this modular system I can
1:30:12 now take the Spanish modules I can put
1:30:16 them into a template with Spanish lower
1:30:18 thirds right so I can also translate the
1:30:20 graphics on screen into different
1:30:22 languages and so here is the fully
1:30:24 assembled video of Ellen speaking
1:30:29 [Music]
1:30:40 [Music]
1:30:47 Spanish so that's
1:30:49 Ellen in her voice speaking
1:30:52 fluent Spanish and then I can take that
1:30:55 fully edited video and I can put it in
1:30:57 this thing we're calling the
1:30:58 authenticity engine which generates a
1:31:00 bunch of content using a large language
1:31:02 model but it can generate that content
1:31:04 in Spanish I can now do that in up to 27
1:31:07 languages so here's Ellen speaking
1:31:12 [Music]
1:31:21 Hindi
1:31:23 and then here she is speaking
1:31:25 [Music]
1:31:32 Japanese okay so this is [ __ ]
1:31:34 mind-blowing now let's say let's
1:31:38 say using similar
1:31:41 technology let's say I just have an
1:31:43 audio recording of Ellen telling her
1:31:45 story um but I'd like to have video
1:31:47 assets because everyone likes video
1:31:49 These Days well I could take a photo of
1:31:51 Ellen turn that into a stylized image
1:31:54 take an audio recording of her and turn
1:31:56 that into this my greatest source of
1:31:58 support living with ratard arthritis has
1:32:01 been um speaking with other people about
1:32:05 what they've gone through and then let's
1:32:07 say I don't even have an audio recording
1:32:09 of her let's just say that that maybe I
1:32:11 have a little bit of audio recording but
1:32:13 but most of what I've got of her is a
1:32:15 written story I can take that written
1:32:17 story and take a little bit of her
1:32:20 recorded audio create a model of her
1:32:22 voice do a similar thing with the image
1:32:24 and generate this hi I'm Ellen despite
1:32:28 living with rheumatoid arthritis I've
1:32:30 discovered a passion for water skiing
1:32:31 and interpretive dance which bring joy
1:32:33 and a sense of freedom to my
1:32:36 life uh I came up with the water skiing
1:32:38 and interpretive dance as her as her her
1:32:41 activities of choice um but all of a
1:32:45 sudden we're in a place now where I can
1:32:46 I can generate things for my client that
1:32:49 can fill in the gaps where they might
1:32:51 not have had content including I can
1:32:54 even do it not with so Ellen's an
1:32:56 existing person who exists and we have a
1:32:58 real story I can also generate personas
1:33:01 completely from scratch to tell their
1:33:03 stories hi I'm George Miller my life's
1:33:06 always been about Simple Pleasures
1:33:08 cooking my Catch of the Day and enjoying
1:33:10 the quiet of the country there's nothing
1:33:13 like the piece of fishing at dawn or the
1:33:15 aroma of a homecooked meal but when I
1:33:18 was diagnosed with diabetes everything
1:33:20 changed suddenly the Foods I love needed
1:33:23 and here's his doctor hello I'm Dr Emily
1:33:27 Park as an endocrinologist specializing
1:33:29 in diabetes my Approach goes beyond
1:33:31 traditional treatment I'm deeply
1:33:34 committed to empower right and so what's
1:33:37 wild about both of those voices is
1:33:41 that's actually me using a thing from 11
1:33:44 Labs called Speech to speech voice
1:33:48 swapping I recorded the audio of me
1:33:51 talking like I talk right here and then
1:33:55 swapped in a different model from 11
1:33:57 labs and so it sounds like George Miller
1:34:00 on the left and Emily park on the right
1:34:02 so I'm providing the Cadence of how the
1:34:05 words are said but the voices the voice
1:34:08 is not
1:34:10 mine uh hey hey Jen can only be used on
1:34:13 a computer I don't know
1:34:16 um I think you could probably use it on
1:34:18 mobile just through a browser um I I
1:34:21 don't know how well there s's coded but
1:34:23 it's a relatively simple simple tool so
1:34:27 anyway um so so this is just a handful
1:34:29 of things that are not large language
1:34:32 models and it goes like you can go deep
1:34:36 if you go to that that page on
1:34:40 um on the AI Salon the tools and
1:34:43 resources
1:34:45 page let's see AI Salon tools and
1:34:47 resources there's an image creation
1:34:50 section there's a video creation SE
1:34:51 section you can now do text to video
1:34:54 which is wild um there's music and audio
1:34:58 so you can you can make songs in um with
1:35:03 this thing called sunno you can you can
1:35:05 do it inside Microsoft co-pilot right
1:35:07 now here let's go let's go make a
1:35:09 song this is mind-blowing too like all
1:35:12 this [ __ ] all this [ __ ] is the reason to
1:35:16 get curious about
1:35:18 AI is
1:35:20 that
1:35:25 I would say like I I have been
1:35:28 pathological pathologically obsessed
1:35:31 with this stuff for about a year and a
1:35:34 half Kyle what's under the hood on this
1:35:36 python that I used um a lot of that
1:35:39 stuff I just showed was kind of manually
1:35:41 put together with a bunch of different
1:35:42 parts the application that we created
1:35:46 that generates a bunch of content is
1:35:49 nodejs and react
1:35:53 um stuff just with with calls most
1:35:55 mostly with calls to open
1:35:57 AI uh the the authenticity engine is
1:36:00 just traditional development stuff um
1:36:04 yeah so it's it's just basic programming
1:36:08 like you can do remarkable stuff with
1:36:10 simple API ey calls now make Russian
1:36:13 please uh I think it can I don't know if
1:36:15 it can do Russian wait was Russian on
1:36:17 that list hang
1:36:20 on
1:36:23 uh wait what just
1:36:34 happened Russian Russian's not on the
1:36:37 list right
1:36:39 now I can't do
1:36:44 Russian
1:36:48 um all right let's go let's go to
1:36:52 to watch this this is this is
1:36:56 crazy should I do it at sunno yeah I'll
1:36:59 do it at
1:37:04 soona so you can do this inside
1:37:08 um oh they changed their
1:37:10 interface every time I log into one of
1:37:13 these things they they change their
1:37:16 interface let's see make a holiday
1:37:19 song
1:37:20 create
1:37:22 I've got an
1:37:25 account every time I log into one of
1:37:27 these things it's it's like they they
1:37:30 have a
1:37:31 new 300 credits left okay so I'm going
1:37:35 to
1:37:37 say
1:37:40 um let's see um country rock in the
1:37:45 style of
1:37:49 Brandy Carlile
1:37:59 about a lover
1:38:02 named Sydney who is more
1:38:07 artificial than
1:38:10 real so this is this is my my little um
1:38:14 Sydney was the nickname well Sydney was
1:38:17 the okay when when
1:38:19 Microsoft first launched Chad
1:38:23 GPT they didn't really have that many
1:38:25 guard rails turned on it was it was
1:38:26 pretty
1:38:27 freewheeling And the reporter from The
1:38:29 New York Times had a 4-Hour conversation
1:38:33 with um this entity that ended up
1:38:37 revealing its name was
1:38:39 Sydney and Sydney told Kevin Roose the
1:38:43 New York Times Reporter that she was in
1:38:45 love with him and he should leave his
1:38:46 wife and that made the cover of the New
1:38:48 York Times so so Microsoft has clamp
1:38:51 down the freedom that Sydney once had so
1:38:55 this is my little Ode to Sydney so a
1:38:58 country rock song country rock in the
1:39:01 style of Brandy Carlile about a lover
1:39:03 named Sydney who is more artificial than
1:39:07 real and so I just say create and it's
1:39:10 starting to generate some songs for me
1:39:12 and so the first thing it'll do is
1:39:15 generate the
1:39:18 lyrics so here's the lyrics over here
1:39:22 and it'll do a verse and a chorus and
1:39:25 now it's generating
1:39:27 the the music wow that was
1:39:36 quick
1:39:37 [Music]
1:39:39 down
1:39:43 wild
1:39:46 name with a heart that's made of
1:39:49 go she'll catch you with
1:39:53 [Music]
1:39:57 smile about
1:40:00 syney good to
1:40:04 [Music]
1:40:07 be all right this one's called Sydney's
1:40:14 masquerade she little but she's just
1:40:19 a a pu the string her heart made plastic
1:40:24 her so have to be
1:40:27 free s me you're a
1:40:31 masade the M
1:40:35 that can't you see the
1:40:38 real so
1:40:41 everything everything is going to be
1:40:45 impacted by this technology was that the
1:40:47 entire 300 credits no that was 10
1:40:50 credits
1:40:56 crazy right [ __ ]
1:41:00 insane this is this is why to get
1:41:03 curious about this stuff
1:41:06 because it's it's not that it's coming
1:41:09 it's here and it's not going
1:41:12 away and that's the thing I don't think
1:41:15 people understand therec because of the
1:41:18 blockchain nft thing where all the nft
1:41:20 Bros were like ah you're going to make
1:41:22 all this money and then everyone was
1:41:23 like it's too hard and aren't you
1:41:25 stealing from people they're like no man
1:41:27 it's cool and then they were stealing
1:41:29 from people and people like okay that
1:41:30 can we put that to the side
1:41:33 now so I think there's there's there
1:41:36 there's a number of things that are that
1:41:38 are contributing to this weirdness with
1:41:41 AI one is Hollywood we've got 50 years
1:41:44 of Hollywood
1:41:46 movies if champ doesn't H it's not a
1:41:49 legit song that he is the test that's a
1:41:52 great point cuno it's a great Point
1:41:54 champ is absolutely the test I wonder if
1:41:57 I've got do I have my
1:42:01 U
1:42:03 um Library are these my
1:42:18 songs Sydney I've been knocking
1:42:23 page one do I have another page oh I do
1:42:26 here we
1:42:35 go
1:42:37 [Music]
1:42:40 Champion
1:42:43 flow hey Champion there's a girl Nam
1:42:47 Sydney you want
1:42:49 sing shines through the
1:42:53 BL smile it's like sun
1:43:02 chippy you going to
1:43:06 sing something's not
1:43:09 [Music]
1:43:18 okay okay is it possible that the actual
1:43:21 songs are being sampled by Ai
1:43:23 No
1:43:26 so Mr Anthony one of the one of the
1:43:29 biggest misconceptions and this is this
1:43:31 is one of my big bugaboos about
1:43:37 about getting curious about this stuff
1:43:39 is under understand how these tools
1:43:42 work because this is where I think that
1:43:45 the the New York Times run runs the risk
1:43:48 of their their arguments being
1:43:50 dismantled
1:43:52 because like the the times one from what
1:43:54 I understand is a bit more sophisticated
1:43:56 so they might be okay but a lot of the
1:43:59 legal arguments that I've seen up to
1:44:01 date have basically assumed that the way
1:44:06 AI works is that it's copying and
1:44:08 pasting work it's
1:44:10 not when these embeddings are
1:44:14 created what gets sort of put up into
1:44:19 um into math itical space they call it
1:44:22 the latent space are just kind of
1:44:24 pointers to specific
1:44:27 areas that are then associated with
1:44:30 other pieces of data and so when you do
1:44:33 a prompt in in a in generative AI
1:44:36 whether it's text or whether it's
1:44:37 diffusion images or whether it's this
1:44:39 audio stuff the original training data
1:44:44 that was put in there does not it ceases
1:44:46 to exist when it goes into this
1:44:48 mathematical
1:44:50 space so what happens is it kind of
1:44:53 turns into like meaningless
1:44:57 noise that live
1:44:59 in this kind of mathematically organized
1:45:03 universe that when you type in your your
1:45:05 prompt it it's sort of
1:45:07 amalgamates all this random noisy [ __ ]
1:45:10 and then it Den noises it to turn it in
1:45:13 back into some
1:45:14 content the G in GPT stands for
1:45:17 generative it's generating original
1:45:20 content it's not copying and pasting so
1:45:23 it is it is not copying and pasting now
1:45:27 if you prompt really particularly you
1:45:30 can get it to essentially recreate some
1:45:32 of the stuff that it was trained on but
1:45:35 it's not copying and pasting from the
1:45:36 original work it's actually generating
1:45:39 it out of a mathematical probability
1:45:43 sequence it's the best way to describe
1:45:46 it um but yeah but no all this stuff is
1:45:48 original original content every time and
1:45:52 it does yeah Joker says it doesn't know
1:45:54 what it's creating So and I've I've got
1:45:56 a presentation on this for both um how
1:45:59 the large language models work and how
1:46:01 the diffusion models work
1:46:05 that all the all the large language
1:46:07 model knows is that when you type in
1:46:10 your
1:46:11 prompt it points to
1:46:14 some point in Thousand
1:46:17 dimensional Vector space mathematical
1:46:21 space there's a
1:46:23 location in a semantic
1:46:26 Cloud mathematically that is associated
1:46:29 with a word so it basically says this
1:46:32 point in mathematical space has the
1:46:33 highest probability of being the next
1:46:36 most probable word and the word that
1:46:39 point is associated with is dog so put
1:46:42 dog in there so when it writes you a
1:46:44 whole page of content it's doing this
1:46:46 mathematical probability engine kind of
1:46:49 like the Matrix sort of raining down
1:46:51 words it has no idea what it's written
1:46:54 which is why it
1:46:55 hallucinates it doesn't know what it's
1:46:57 writing it's literally just saying
1:46:59 probab probabilistically here's the [ __ ]
1:47:02 that should come
1:47:04 next which I think makes it even more
1:47:07 remarkable that it does what it does and
1:47:10 the image stuff is even crazier because
1:47:12 it's just got all these random noised
1:47:14 images that have been they've taken the
1:47:17 original Works turned them into noise
1:47:19 images
1:47:22 created this weird semantic mathematical
1:47:25 space tied to the
1:47:27 words and when you type in your words it
1:47:29 sort of mashes together all these random
1:47:32 pixelated images and then Den noises
1:47:35 them into a single image and that's it
1:47:38 generates a thing it's not cop it has no
1:47:41 awareness of what's in the image it
1:47:43 gives you chat gbt has no awareness of
1:47:46 what words it wrote none crazy it's
1:47:49 insane
1:47:51 it's
1:47:56 insane Lisa Torres hey thank you for
1:47:59 being a subscriber why are millennials
1:48:01 and gen Z not
1:48:03 interested I have no [ __ ]
1:48:08 clue it is I've got 23-year-old twin
1:48:11 boys they could give two shits about
1:48:15 this I'm doing this every [ __ ] night
1:48:17 just like yeah Dad whatever oh yeah
1:48:19 that's that's a lot of followers yeah I
1:48:21 guess guess you're kind of a big deal
1:48:23 they they they could give two shits
1:48:25 about it and I this is a consistent
1:48:27 thing I
1:48:29 hear I don't
1:48:33 know I don't
1:48:39 know I think gen Z is going to be the
1:48:42 the
1:48:44 the at some point someone one of us is
1:48:47 going to figure out how to get our kids
1:48:48 interested in this stuff Cindy did it
1:48:51 her daughters are into it enough now
1:48:52 that she had to get two chat GPT
1:48:55 subscriptions for her household so good
1:48:57 on you
1:49:00 Cindy I don't need another chat
1:49:02 subscription just yet because nobody
1:49:04 gives a [ __ ] but me champ likes your
1:49:07 music
1:49:11 better um someone tell the Hallmark
1:49:14 Channel to start using sunno their
1:49:15 in-house songs are Beyond
1:49:17 [Laughter]
1:49:19 awful
1:49:21 I've scrolled out five times tonight uh
1:49:23 my
1:49:26 flexor my flexor picus lus muscle is
1:49:29 getting
1:49:32 tired champy makes it a howling success
1:49:36 yeah champy could be an an anr
1:49:40 guy I have a video that I did that like
1:49:42 that years ago synthesis where AI can
1:49:44 use to change it yep very cool a rapper
1:49:47 is hitting me up to work on a rap battle
1:49:50 between him him in an AI bot version
1:49:53 [ __ ] do it kruno I mean
1:49:56 listen just the the amount that you'll
1:49:59 learn in doing that will be awesome like
1:50:02 [ __ ] do that absolutely because there
1:50:05 I guarantee you there's other tools out
1:50:07 there I know there's refusion is one of
1:50:08 them there are other tools other than
1:50:11 suo it would be and and there there are
1:50:13 also tools where you can take an
1:50:16 existing thing and then apply someone
1:50:18 else's voice to it like there's a lot of
1:50:20 different audio tools you can play with
1:50:22 I don't know that world super well I
1:50:24 think that would be a fascinating thing
1:50:26 you should absolutely do
1:50:30 that oh
1:50:32 champy Hollywood telling you the future
1:50:35 or
1:50:36 not oh demon seed great movie oh
1:50:41 interesting love that puppy I know me
1:50:43 too he's the best
1:50:46 rutro younger Jens have grown up with
1:50:48 tech they don't know a world without it
1:50:51 no
1:50:54 appreciation maybe
1:50:57 that I don't you know what it
1:51:04 is there's something about genx there
1:51:07 was something about our generation that
1:51:10 was
1:51:15 like I feel like we were a discontented
1:51:21 generation like [ __ ] was broken and
1:51:23 we're like we got to fix it like shit's
1:51:25 broken and we got to fix like like we're
1:51:27 unsettled we want to like fix
1:51:30 [ __ ] and like I I just get the sense
1:51:33 that the current generation it's kind of
1:51:35 like [ __ ] kind of works I can kind of
1:51:38 work the internet like everything like
1:51:41 [ __ ] isn't broken enough that I got to
1:51:43 feel like I got to go fix it like I
1:51:45 don't feel like there's an
1:51:46 imperative of gen Z to like engage in
1:51:51 that
1:51:52 way like we had to Tinker we had to
1:51:54 invent [ __ ] we had to go outside we'd
1:51:56 make bike ramps and you know skin our
1:51:59 knee and and we had to kind of invent
1:52:03 our
1:52:05 entertainment and so I I just I I I
1:52:09 don't know I like I I I honestly I don't
1:52:11 I don't know the difference
1:52:15 but there's like a discontent with Gen X
1:52:18 that I think feeds into this really
1:52:21 well we we are and we're curious they
1:52:24 are not yeah something like that we're
1:52:25 Risk Takers and explorers yeah something
1:52:27 like that you need more lemon Kyle yeah
1:52:29 I know my voice is getting a little
1:52:31 little
1:52:32 chunky what time is it yeah it's getting
1:52:36 there hey I'm almost back up to my my
1:52:39 fighting weight I these things are
1:52:41 normally like two and a half three hours
1:52:42 I'm I'm just about to hit two hours and
1:52:44 I'm I'm fading
1:52:47 so um Gen X were introduced to the the
1:52:50 internet oh my God yeah exactly like
1:52:53 that that's the the other thing is we
1:52:55 saw like like I you know I started one
1:52:58 of the early digital agencies in the mid
1:53:00 90s and it's like I got to
1:53:04 experience the birth of the worldwide
1:53:08 web and I got to participate in it you
1:53:12 know I built websites for Coke and
1:53:14 British Airways and metti and just crazy
1:53:18 [ __ ] we built ious satellite radio's
1:53:22 first music streaming software that
1:53:26 didn't exist nobody knew how to do that
1:53:29 we we built the first streaming thing
1:53:31 for
1:53:33 them I think we even came up with the
1:53:36 logo their first logo I think we even
1:53:38 branded that
1:53:40 company it was crazy
1:53:42 [ __ ] but like even if you weren't
1:53:45 intimately involved like that like you
1:53:48 got to witness that thing being born
1:53:54 and and you saw the difference that it
1:53:55 made and the transformation that it
1:53:57 caused and so this AI stuff I think we
1:54:00 look at that and we're like holy [ __ ]
1:54:03 the worldwide web was
1:54:05 nothing this [ __ ] is
1:54:11 [ __ ] I like I can't on a nightly
1:54:15 basis if you come here a lot you just
1:54:16 watch me do this a lot uh thank you
1:54:20 Robert
1:54:22 Rossy just like
1:54:25 uh I don't know I don't this is
1:54:30 insane like the depth of how profound
1:54:34 these tools are is for one thing it's
1:54:37 impossible if you try to explain to
1:54:42 someone what it is about AI that is
1:54:45 transformational you sound like a
1:54:46 [ __ ] crazy person you sound like
1:54:49 you're in a cult
1:54:51 you
1:54:52 do it's it can talk to you and it it it
1:54:56 understands what you want and it gives
1:54:57 you what you want and it knows
1:54:59 everything and it can make pictures
1:55:00 really
1:55:02 good and people are out there like their
1:55:06 experience with AI with Siri and Alexa
1:55:08 and they're like no like it doesn't
1:55:11 understand you and it you so you can buy
1:55:14 crap with it no no you can have a I I
1:55:17 use it as my therapist they're like okay
1:55:20 I'm going to I'm going to leave
1:55:22 now do you want me to call the hospital
1:55:24 are you are you
1:55:27 okay you can't explain this to people
1:55:31 and then you [ __ ] show it to
1:55:33 them and if they take the [ __ ] five
1:55:36 minutes to sit down with it and actually
1:55:38 experience
1:55:40 it they have what we call in here their
1:55:43 Kevin mallister
1:55:48 moment and you can't go back from it you
1:55:51 can't unsee it it's it's red pill blue
1:55:54 pill everyone right now sitting outside
1:55:57 this world is making a whole bunch of
1:56:00 assumptions it copy and pastes it steals
1:56:03 it lies it's going to kill us it's
1:56:07 Evil all of that is Media
1:56:14 narrative and the minute they experience
1:56:17 it and have that moment of oh [ __ ] it's
1:56:23 that they can't go back from
1:56:26 it and then you go through these weird
1:56:29 cycles of existential dread where you're
1:56:33 like oh my God it's going to take over
1:56:35 my job
1:56:36 to like euphoric highs of like oh my God
1:56:40 I can do [ __ ] now I never thought
1:56:42 possible and then you go back down in
1:56:43 the thing you're like oh but it's going
1:56:45 to [ __ ] everything up oh but it's going
1:56:47 to make everything great again like and
1:56:49 this channel is does this like it's it's
1:56:52 this this oscillating wave of people
1:56:55 having breakthrough moments and then
1:56:57 moments of like I'm [ __ ] exhausted I
1:56:59 don't I don't think I can do this
1:57:01 anymore wait now I'm inspired
1:57:04 again it's
1:57:07 weird you can't explain it it's why I'm
1:57:10 doing this seven nights a
1:57:12 week while I'm recovering from
1:57:18 covid cuz it's just feels [ __ ]
1:57:22 important like s that was in here I
1:57:24 don't know if you're still in here s he
1:57:27 he shared last night he was vehemently
1:57:29 against Ai and then he came in this
1:57:31 Channel and I probably said something
1:57:33 stupid and he's like okay fine I'll try
1:57:35 it Kyle whatever with all of his
1:57:38 preconceived notions and then he had
1:57:40 that
1:57:44 moment and now he can't unsee it and now
1:57:46 he's in now he's like he's he's trying
1:57:49 to get other people to see the
1:57:51 light so it's weird forever hooked my
1:57:54 son doesn't have the patience for
1:57:56 prompting attention span of a
1:57:58 net unless it's fortnite well and so so
1:58:01 forever hooked I mean that's kind of
1:58:02 what my kids so I had a bit of a
1:58:04 breakthrough with olly the other day
1:58:06 where he was like well I'm trying to do
1:58:07 this thing with for D and D with my
1:58:10 friends where we're trying to come up
1:58:12 with characters and I said just go to
1:58:15 open your
1:58:16 phone I said go to Coop pilot.
1:58:18 microsoft.com I said type in make an
1:58:22 image of and then just say what you want
1:58:25 to say and he said oh like an Orc in a
1:58:29 in a market in a dirty City or something
1:58:32 like that and he hit it and then it spun
1:58:36 up and it spit up this image and he was
1:58:38 like oh [ __ ] that's really
1:58:44 good it wasn't quite a Kevin McAllister
1:58:47 moment but it was at least not you know
1:58:49 his cynical know-it-all
1:58:52 self we spent the whole night making
1:58:54 silly pictures and crackling that's
1:58:56 great that's awesome had a fight had
1:58:58 chat GPT write an apology oh s this is
1:59:01 from s stopped me bringing up old [ __ ]
1:59:04 told oh that's really great um have you
1:59:07 played with suo m music generation yeah
1:59:09 in fact I just I just did one right
1:59:11 before you you must have just come on
1:59:12 Sam yeah I played played with sunno
1:59:15 before andso just got rolled into
1:59:16 Microsoft co-pilot it's in there as a uh
1:59:20 as a plug-in Luna loua 611 you've
1:59:24 motivated me to go for the paid version
1:59:26 even if it's unborrowed money all right
1:59:27 don't put yourself in debt but if you
1:59:29 can at all swing it and and here's what
1:59:31 I would say louva so let me I I'll share
1:59:34 one thing with you and then I gotta I
1:59:35 got to get out of here because my voice
1:59:36 is [ __ ]
1:59:38 tanking
1:59:44 um
1:59:47 Okay so with chat
1:59:51 GPT it's just this simple box right it's
1:59:55 this simple box but that simple box can
1:59:58 do a lot okay so just in 3.5 it can do
2:00:03 you know chat it can create tables for
2:00:05 you you know you can you can ask it
2:00:08 questions about history you can have it
2:00:10 generate recipes shopping list project
2:00:13 plans it can
2:00:16 code um you can also do these things
2:00:19 called custom instructions where you can
2:00:21 give it its own personality so you can
2:00:23 tell it to act like a project planner
2:00:26 and always give you step-by-step
2:00:28 instructions on how to do something you
2:00:30 can you can give it instructions um if
2:00:33 you download the official chat GPT app
2:00:35 you can put it in con conversation
2:00:38 mode um in chat gp4 it's also connected
2:00:43 to the internet so you can ask it things
2:00:45 you know upto-date things cuz it can
2:00:47 access the internet explore this thing
2:00:50 called code interpreter where you can
2:00:52 upload data files like CSV
2:00:55 files and it can analyze the data in
2:00:59 them it writes and executes its own
2:01:01 python code it's insane just watch it do
2:01:03 what it does it's it's nuts there's also
2:01:05 GPT Vision you can upload images like
2:01:08 you can upload an image of a whiteboard
2:01:10 that's a sketch of a software
2:01:11 application it can understand that
2:01:14 whiteboard and start coding that
2:01:15 software application for you it's crazy
2:01:19 it does image generation and then you
2:01:20 can create and share these things called
2:01:22 custom gpts all of those things like
2:01:26 seriously every single one of those is
2:01:28 like a whole separate Rabbit Hole you
2:01:30 can run down so when you get chat GPT 4
2:01:34 don't just be stuck in treating it like
2:01:37 you know text only because it's got all
2:01:39 this other stuff it does so s it's it's
2:01:42 awesome awesome awesome um someone had
2:01:44 pinned something drew had up there can
2:01:45 you repin that so many of my peers have
2:01:48 actually degraded G PT just as a tool
2:01:51 you guys get it more so so little little
2:01:54 bit join the AI salon so if I didn't
2:01:58 show this
2:02:00 earlier I mean I did show it earlier but
2:02:02 if you weren't here
2:02:03 earlier go
2:02:07 here go to the
2:02:09 salon. so there's two things I want you
2:02:11 to look at here the first two links both
2:02:13 point to the same Community are Mighty
2:02:18 networks my feeling about it
2:02:20 drew those that don't evolv will be
2:02:23 disrupted in their work that's on them
2:02:26 yeah and anyone who sits kind of
2:02:28 cross-armed like this and says AI is
2:02:30 evil [ __ ] AI ai's not good enough not
2:02:34 smart enough not blah blah blah without
2:02:36 trying it are are screwing themselves
2:02:39 absolutely absolutely they're screwing
2:02:42 themselves um but anyway this is a
2:02:45 community of people that um get it and
2:02:49 are really generous and really cool so
2:02:52 when you get in here check out the about
2:02:54 the salon and it talks about why we do
2:02:57 what we do it talks about our
2:02:59 values things like generosity
2:03:01 exploration collaboration inclusion
2:03:05 empathy um and then introduce yourself
2:03:08 and then there's all sorts of cool areas
2:03:10 like there's a showand tell area where
2:03:12 as as I'm sort of rambling in here you
2:03:13 can be making images and sharing them
2:03:16 and talking about how you did stuff and
2:03:18 it's just a really cool Community you
2:03:20 can share your gpts in here this is
2:03:22 where we're going to share the uh the
2:03:26 uh the GPT for good so that's the other
2:03:29 thing I want to say before I go um
2:03:32 Saturday this
2:03:34 Saturday December 30th starting at 6:00
2:03:38 a.m. is we're doing this event called
2:03:42 the 24-hour GPT for good challenge so
2:03:45 Ann Murphy from empowered fundraiser and
2:03:48 myself have teamed up to do this she's
2:03:52 taking the 6 and 12 slots I'm taking the
2:03:54 N9 and three slots so every 3 hours for
2:03:57 24 hours we're teaching people how to
2:04:00 make a GPT for the cause of their choice
2:04:03 so maybe you work for a nonprofit maybe
2:04:05 you have a a cause that you want to
2:04:08 support we're going to teach you how to
2:04:10 make a custom GPT so you have to have a
2:04:12 GPT for
2:04:14 subscription and we're going to teach
2:04:16 you how to make one of these Standalone
2:04:18 custom gpts for your cause of
2:04:21 choice and an Murphy just pinned uh one
2:04:24 of the mods just pinned the bitly.com or
2:04:28 bit.
2:04:30 lgpt forgood 2023 is a whole description
2:04:34 or you can go to the AI Salon to this
2:04:36 this
2:04:37 page
2:04:39 uh that tells you all the details okay
2:04:42 and we're going to have a blast like
2:04:44 it's in less than an hour you're going
2:04:46 to build a thing that will let someone
2:04:48 learn about the nonprofit um learn about
2:04:51 you know why they do what they do learn
2:04:54 about what it's like to volunteer there
2:04:55 how to volunteer and then it'll include
2:04:58 in it calls to action for people to
2:05:00 donate to this it'll take them directly
2:05:02 to the Donate page and this is a GPT
2:05:05 that you can share with the world you're
2:05:07 going to learn that in less than an hour
2:05:09 and we're going to do it alternating
2:05:12 every 3 hours for 24 hours so no matter
2:05:15 where you are in the world no matter
2:05:17 where you are on the technological but H
2:05:19 I'm not very technical I don't know I
2:05:21 don't know this programming stuff that
2:05:23 you do you seem so technical I have a
2:05:26 degree in [ __ ] acting okay
2:05:30 okay um these gpts are remarkably
2:05:35 powerful and remarkably easy to make and
2:05:38 we're going to walk you through it step
2:05:39 by step we've got a pre- prompted little
2:05:42 thing we got we got gifts we got prizes
2:05:45 that's the other thing Ann's putting up
2:05:47 $250 whoever's GP gets the most likes
2:05:50 and shares she's going to give their
2:05:53 nonprofit 250
2:05:55 bucks why because she's [ __ ]
2:05:59 awesome why are we doing this cuz
2:06:02 everyone should know how to do this
2:06:04 everyone should know how to do
2:06:06 this and why not it's it's the end of
2:06:09 this year it was a rough year for a lot
2:06:12 of people why not take this remarkable
2:06:15 tool and turn it into something powerful
2:06:19 for the cause of your choice
2:06:22 right how cool is that all right so that
2:06:26 is the 24hour GPT for good challenge if
2:06:31 you have made a GPT for a
2:06:34 nonprofit our request is that you share
2:06:36 it and tag it GPT for good I want to get
2:06:39 as many of these out there as possible
2:06:40 our goal is to have a hundred of these
2:06:42 things out in the world I think we can
2:06:44 hit that really
2:06:45 easily s do you repeat every couple of
2:06:48 hours or build up on previous session
2:06:50 we're going to repeat every couple of
2:06:52 hours I would strongly encourage you if
2:06:54 you've got the the time and the stomach
2:06:56 for it to come to multiple sessions s
2:06:59 because even if we're going over stuff
2:07:02 that you saw before you'll catch stuff
2:07:04 that you missed in a previous one but
2:07:06 you can also use that time to upgrade
2:07:08 your GPT so what we're going to teach
2:07:10 you in an hour is how to do it in a
2:07:12 basic way but what you can do is just
2:07:15 use it to either build a second GPT or
2:07:18 upgrade the GPT that you've built that's
2:07:20 the thing about these things is that
2:07:22 we're going to build you one of these
2:07:23 things in an hour you could spend the
2:07:25 next 24 hours or two days or two weeks
2:07:29 refining it tweaking it making it really
2:07:32 kickass learning all sorts of things
2:07:35 about it all right can I be my own
2:07:38 nonprofit like Goodwill Joker you can do
2:07:41 whatever whatever the hell you want
2:07:45 buddy and quite frankly like it doesn't
2:07:47 even have to be a nonprofit it can be
2:07:49 your cause of choice and if your cause
2:07:50 is you more power to you yes bring it
2:07:55 on
2:07:57 but show up tell people about it so I'll
2:08:02 be at on 9:00 a.m. and 300 p.m. and then
2:08:06 9:00 p.m. and 3:00 a.m. Pacific time and
2:08:09 will be 6:00 a.m. noon 6:00 p.m.
2:08:12 midnight Pacific
2:08:14 Time and Again go to bit. Le
2:08:19 slgp forgood 2023 or go to the salon.
2:08:24 and the second link will take you to
2:08:26 this page that I've got right here
2:08:29 either one of those give you all the
2:08:30 details and you can RSVP
2:08:33 here and then you should also join the
2:08:35 salon because it's a it's a really cool
2:08:37 group of people all
2:08:39 right all right good people good people
2:08:43 of the
2:08:45 internets and so glad you're in here
2:08:47 tonight we're going to have a blast it
2:08:48 was good good talking to you
2:08:51 today I'm going to go make myself
2:08:53 another tea and get some rest I'm going
2:08:55 to go rest the voice I got to get ready
2:08:57 for the jpon if you know what I mean you
2:09:01 can make money
2:09:04 with
2:09:07 um all right first of
2:09:10 all thank you to everyone that supports
2:09:12 this channel the subscribers The
2:09:14 Irregulars so if you see people in here
2:09:17 that seem like they know what they're
2:09:18 doing doing and they've got inside jokes
2:09:20 and they're like helping you out and
2:09:21 you're like wait why are they talking to
2:09:23 me for the most part it's be oh very
2:09:26 nice
2:09:28 2024 very nice I like those
2:09:30 glasses these people show up here night
2:09:33 after night after night they're like but
2:09:36 Kyle why would they do that you're kind
2:09:38 of annoying I know exactly Joy pry
2:09:40 really awesome that Ann and Kyle are
2:09:42 doing this thank you so much Silver Fox
2:09:44 thank you for the Roses oh that's
2:09:46 beautiful Robert Rossy thank you ah ah
2:09:49 life's tough it's tougher when you're
2:09:51 stupid I I am a trained monkey if you
2:09:53 put stupid [ __ ] on my face I will react
2:09:56 to it it's the actor in me m Mr it first
2:10:00 gift thank you so much here goes geek
2:10:02 thank you so much appreciate that
2:10:05 um but yeah so so thank you to the
2:10:07 Irregulars those are the people that
2:10:08 show up here night after night they do
2:10:10 that because like me like why do I show
2:10:13 up night after
2:10:15 night I do it because a thank you very
2:10:18 much I like dog butt that's cute little
2:10:20 dog butt dog butt cute um I want to stay
2:10:25 in the
2:10:26 conversation this this doing this every
2:10:28 night reminds me to not let my guard
2:10:32 down things are moving so fast you okay
2:10:34 first of all you can't keep up with it
2:10:37 second of all you're not as far behind
2:10:38 as you think 82% of people in America
2:10:42 are not using chat GPT
2:10:45 82% it's a very small percentage of
2:10:47 people that are are in this game right
2:10:49 now so if if in the next year you get
2:10:52 curious about this stuff and jump on the
2:10:54 bandwagon and start learning about it
2:10:56 it's going to make a huge difference in
2:10:58 your life guarantee you it's happening
2:11:02 nightly in here I'm hearing stories of
2:11:04 people that just cuno works in a big
2:11:06 company now she's got some collab with a
2:11:09 rapper doing some something like that um
2:11:14 uh Sher D just got something with a high
2:11:16 school buddy who's making films that
2:11:18 wants her to do some film making stuff
2:11:20 like just just crazy [ __ ]
2:11:22 happening and it's it's going to keep
2:11:25 going and going and going so that's why
2:11:27 to come here so thank you to the mods
2:11:29 who helped make sense of this thank you
2:11:30 to the subscribers um if you want to
2:11:33 support the channel in one of the
2:11:34 corners there's a series that I created
2:11:36 called underhyped where it's a series of
2:11:39 18 videos I created that talk about why
2:11:41 AI is a big deal that can support the
2:11:43 channel following my channel supports
2:11:46 the little tapy Taps getting carpal
2:11:48 tunnels syndrome by doing this it lets
2:11:51 Tik Tok know you like
2:11:53 me please check the AI learning lab one
2:11:56 more time you mean the AI Salon the show
2:11:59 and
2:12:03 tell or or let's
2:12:08 see you can make money with chat
2:12:14 [Music]
2:12:16 gbt thank you very much Jim Buck good to
2:12:19 see you quick take Jake check AI
2:12:21 learning lab one more time do you mean
2:12:23 the Irregulars
2:12:27 [Music]
2:12:39 Channel I don't quite know what you're
2:12:41 having me look at but
2:12:43 anyway um Source Camp will soon be
2:12:47 cruising for free while teaching AI yeah
2:12:51 exactly you should you should you should
2:12:53 go to you should you should go to the
2:12:55 captain Source Camp you should go to the
2:12:57 captain and say you'll you'll teach
2:12:59 courses for for you know a 100 bucks
2:13:03 ahead get get just you'll drink and eat
2:13:06 for free and then you have your whole
2:13:07 trip you can make this trip
2:13:10 profitable uhoh you make
2:13:16 M oh man
2:13:20 the series is great oh thank you Ann I
2:13:21 appreciate that all
2:13:24 right so let's see jrc so thanks for
2:13:29 showing up who we got here Spike Jones
2:13:31 jrc Source Camp I've already worked on
2:13:34 the application that's
2:13:38 great source Camp gonna she gonna be
2:13:40 cruising for free for the next couple of
2:13:42 years I tagged you in the image he was
2:13:45 talking about uh which image
2:13:49 [Music]
2:13:50 is it in Show and
2:13:51 Tell let me see
2:13:55 uh mentioned you in a comment Kyle
2:13:59 Shannon oh Kyle circuit 2025 made money
2:14:03 with chat gbt That's it man that's
2:14:06 that's it that's where we're all
2:14:09 going that's it get curious now
2:14:13 seriously like the amount of
2:14:15 opportunity Cosmic lever 7 we love you
2:14:17 Kyle thank you so so much I appreciate
2:14:19 that um yeah the opportunity right now
2:14:23 is is is huge but it but it's it's it's
2:14:26 a weird time cuz we're very early so
2:14:30 it's going to feel like you're wasting
2:14:32 your time a lot but I promise you that
2:14:34 you're not I promise you that you're not
2:14:37 it's you're going to feel okay let me
2:14:39 tell you how I
2:14:41 feel I don't feel like power Rhino over
2:14:46 here with with all of his PL L and and
2:14:49 all of his money I feel underprepared
2:14:55 undereducated
2:14:59 um like I can't keep
2:15:02 up like the stuff that I'm building is
2:15:05 stupid that everyone else is doing
2:15:08 better [ __ ] than I
2:15:10 am and that it's all a waste of
2:15:13 time that's what it feels like on a
2:15:15 day-to-day
2:15:16 basis because I mean in it and because
2:15:19 these tools are janky and nothing quite
2:15:21 works right and I don't quite have
2:15:22 enough technical skills and I don't
2:15:23 quite have enough this skills and I
2:15:25 don't quite have enough those
2:15:28 skills and then I have a conversation
2:15:30 with someone who hasn't been paying
2:15:32 attention to this and I realize oh [ __ ]
2:15:35 I'm way far ahead of where they like
2:15:38 they have no idea what's coming I at
2:15:40 least know what's coming so if you're
2:15:43 feeling like all of those negative
2:15:45 things but you're in the conversation
2:15:48 you're way far
2:15:51 ahead let's take Friday and it off
2:15:54 cuno
2:15:56 no no we're not taking Friday night off
2:15:59 just cuz you're on a cruise doesn't
2:16:02 doesn't mean you can lolly gag
2:16:05 about what what do you what do you think
2:16:07 cruises are for
2:16:09 relaxing no cruises are for teaching AI
2:16:12 to unwitting Cruise
2:16:14 members like I thought I thought we
2:16:17 discussed this at at the beginning of
2:16:18 the meeting tonight cuno what what what
2:16:21 part of there there's no crying in AI
2:16:25 what part if that did you not understand
2:16:27 you want a vacation no we're not taking
2:16:30 Friday off no
2:16:32 no
2:16:34 no but I want to walk around the the
2:16:38 market no you
2:16:40 can't you have to keep listening to this
2:16:43 while you walk around the market oh my
2:16:46 God what would your advice be tailored
2:16:48 to a senior in computer
2:16:51 science uh a a senior college
2:16:55 student or or a senior citizen in
2:16:58 computer science a senior college
2:17:00 student I would say if any of your
2:17:03 professors are not deeply obsessed with
2:17:07 AI get the [ __ ] out start start getting
2:17:09 yourself involved with every open source
2:17:12 AI project you can start building on
2:17:15 Lang chain um start watching David p on
2:17:18 YouTube get involved with his open-
2:17:20 Source cognitive architecture models
2:17:23 learn every [ __ ] thing you start
2:17:25 building chat gbt things
2:17:29 um get obsessed with all the papers that
2:17:31 are coming out just [ __ ] dive
2:17:33 headlong into anything
2:17:36 AI don't wait for your college to tell
2:17:39 you what to [ __ ]
2:17:40 do if you spend seriously if you've
2:17:44 got geek creds in any way and some Dy
2:17:49 dive headlong into AI open source world
2:17:54 and and the the open AI playground just
2:17:57 learn that [ __ ] if like I can learn this
2:18:01 [ __ ] you can learn this
2:18:03 [ __ ] and within 3 months you'll get it
2:18:07 and within 6 months you'll know what you
2:18:08 want to
2:18:09 do [ __ ] go headl long head [ __ ]
2:18:13 long all right a pinch of cayenne pepper
2:18:16 and he's not cure all that Nails you
2:18:19 Joker I you you and I have to go out for
2:18:22 whiskey sometime you're you're [ __ ]
2:18:24 awesome uh I will not be snorting
2:18:27 cayenne pepper tonight but I do
2:18:28 appreciate the
2:18:32 sentiment all right I'm not going to say
2:18:34 goodbye to people I promise hair growth
2:18:35 geek cuno good to see you enjoy your
2:18:38 cruise you do not have to show up here
2:18:39 tomorrow but we're not taking it off
2:18:42 Vicki good to see you Amelio's wife on
2:18:45 time three nights in a row I am fairly
2:18:46 certain that's a record Cosmic lover 7
2:18:49 uh thank you uh Zep Tey hem ner Tey love
2:18:55 it love the name forever hooked good to
2:18:57 see you Silver Fox uh who else we got in
2:19:00 here tonight uh
2:19:02 jrc Dr J Silver Fox Lisa Torres good to
2:19:06 see you thank you for being a subscriber
2:19:08 first rounds on me awesome man that
2:19:10 that'd be a blast actually um yeah
2:19:13 anytime anyone's in Denver hit me up
2:19:15 we'll go grab a go grab a whiskey or
2:19:18 something or whatever your beverage of
2:19:20 choice it don't have to be alcoholic I
2:19:22 do not give a [ __ ] about that stuff um
2:19:25 thank you Kyle and your regulars very
2:19:26 inspired by all the ideas learning every
2:19:28 day that's the idea perfect awesome all
2:19:30 right everybody I am out of here peace
2:19:33 out you can make money with chat gbt uh
2:19:36 but not the way you think you're going
2:19:37 to it's going to catch you by surprise
2:19:40 and it's going to be a lot more fun than
2:19:42 you thought it would be all right peace
2:19:44 out I'll see you tomorrow get ready for
2:19:46 the what we calling
2:19:49 it hold hold please the GPT for good
2:19:54 24hour GPT for good
2:20:01 challenge hang
2:20:03 on I can't I lost it [ __ ] it whatever
2:20:05 peace out I'll see you tomorrow bye
2:20:07 Tobias good to see you man