aiLL 6/16/23 - How I'm Using ChatGPT to Transform My Business

Description

Chapters

Transcript

0:02 um
0:23 hello hello it's Kyle Shannon with the
0:26 AI learning lab
0:28 welcome welcome welcome
0:30 happy Friday night and we'll wait for
0:32 some people to get in here before we
0:34 start the excitement
0:36 hello Tobias how are you
0:39 good to see you Amelia's wife good eve
0:43 indeed it's Friday
0:46 here we are here we are doing what we do
0:51 uh welcome everybody
0:53 the uh the AI learning lab is where you
0:57 have stumbled upon
0:59 I am your AI learning lab
1:02 uh let's see host
1:04 [Laughter]
1:07 um yeah this is the AI learning lab the
1:09 salon.ai is is a salon that I started
1:12 which is a community that gets together
1:16 um bi-weekly at meetings where we have
1:18 guests we have a cool guest coming this
1:20 week a guy that runs a post-production
1:22 company and he's been rolling AI into
1:25 his post-production work
1:27 and everyday AI news that's our Twitter
1:29 handle for a newsletter that I put
1:32 together with Greg mushin and this is
1:34 the AI learning lab
1:36 feel free to drop uh
1:39 drop uh whatever you call it oh I
1:41 learned so much last time from you thank
1:42 you for your time thank you Tobias I
1:44 appreciate the uh appreciate the kind
1:46 words if you have questions about AI pop
1:49 them in the comments below if you're
1:51 sitting there thinking who the hell is
1:53 this guy and what the hell does he know
1:54 about AI not a lot I'm learning with you
1:57 I'm like as I'm learning I'm sharing
2:01 um
2:02 I am an entrepreneur for of many many
2:05 years I still don't have dual camera
2:06 they did not give it back to me no no so
2:09 I have to flip back and forth
2:11 like this
2:13 like a plebe
2:15 I know I know I'm right there with you
2:17 wow looks like a bunch of people just
2:19 joined so hello my name is Kyle Shannon
2:22 this is the AI learning lab uh we talk
2:25 about things I AI here uh if you have
2:28 any questions comments thoughts about AI
2:31 pop them below I do my best to answer
2:33 everyone I can I'm an entrepreneur
2:36 I've started I don't know 12 or 15
2:38 companies I'm a co-founder and CEO of a
2:40 company called storyvine which is a
2:42 automated video platform we've been
2:44 around for 11 years or so and uh
2:48 interestingly enough well these days is
2:50 uh we automate video production but we
2:53 have not used AI to automate the video
2:55 production
2:57 um we're starting to use AI now to do
2:59 things like do automatic automatic
3:01 transcriptions
3:02 and create omnichannel content off of
3:05 video so lots and lots of things like
3:07 that
3:08 and
3:10 um and what else let me let me show you
3:12 some things I don't have dual cameras
3:14 which is the bane of my existence these
3:17 days but if that's my only problem we're
3:18 doing all right so so these URLs here if
3:22 you've not played with chat gbt that top
3:25 URL chat.openai.com that's the official
3:28 chat GPT website go there and start
3:30 playing while I ramble on and then
3:33 bing.com is also chat EBT if you go to
3:36 the chat button there it's gpt4 it's
3:38 connected to the internet and it's free
3:40 um
3:41 I don't think it's quite as good as the
3:43 official chat GPT website but it's still
3:45 there and it's free and then poe.com has
3:47 lots of options as well a lot of
3:49 different models you can play with
3:51 prompts.chat is super cool that will
3:53 teach you about prompting in general if
3:55 you don't quite know what's going on
3:56 with that and then future PD is just a
3:59 big old list of tools I'm not affiliated
4:00 with any of those I just think they're
4:01 super cool
4:03 um You probably don't have the lights on
4:05 quite as bright you babe you may be more
4:09 comfortable
4:10 um I don't know they're still pretty
4:12 bright here do I look darker
4:14 interesting I don't know possible let me
4:17 see
4:18 let me see am I looking at them yeah you
4:20 know let's see
4:22 no okay A little brighter
4:26 anyway welcome to the AI learning lab
4:29 um one of the things I like to kick off
4:31 if you've got questions about AI pop
4:33 them in the in the comments below and
4:34 I'll do my best to answer them but one
4:36 of the things that
4:38 um I think it's important why why I
4:40 started this Channel and why I'm doing
4:42 this
4:43 um I was
4:44 um
4:45 I was lucky enough to to be in New York
4:48 City uh when the World Wide Web kind of
4:50 came on the scene and started one of the
4:52 early digital agencies and you know kind
4:54 of had some big epiphanies around the
4:56 world wide web and what it was going to
4:57 do with the world and uh started this
5:00 company called agency.com and what's
5:02 happening right now with generative AI
5:05 um there are some massive parallels um
5:08 to what was going on then I think this
5:10 is a big deal like that was a big deal
5:12 but I think that this AI stuff is coming
5:15 faster and more profoundly so
5:19 um let's CC say Greetings Kyle bourbon
5:24 happening I might do bourbon tonight
5:26 although I just ate dinner I'm a little
5:28 full
5:29 I'm a little full I've got a an ice cold
5:32 aqua
5:35 agua
5:37 I guess some country might say Aqua agua
5:41 but I might do I might do a little
5:43 bourbon later
5:44 saying you can darken the lights sir ah
5:47 I see I'm a little too bright
5:49 all right there I'm a little more more
5:52 colorful wait I can turn this one off if
5:54 I turn this one off and I'll be more
5:56 dramatic
5:57 drama the drama of AI is very dramatic
6:02 all right very lovely can I day trade
6:06 with AI
6:07 well you can day trade with day trading
6:09 platforms you can use AI to help you
6:11 come up with strategies and Bots and
6:14 stop-loss thingies and
6:18 banned
6:19 engorgement strategies I don't know [ __ ]
6:22 about day trading but yeah you can I
6:24 mean
6:25 it can do a lot if you haven't played
6:27 with chat gbt go play with it will
6:29 companies be able to tell when their
6:31 employees use AI nope
6:35 and and I guess my question will be why
6:38 would they care so so right now chat GPT
6:41 is kind of this thing off to the side
6:42 and it's kind of been it's a little
6:44 demonized like what if someone uses
6:45 Chachi pite it's a very similar concern
6:49 when the calculator came out
6:51 all the math professors were like oh my
6:52 God people won't be able to add what do
6:54 we do let's protest
6:56 um so in a very similar place right now
6:58 here's something that's happening you
7:00 know similar to how
7:02 calculators and spreadsheets uh you know
7:05 just became tools that we use and people
7:08 still learn math and know how to do math
7:10 same kind of thing with these tools
7:12 Microsoft and Google both are taking
7:14 their large language models chat GPT and
7:17 Bard uh respectively and they're rolling
7:20 them into Office 365 and Google Sheets
7:24 and docs and the office suite for Google
7:28 um these tools are going to be
7:29 everywhere they're going to be
7:30 ubiquitous and it's going to happen you
7:32 know probably pretty soon Microsoft's
7:33 are already announced theirs Google's
7:35 already announced theirs
7:37 um
7:38 Microsoft is enjoying
7:41 um being a thorn on the side of Google
7:43 for the past six months I have a feeling
7:44 they're going to be pretty aggressive
7:45 about it once it's ready they'll
7:47 probably they'll probably push it out
7:50 um I know I know I said that kind of
7:51 glibly like does it matter if if people
7:54 use AI
7:56 um what matters if is if they use Ai and
7:59 they're not thoughtful about it right if
8:02 they just say you know here's a letter
8:03 I'm going to send it without reading it
8:05 um there was a lawyer two weeks ago that
8:08 wrote a legal brief with chat GPT and
8:10 handed it to a judge
8:12 and the judge went ahead and checked the
8:14 citations of the legal brief and they
8:16 were all fake
8:17 and the judge said to the lawyer uh dum
8:20 dum come here
8:23 these are all fake
8:25 um so so if you do bad work you can do
8:29 bad work faster
8:31 with chat EBT you'll give me money like
8:34 hey and you can do shitty work faster
8:36 with chat DBT
8:38 um but you could also do really good
8:40 work faster with chat GPT so if you have
8:43 a tool that increases your employees
8:45 productivity by 30 or 40 percent and
8:47 it's the most powerful tool you know
8:50 known to the history of mankind why
8:53 would you not want your employees to use
8:54 that is a question I have and listen
8:56 there are reasons right now why you
8:58 wouldn't want to use it it's new it's
9:01 early it hallucinates it's not perfect
9:04 you know it can be a crutch all that
9:07 sort of stuff is true but this is a tool
9:11 that's going to be quite ubiquitous
9:13 in my humble opinion but Kyle you're not
9:17 humble oh that's rough that hurts
9:21 it's true true true true
9:24 I'm gonna give you that one
9:26 gonna give you that one but uh
9:30 all right
9:31 uh
9:33 uh AI makes decisions web pages can't
9:36 well web pages that are programmed to
9:38 make decisions Ken but uh yeah AI is uh
9:42 AI can do some [ __ ]
9:43 it's not reasoning but it can certainly
9:46 help with that do you know the muffin
9:48 man the muffin man
9:51 Italian say Aqua ah
9:55 I I don't know enough Italian to even
9:58 make a stupid saying aqua
10:01 oh Jesus
10:03 it's like I'm worldly without the world
10:10 Mercedes-Benz too yeah yeah if you
10:13 didn't hear that Amelia's wife
10:15 tuned me in on this one yesterday that
10:17 uh Mercedes-Benz is rolling Chachi BT
10:20 inside all the mercy Doodles so you can
10:22 just talk to your car hey treacherypt
10:25 can you teach me how to make money watch
10:27 out
10:28 I need to make a payment on a share of
10:30 Mercedes
10:32 um
10:33 but yeah like these tools are going to
10:35 be
10:36 they're going to become the interface
10:38 for
10:40 all the complicated systems we use right
10:42 imagine the things you know you know the
10:45 you know those car entertainment systems
10:47 that are like more complicated than any
10:48 computer you've ever used and you'll you
10:50 can only figure out like two percent of
10:52 them because they're so damn complicated
10:53 and you almost crash the car every time
10:55 you try to change the screen well
10:57 imagine having a system smart enough
10:59 that you could just ask it to do [ __ ]
11:00 and it would do it
11:02 including like booking you a restaurant
11:03 for the night you know setting your map
11:06 to take you there sending flowers to
11:09 your wife before dinner all that sort of
11:11 stuff right that's that's the kind of
11:13 thing that's coming so these are going
11:14 to be these are going to be everywhere
11:17 oh boy someone just spoke Italian
11:19 Emilio's wife I can't read Italian but
11:22 thank you I bet you use chat to
11:24 translate that didn't you
11:26 not saying I don't like the idea of AI
11:28 but what do you think about the impact
11:30 of an AI in a society where
11:34 wait what clone X
11:38 where's the rest of your sentence you
11:40 didn't finish your sentence
11:43 what do I think about the impact of AI
11:46 in a society where
11:50 what's the second half of the sentence
11:52 clonax I don't know how to answer you
11:58 um
12:02 welcome everyone oh we're we're a small
12:06 intimate group on Friday night many
12:08 people thought better of this they're
12:10 like yeah this doesn't seem like it's
12:12 going anywhere I'm out of here I'm gonna
12:14 go get a bourbon
12:17 um dude I know when Taffy will go live I
12:19 don't but it's super cool
12:20 um plumbers will have have work
12:22 Radiologists will go home with AI
12:24 although you know they've been calling
12:26 Collins Radiologists for about 10 years
12:28 now
12:30 I think anywhere where you've got
12:31 people's lives on the line I think
12:33 insurance companies are gonna are gonna
12:36 um
12:37 ensure that human beings are in the loop
12:39 for some time
12:44 if you've got questions about AI pop
12:46 them in the comments
12:49 oh clonex your your second half of your
12:52 comment was right next to the first one
12:54 I was looking for it somewhere below
12:56 so so the question is what do I think
12:59 about the impact of AI in a society
13:01 where profit is more important than
13:03 human lives well that is a mighty
13:05 cynical point of view don't necessarily
13:07 disagree with it
13:09 um
13:10 so so I think I think it goes a couple
13:13 of directions
13:15 I think companies that are owned by
13:18 private Equity firms and hedge funds and
13:20 things like that that are only looking
13:21 at the bottom line
13:24 um they're gonna slash their staffs
13:26 right they're going to say oh 30 to 40
13:28 percent productivity increases 30 to 40
13:31 percent they'll probably say okay we now
13:33 decree 50 cuts across the board deal
13:37 with it let AI make up the rest so there
13:39 will be some of that
13:41 um
13:44 a lot of those displaced people also
13:46 have access to these tools
13:49 so some percentage percentage of those
13:52 displaced people are going to use AI to
13:54 create companies to go to compete with
13:56 the companies that just laid them off so
13:58 that's going to be a thing
14:00 some of those people are going to
14:02 completely ReDiscover themselves and
14:03 take you know completely new career
14:05 trajectories or you know contribute to
14:08 society in other ways so there's there's
14:10 that some of those people are going to
14:12 just be displaced and it's going to be a
14:14 [ __ ] [ __ ] show and that's going to be
14:16 horrible
14:18 um
14:18 if it's at a big enough scale then the
14:21 government's gonna have to deal with it
14:22 kind of like they'd you know dealt with
14:23 kobit where here's a check or here's
14:25 Universal basic income or here's the
14:27 something
14:28 um I don't think this is going to be
14:29 long term because AI like like the
14:32 Industrial Revolution
14:34 um got rid of a bunch of jobs but it
14:35 also created a bunch of new jobs this is
14:37 going to do something very very similar
14:41 so so that's that's the the money
14:44 grubbing pigs the capitalist pigs are
14:47 going to do that I'll give you that I
14:48 don't think it's a hundred percent
14:49 blanket statements like you know
14:52 um everybody you know uh is for profits
14:56 over people I don't think that's a
14:57 generally true statement
15:00 um
15:02 some companies will say oh wait a minute
15:04 we could take on 30 or 40 percent more
15:07 clients or we could deliver 30 to 40
15:10 percent higher quality work or we could
15:12 cut our prices to sell more
15:16 um or we could have our people that
15:18 right now 30 to 40 40 of the work that
15:21 our people do is low level soul-crushing
15:24 work so we're going to use AI to upskill
15:27 them to do higher level solar energy
15:30 enriching work and we're going to let
15:32 the AI do the shitty work that no one
15:34 wants to do anyway
15:36 and the the quality of work and the
15:40 quality of work experience for those
15:41 employees is going to go up it's going
15:43 to get better so that's a possibility as
15:45 well
15:47 um
15:49 I mean these are just tools right this
15:52 is a tool and this is a tool that that a
15:55 boss I'm a CEO of a company right now I
15:57 could look at this and I could say okay
15:59 like I don't need these stinking people
16:00 I have a small company those are my
16:02 family
16:03 I want to do everything I can to support
16:05 them and I want to make sure that our
16:06 clients are getting great work too so
16:09 um yeah I so so I don't think it's quite
16:12 as black and white as as you positioned
16:16 it there
16:17 um but um
16:19 but yeah that's that's my thoughts on it
16:23 all right let's get to some other
16:25 questions shall we
16:27 what do I think of AI certifications if
16:30 you're going into machine learning and
16:33 um
16:34 the computer science side of things into
16:36 the into the infrastructure side or the
16:38 the
16:39 development side of AI sure
16:42 if you're talking about certifications
16:44 for like chat GPT
16:46 you give me money with chat gbt buy my
16:48 certification now I think those are all
16:50 [ __ ] because
16:52 things are moving too fast and nobody
16:54 knows anything like like we don't know
16:56 you know it's always talk about you know
16:58 prompt engineering is going to be the
16:59 new job well the the tools are starting
17:01 to prompt themselves and the tools are
17:03 going to get better and better at
17:04 prompting themselves so I think
17:05 understanding
17:08 how to do prompting is really important
17:10 but I think the specifics of how to do
17:11 prompting today is going to be
17:13 irrelevant three months from now so I
17:15 would be very careful of anyone
17:17 positioning themselves as an expert
17:19 in technology that's only been around
17:21 for six months that is literally
17:23 changing week by week
17:25 um if you're if you're going into the
17:27 computer science side of things then of
17:29 course go to Stanford and go to or go to
17:30 MIT or go do the free courses at
17:32 Stanford MIT
17:34 um I don't think you need to go to
17:36 college anymore because we have tools
17:37 that can do that [ __ ] for us now
17:39 plumbers will work I agree with that I
17:42 think the trades the value of the trades
17:44 are going to go up
17:47 um they did finish the sentence Tick
17:50 Tock is just trash Tick Tock took away
17:52 my dual camera feature that Dum Dums
17:55 there used to be a little thing here
17:56 where I could show you
17:58 this on a regular basis that would just
18:01 be sitting there and then I could
18:02 occasionally go oh there's new people
18:04 here go to those sites and do that [ __ ]
18:06 and I could point to stuff but I can't
18:07 point now
18:08 why because Tick Tock hates me
18:14 how do I use AI to learn code for data
18:17 analysis well
18:19 so you can just go to Chachi PT and
18:23 start
18:25 learning it like have it teach you what
18:28 it can do so let's go so right now I'm
18:31 on just regular old chat EBT so this is
18:34 GPT 3.5
18:37 so I'm going to say
18:38 um I want
18:42 to
18:43 learn
18:45 data analysis where do I begin question
18:52 mark That's great data analysis is a
18:55 valuable skill
18:56 understand the basic basics of
18:59 Statistics learn a programming language
19:01 like Python and R are popular languages
19:04 pandas numpy Matlab explore data
19:07 visualization get hands-on experience
19:09 experience with real data okay so I'm
19:13 going to say
19:14 uh let's see can you give me some fake
19:20 data
19:21 to learn on
19:24 certainly
19:26 here's a sample of fake data
19:30 um
19:31 what might I do with this as a first
19:36 step so
19:39 you don't need to watch me keep making
19:41 up stupid things because I know nothing
19:43 about data analysis but
19:46 you can do if if you've got a curiosity
19:49 in this and even a small amount of
19:51 competency
19:53 in data analysis or even an interest in
19:55 it you can run down some significant
19:58 rabbit holes you know I can say make me
20:01 a table
20:03 um of I don't know eight Columns of
20:08 artificial
20:10 data
20:12 um
20:14 that would be interesting to analyze
20:23 so it made me a python a little python
20:27 script there
20:28 um I want to say
20:30 can you put this in table format
20:36 and it will do that
20:43 um so so one thing you can do is is you
20:47 know think of chat GPT you know less as
20:50 oh that's a thing people write songs and
20:52 poems with and more is like the smartest
20:55 co-worker you've ever worked with and
20:58 just have that co-worker start you know
21:00 tutoring you
21:01 there's also go watch
21:04 um go to YouTube and search for code
21:08 interpreter for chat EBT
21:11 um so what code interpreter is is it's a
21:14 full-on data analysis engine
21:17 that will be built into chat gbt it's
21:19 not out yet I don't have access to it I
21:21 don't think let me just go
21:23 look and make sure that he didn't give
21:24 it to me overnight because they do do
21:26 that sometimes it's just turn [ __ ] on no
21:28 it's not there
21:30 um so a code interpreter is is in your
21:32 little chat box of chat CPT there's a
21:34 little upload button
21:36 and you can upload data you can upload
21:38 like 100 megabyte Excel file full of
21:40 data and then you could say to it what's
21:43 interesting about this data and it will
21:44 give you a whole data analysis report
21:46 you can say find any anomalies in the
21:48 data and it will find them you can say
21:51 um create these kind of uh charts that
21:54 compare Trend a to Trend B and it will
21:57 write the python code generate the chart
21:59 and hand it to you that's coming
22:01 like that's coming soon like there's
22:03 YouTube videos code interpreter for chat
22:05 GPT just go to YouTube and watch them
22:07 and then when you pick your brains up
22:10 that dripped out of your ears as as you
22:13 were watching the video then go to chat
22:15 gbt and say teach me how to do this
22:19 because it's crazy what's coming is
22:23 crazy call centers will have exact same
22:26 accent as the region of the person
22:28 calling do the voice Wicked yeah I think
22:30 that's going to happen
22:32 pretty quickly I like my optimistic view
22:35 of call centers is that the call centers
22:38 that are these really frustrating uh
22:41 experiences right now where you can't
22:42 understand the people and they're
22:44 clueless and they're they have no power
22:45 to do anything
22:47 are going to be replaced with uh
22:50 synthetic voices that know everything
22:53 and then you take some of that money
22:56 that you spend on a call center and you
22:58 upskill your employees
23:00 um to be high level customer service
23:02 people so you have you have essentially
23:04 chat Bots with synthesized voice that
23:07 people talk to
23:08 but the chat Bots are good enough to
23:10 understand that you know if this is if
23:12 this is a normal problem I'll solve it
23:14 and they'll have some authority to solve
23:17 it but if this is a problem that's out
23:19 of the ordinary which is that's always
23:20 the one when you when you know you have
23:22 something out of the ordinary and you
23:23 need to get to a person soon that's the
23:25 one where you're three hours on hold
23:27 with six different people like that's
23:29 just maddening so imagine you're on with
23:32 a chat bot it quickly figures out that
23:34 you have an anomaly in customer service
23:36 and it escalates you to a person that
23:38 can actually help you like that could be
23:40 a future that's coming so I totally
23:42 agree I think I think the customer
23:44 support business is going to transform
23:46 completely completely
23:50 that can be used more on product what
23:55 uh uh
23:56 [Music]
23:58 I don't know what that what that was in
24:00 uh
24:01 response to
24:04 let's see
24:05 you could put a background up silly
24:08 I don't know what you're talking about
24:11 what do I think of prompt libraries
24:15 um I think they're interesting I don't
24:17 know
24:21 I just I I just heard uh that IBM uh
24:25 just created a a generative AI center of
24:29 excellence and they have a thousand
24:31 people at IBM writing prompts
24:34 so they're creating a massive prompt
24:35 Library
24:37 um I don't know how necessary they're
24:39 gonna be
24:40 um let me show you something I'll show
24:42 you something that that is a
24:44 um
24:47 that for me kind of points to to where
24:50 this is headed so there's a
24:52 there's a plug-in called prompt perfect
24:55 so I'm going to go to plugins I'm on
24:57 gpt4 I'm gonna go to plugins
24:59 and I'm going to go turn off these other
25:02 things I'm going to turn on prompt
25:03 perfect
25:06 I'm going to turn off trip.com what else
25:08 did I have turned on so what I'm doing
25:11 right now with with
25:13 um with prompts you can install prompts
25:16 so what I see in this list here are
25:17 things that I've installed and then in
25:19 any given session you can activate up to
25:21 three prompts so I'm just turning on one
25:23 called prompt perfect and the way it
25:26 works is if I type the word perfect in
25:28 front of a prompt perfect
25:32 and then I type in a crappy prompt like
25:35 um
25:36 let's see
25:38 right uh let's see
25:41 um write me
25:43 a
25:46 marketing plan for my
25:52 gardening business
25:54 that's about as bad a prompt as you can
25:56 write
25:58 but because I put perfect in front of it
26:00 it's going to fire up that plug-in
26:03 and then it is going to what did it do
26:06 it rewrote it rewrote my crap crappy
26:09 prompt please develop a comprehensive
26:12 strategy for my gardening business that
26:14 includes specific tactics and actions to
26:17 promote our services attract new
26:19 customers increase Revenue please
26:21 consider our target audience competitive
26:23 landscape budget unique selling points
26:25 of our business the plan should outline
26:27 clear goals timelines metrics and for
26:30 measuring success
26:31 and so then it writes me
26:34 a big ass marketing plan on a much
26:36 better prompt so
26:39 the the the whole idea of creating a a
26:42 prompt Library based on
26:45 today's capabilities just just feels to
26:48 me like some piece of that is going to
26:51 be
26:52 made obsolete probably sooner than later
26:56 um I think what's what I here's why I
26:58 think there's value in creating a prompt
27:00 library is that in creating a prompt
27:02 Library you'll you'll really explore all
27:05 the different ways you can prompt I
27:06 think the the act of learning
27:09 how to think of your native language as
27:12 a programming language that's a skill
27:14 that is going to remain important but I
27:16 think the actual tactics of how we
27:18 prompt today I think that gets much more
27:20 simple as as the tools I don't know take
27:23 over that that task
27:30 that was very helpful great I don't know
27:33 do I have do I have to have python
27:37 no so so so here's a cool thing so this
27:41 was this was my my big Epiphany with
27:43 with chatgpt
27:46 um you know let me show you this
27:48 the the thing that made me realize holy
27:50 [ __ ] the world just changed so so I was
27:53 like you where I kind of knew what I
27:55 wanted to do but I didn't really know
27:56 python so
27:58 I said right right
28:01 me a python
28:04 program that uh I don't know whatever uh
28:08 takes and takes
28:12 inputs
28:14 by asking
28:17 three questions
28:19 four questions and so it right it writes
28:22 that thing and then
28:24 it even gave me an example answer so
28:27 here here's the code and I say
28:29 um
28:30 I don't
28:32 know how to run python
28:36 can I run
28:40 it on a website question mark yes you
28:45 can run python code on websites there's
28:46 some websites replit python anywhere
28:49 trinket.io online right
28:52 so so this is what I ended up doing and
28:55 I I was just kind of [ __ ] around with
28:58 this thing and within 90 minutes I had
29:00 written a python application that was
29:02 running in a website I didn't have to
29:04 install anything I had to install a
29:06 library on that website but it taught me
29:08 how to do that
29:10 um and and I really quickly got up to
29:13 speed so so depending on what you're
29:15 trying to accomplish just have uh chat
29:18 GPT you just keep it like there's
29:22 there's there's literally no shame in
29:24 chat GPT like it doesn't care if you
29:26 don't know anything
29:28 um one of the things that I've done with
29:29 it is you can if you've heard of GitHub
29:32 so GitHub is is where you know Geeks and
29:35 engineers and programmers go to store
29:37 their code and share their code and do
29:39 all the Kodi Cody [ __ ] right if you
29:41 don't know coding you go to GitHub and
29:43 it is a confusing nightmare of Hell want
29:45 to see
29:46 watch
29:48 this will make your head hurt if you if
29:51 you thought you made a bad decision
29:52 hanging out with me on a Friday night
29:54 now let's go to GitHub and check out the
29:56 interface here GitHub
30:00 oh wait let me go
30:04 github.com slash
30:08 okay here's one here's a project okay so
30:11 you get go to GitHub
30:13 there's all this there's there's all
30:16 these links there's folders there's like
30:18 dot ESL something ignore major overhaul
30:22 and flow standardization number one one
30:25 three and these are all hyperlinks three
30:27 weeks ago and then up here there's
30:29 notifications and forks and stars and
30:32 there's insights and security and
30:33 projects and like you don't know these
30:35 are all related to this one project but
30:37 if you don't know what you're doing here
30:39 it's it's completely overwhelming and if
30:42 you know a little bit about GitHub you
30:44 know that there's one file here this one
30:47 the readme file that is kind of the key
30:49 to the kingdom you wouldn't know that if
30:51 you didn't know this right so if you go
30:54 to the readme file it will tell you
30:56 basically how to install this and what
30:59 it does and it tells you and it gives
31:00 you some code and then there's all these
31:02 instructions and if you know what you're
31:04 doing
31:05 this makes it really easy to leverage
31:08 this code right
31:10 but here's what you can do you can just
31:12 copy this readme file copy all the the
31:15 all of it
31:17 paste it into chat gbt and then tell
31:21 chat GPT what system you're on and
31:23 please make you a step-by-step tutorial
31:25 for how to install that GitHub
31:27 repository and do not move on to the
31:30 next step until you've said okay to the
31:32 previous step and so it'll give you the
31:35 first step and if you don't understand
31:36 it you're like I don't understand and
31:38 it'll give you a little tutorial of that
31:39 step and then you do that step and when
31:41 you're done with it you say okay and it
31:43 moves on to the next step so
31:46 so it's a kind of thing that if you're
31:48 completely clueless about this
31:49 technology stuff but you're curious
31:51 enough to try to figure some [ __ ] out if
31:53 you if you can kind of Tinker with
31:55 computer [ __ ]
31:56 um chat EPT can absolutely help you
31:58 solve almost anything
31:59 and if you're good with code then you're
32:02 going to be able to just knock this [ __ ]
32:03 out you can go do some stuff I am not I
32:07 am I'm a tinkerer like so when I went to
32:10 install stable diffusion which is one of
32:12 the image generation tools I failed nine
32:14 times to get uh this thing called
32:19 automatic 11 11 which is a a web
32:21 interface for stable diffusion I failed
32:24 nine times trying to get it to run on
32:26 Google collab before I finally got it
32:29 and it was over a couple of months like
32:30 like I would try and spend a couple
32:33 hours it would fail I'd feel full of
32:36 Shame like I'm a dumb dumb I don't know
32:39 how to do anything on computers anymore
32:41 I'm just old
32:43 that part's true and then uh I'd go away
32:47 for a few days and I'd come back and
32:48 they'd fail again and then and then I
32:50 finally got it so
32:52 um all of this stuff is depends on how
32:54 how hungry you are one of the other
32:55 things going on right now
32:57 we're very very early in this inflection
33:00 point so Chachi PT kind of marks this
33:03 this new moment where
33:06 these large language models are made
33:08 available to like the rest of humanity
33:10 previously you know machine learning and
33:12 that was kind of researchers and data
33:14 scientists and and things like that chat
33:17 GPT kind of marks this moment in time
33:18 where in six weeks a hundred million
33:21 people started using chat EBT 100
33:23 million six weeks it took the World Wide
33:26 Web six years to get to that many users
33:29 um so so
33:31 um
33:33 so so there's a bunch of people right
33:36 now using these tools that are
33:37 incredibly powerful
33:40 um that don't necessarily know you know
33:42 all these complicated ways to stitch
33:44 these tools together and that's okay
33:46 we're very early it's only it's only
33:48 been six months since since this stuff
33:49 came out
33:51 um give it a year and all of the tools
33:53 we we we're interfacing with right now
33:56 will drift to the background and people
33:58 come up with really slick interfaces on
34:00 front of them like Mercedes-Benz in
34:03 their cars
34:04 I have an idea about a company that uses
34:06 stories from humans to educate AI well
34:08 that's kind of interesting my only
34:10 question on something like that is who
34:12 would pay for it and what what's the
34:13 value where's the where's the value
34:14 there and who who's going to actually do
34:16 the paying for that but it's but it's
34:18 interesting because I think that there's
34:20 there is some fear that
34:22 as the developer Community uses AI to
34:25 generate data to train other AI there's
34:28 a thing called Model collapse where
34:31 where the the models just collapse
34:33 because uh there's sort of this point of
34:35 diminishing returns with synthesized
34:37 data so there might be something there
34:40 um but I just don't know if it's a big
34:41 enough problem yet um or who would pay
34:44 for it but if you can solve that that
34:45 could be interesting I think that one is
34:48 interesting I think a big one in two
34:52 years is going to be authenticity and
34:54 verification
34:55 I think authenticity the the ability to
34:58 authenticate is this article reel is
35:01 this person real is this live stream
35:03 with Kyle real
35:06 um that's going to get really important
35:08 as as synthesized you know people in
35:11 data gets way more sophisticated wait
35:13 for the 2024 presidential election in
35:16 America people that's going to be a
35:18 [ __ ] party
35:21 what do you mean Joe Biden's French kiss
35:24 in the gorilla at the Cincinnati Zoo
35:26 is he there's video
35:29 really
35:30 have you seen it
35:32 two days before an election you watch
35:34 it's gonna be insane
35:40 it's gonna be crazy gonna be crazy but
35:42 there's gonna be whole new industries
35:44 that pop up because of that stuff
35:46 all right let's see I'm trying to find
35:48 my place in the questions this is one of
35:51 my hobbies in these lives is finding
35:53 finding finding the questions I'm also
35:55 wearing again a light shirt that it's
35:57 impossible to read text in front of so I
36:00 just don't learn my lessons
36:02 he's nice but he's he seems a little
36:05 slow he seems a little thick in the head
36:08 he said he can't he can't read it he
36:10 can't read the text and he gets lost all
36:13 the time it doesn't
36:15 he seems clever but he seems a little
36:16 slow
36:19 I don't know
36:20 all right
36:24 what is the most effective way to use AI
36:27 that you recommend
36:29 um let's see Wilbert
36:33 I I mean it depends so so here's the
36:35 most effective way to use AI that I know
36:37 is start with a problem don't start with
36:40 the tool I I think so look this is let
36:44 me show you what I find very
36:46 intimidating and I have 80d so I'm
36:48 generally good at coming up with [ __ ] to
36:50 do on the Fly
36:54 that right there
36:57 is intimidating it just looks like a
36:59 Google search box and it can do almost
37:02 [ __ ] anything
37:04 and so I sit here and I'm like
37:06 um uh make me a poem
37:11 all right there's a poem uh
37:14 uh then make it a song right so so
37:19 like like when you're just faced with a
37:21 tool that can do almost anything and you
37:24 know like
37:25 um write me a computer
37:31 program
37:33 that plays that song
37:38 all right
37:39 there you go
37:41 put it in a minor key
37:47 so
37:50 like like
37:52 these are just parlor tricks and it's
37:54 like what do you do with it so so the
37:56 the thing that I've found that makes the
37:58 biggest difference is figure out
38:00 something you're trying to do in your
38:01 life it can be really small it could be
38:06 um
38:06 I don't know when to plant the tomatoes
38:08 in my region
38:10 go to church EBT and have it give you a
38:13 planting schedule and then have it tell
38:16 you natural ways to fertilize tomatoes
38:19 and then say what are some awesome
38:21 tomato recipes and when am I going to
38:23 have a lot of tomatoes to eat given my
38:26 region in average picking time like like
38:29 the minute you give yourself a problem
38:31 then you can say oh let me go see if I
38:33 can solve that with chat GPT or if I
38:35 need an image maybe I go to Mid Journey
38:36 or one of these other tools to play with
38:38 it
38:39 so that's kind of the most effective way
38:42 to use it is don't start with the tool
38:44 that said if you've never played with
38:46 the tool go play with the tool right go
38:49 start understand start understanding the
38:53 incredible depth of what that thing can
38:56 do that that top URL there
38:58 chat.openai.com
39:00 is the official chat CPT website just go
39:02 there right now if you haven't gone
39:04 there and just start playing you can't
39:06 break it it can do almost anything
39:10 um
39:11 you know have fun with it do me a favor
39:13 if you could tap the screen I know a
39:15 bunch of you are doing it now but tap
39:16 the screen so we get some hearts and if
39:18 you would share this live so other
39:20 people see it because uh Tick Tock has
39:22 decided that Friday night they're not
39:24 gonna they're not gonna make this live
39:26 available to anyone because
39:28 he doesn't get a dual camera and he
39:30 doesn't get an audience
39:31 [Laughter]
39:33 because he's boring and he doesn't know
39:35 how to wear shirts that make his text
39:37 readable until your competition starts
39:40 using AI instead of humans you will have
39:43 to wait until your competition starts
39:45 using AI instead of humans you'll have
39:47 to do likewise well I think that's right
39:49 I mean
39:50 that that's why I think this stuff is
39:52 coming fast it's not it's not so much
39:54 that every company has to do it it's
39:56 that
39:58 the companies that don't do it are going
40:00 to get smacked upside the [ __ ] head
40:02 by companies that do
40:03 they're gonna be like wait how did they
40:05 deliver that so fat wait they're
40:08 charging how much to deliver that
40:10 and they're delivering it twice as fast
40:14 we can't do that how are they doing oh
40:16 it's AI oh Jesus Christ Marge We need to
40:20 do AI
40:22 does anybody know AI I don't know Sally
40:24 is chat GPT wants Sally your VP of AI
40:27 let's figure this [ __ ] out
40:30 like that conversation's probably
40:32 happening you know right now
40:35 across the world
40:37 because somebody's got their ass handed
40:39 to them because someone else was using
40:41 Ai and if it's not happening now a year
40:43 from now it's going to be happening so
40:45 the like the the purpose of this channel
40:47 is to get as many people as possible
40:50 trying right giving it a shot
40:55 um
40:56 all right what do I think a prompt
40:57 libraries talked about that one
41:00 you could put a background up silly me
41:03 possible why would I put a background up
41:05 when I have this loveliness
41:07 [Laughter]
41:09 uh
41:10 uh behind on my questions I've answered
41:12 all these all right the website
41:13 addresses oh you could put them on a
41:16 background oh
41:18 and with the cleverness oh
41:22 hang on wait do I have it
41:27 um
41:30 I'll have to
41:35 that's a good idea
41:37 that's good
41:38 that's see this is this is
41:41 I mean Kyle's nice but he seems to need
41:44 remedial help every once in a while
41:48 that's a good good idea I like it I'm
41:51 gonna have to turn this into hang on
41:54 here I can you can watch me as I as I do
41:57 my my uh my live graphic uh support
42:02 um
42:03 what's it called
42:05 um
42:06 housekeeping
42:08 so we're gonna go export
42:11 quick export is PNG
42:15 TT live image we'll go to desktop we'll
42:19 save it there we're gonna go to
42:23 uh
42:24 we're gonna go to Gmail
42:29 and I'm gonna send it to myself because
42:32 I could
42:35 do it in a more sophisticated way but it
42:38 takes too long so my name's Kyle and
42:41 then
42:42 I'm gonna go ttbg
42:46 and then we're going to attach
42:49 that ttbg
42:52 that we put right there
42:54 I'm gonna send it to ourselves all right
42:57 now it's now it's at least on my machine
42:58 all right that's a really good idea and
43:00 thank you very much
43:02 thanks thanks for helping me uh uh you
43:06 know do the basics
43:09 I'll get to that as soon as I shut the
43:12 the live down I'll do that on the next
43:14 lives and then you know what's gonna
43:15 happen the minute I start putting the
43:17 that background behind me then I'm gonna
43:19 give my live camera back and then I'm
43:20 not gonna know I'm not gonna have to set
43:22 up for it it's gonna be a mess
43:24 all right I feel like plugins aren't
43:26 that impressive what am I missing
43:29 you're missing the fact that they're not
43:30 that impressive oh wait that's not
43:32 missing anything they're not they're
43:34 they're kind of janky and shitty and
43:37 most of them so so a lot of them
43:40 so there was one that I was using called
43:41 link reader which all of a sudden just
43:43 stopped working and and here's my my
43:45 thesis on it is this it was like some
43:48 dude that was like oh I'll know I'll
43:50 make a plug-in that can read a um
43:52 someone's hyperlink and I'll put that
43:54 out there and they probably just built
43:56 it on some sort of API that charges them
43:58 a nickel every time someone used it and
44:00 then a bunch of people used it and got a
44:02 bill for like 150 bucks he's like I'm
44:04 not that's I'm not paying that for these
44:07 people so the the plugins are really
44:09 easy to build right now and there's not
44:12 really a ton of vetting and it's not
44:14 clear how they work and
44:17 um so so here's how I'm looking at
44:19 plugins right now if you find one that's
44:21 actually useful like I think I think
44:23 prompt perfect is actually useful I
44:26 think the PDF reader things are useful
44:29 um if you find things that are used like
44:31 wolf from alpha where you can you know
44:33 ask it a sophisticated math problem and
44:35 it fires off to wool from alpha and
44:37 solves it and brings you back the answer
44:38 cool
44:40 so so right now right now they're just
44:43 you know it's kind of like when the I
44:45 don't know if you remember this but when
44:46 the when the iPhone first came out they
44:48 didn't call them apps they called them
44:49 widgets and widgets were just basically
44:51 shitty little websites they were like
44:53 icons that that took you to a website
44:55 that couldn't do anything
44:58 um and they were all just super simple
44:59 and people were like what's the point
45:01 why would I need a widget I don't need a
45:03 widget for this because I got a website
45:05 that does that and people are all
45:07 bitching and whining and then it turned
45:09 into the App Store and it got really
45:10 sophisticated and and all of a sudden it
45:12 changed the way we worked and so I think
45:14 I think plugins are going to be the same
45:16 way
45:17 open AI also just
45:20 um announced or released
45:23 a big update to their apis so for
45:26 programmers who use it there's a thing
45:28 that they can do do now where they can
45:30 call functions which is basically like
45:32 calling a plug-in but you can have chat
45:34 you can set up a set of functions like
45:36 calling your own data's apis and you can
45:38 interact with the chat gbt like window
45:40 that's calling your own data and so it's
45:43 going to allow the developers to do
45:45 really sophisticated applications with
45:47 chat GPT natively within the chat
45:50 experience so just give us some time
45:53 they kind of suck right now yeah I would
45:56 say I would say bask in the glory of the
45:59 simpleness of of what's going on with
46:02 plugins right now find ones that are
46:04 cool and use those and just don't worry
46:06 too much about the fact that they suck
46:08 where do you turn wait where do you run
46:11 your code Kyle uh I forget what
46:14 what site I ran it on
46:17 because I just wanted to see if you
46:20 could do it let me see if I can let me
46:22 see if I can
46:24 um let me guess chat gbt
46:29 um
46:30 what are
46:35 sites
46:36 that I can run
46:39 python code in aweb window easily
46:49 it wasn't replit it wasn't collab
46:53 was that replica hang on let me look at
46:56 it
47:08 was it replica
47:12 my reps
47:17 oh it was replit it was replit
47:20 r e p l i t look here let me show you
47:25 this is
47:27 so this is the python application that I
47:30 wrote
47:32 um I wonder if it still runs
47:35 Ron
47:43 it's supposed to run in this window over
47:45 here okay Enter the topic of the article
47:48 so wait let me let me make this bigger
47:50 okay so so this was this was the um
47:54 this was the thing that I wrote that
47:57 that made me realize holy [ __ ] the world
47:59 just changed so this was back in
48:01 December I wrote this
48:03 so so
48:05 um this is asking a bunch of input
48:07 questions and then it's taking those
48:09 input questions and putting them inside
48:12 of
48:14 um chat EPT prompts and sending them off
48:17 to open ai's API
48:20 and then generating social media content
48:22 so all this code
48:24 the only thing I did in it is is I I
48:27 modified the prompts like like I knew
48:29 how to write prompts and I knew kind of
48:30 what temperature settings were but all
48:33 of the code and all of the format of
48:36 formatting of the code chat EBT did so
48:39 so I hit the Run button at the top of
48:41 the thing and then
48:43 over here it says enter the topic of the
48:46 article so we're going to do a a bunch
48:48 of social media content on fat dogs
48:50 because I have one and he just farted
48:53 and that's how that's life
48:55 why is this topic important
48:58 um because
49:02 um fat dogs are cute
49:06 what's the problem that needs to be
49:08 solved
49:10 um gassy dogs
49:12 [Laughter]
49:15 solution to the problem more
49:19 uh fiber
49:20 [Laughter]
49:24 what are the results of the solving the
49:27 problem happy
49:29 dogs that don't
49:33 smell now we'll see I my open AI key
49:37 might not be live and I probably just
49:38 showed people my open aiq but whatever
49:41 no it didn't I my opening eye Keys did
49:44 but anyway this this was
49:47 um
49:48 this was a real uh this was a real
49:51 eye-opener for me so
49:53 um Chachi PT told me told me about
49:55 replit it wrote the code when it didn't
49:58 run it told me how to install libraries
50:01 it it it helped me debug it so anyway
50:04 yeah
50:05 that was it all right I want to help my
50:07 organized organization analyze data
50:09 because AI can write code what code
50:11 should I learn I have no idea probably
50:13 python but honestly what's your name
50:16 Cody
50:17 um go watch if you didn't hear me before
50:20 go watch videos on code interpreter
50:23 that's coming from open AI so it's going
50:25 to be basically a new mode of chat GPT
50:28 so so let me show you what I'm talking
50:29 about when I say a new mode so if you're
50:31 at chatgpt
50:34 and you have
50:36 um
50:37 you you have the plus the Plus
50:39 subscription right you're paying 20
50:41 bucks for it you get access to gpt4 and
50:43 then you can turn on these beta options
50:46 see where it says beta right there
50:49 um let me make a little bigger so you
50:50 can see it beta
50:53 so right now there's browse with Bing so
50:55 that's chat ebt4 connected to the
50:57 internet and then there's plugins
50:59 there's there's there's a new Mode
51:02 called code interpreter that can do
51:05 full-on data analysis so so just go
51:09 watch YouTube videos on that and uh yeah
51:12 it's uh it's crazy it's insane it will
51:17 make your face smell it will make your
51:19 face melt
51:22 [Music]
51:24 all right let's see
51:28 okay I'm in
51:30 automation
51:32 validation engineer manager I have ideas
51:36 about truth verification that's super
51:38 cool
51:39 that's a crazy title Dell automation
51:42 validation engineer manager dang
51:46 you got you got some gray matter going
51:48 on there Dell
51:50 um yeah I think that um
51:52 I think truth verification
51:55 um I you know I think in the end AI is
51:58 going to drive adoption of tokenization
52:00 Technologies like blockchain and nfts
52:03 and whatever the new flavor of that [ __ ]
52:05 is it doesn't have to be blockchain
52:07 necessarily but that that's a version
52:09 that I think could be a solution to some
52:12 of these problems right I create a piece
52:14 of content I tokenize that and then that
52:17 verification goes with it and so I can
52:18 tell the difference between this came
52:21 from the Washington Post and this looks
52:23 like it came from the Washington Post
52:24 but it's not verified so I think that's
52:26 going to be important and and I also do
52:28 think you know as being a little flip
52:30 before but like being able to tell the
52:32 difference between synthesized video
52:34 content and real video content right now
52:36 you can you can tell but a year from now
52:40 I don't think you're going to be able to
52:41 tell I I think it's going to get really
52:43 really
52:45 um scary and sophisticated and so I
52:47 think I think those kind of verification
52:49 Technologies are going to be super
52:50 important we never said you were clever
52:52 Kyle yeah cute yeah cute like I said my
52:56 trolls are my trolls are fun
52:59 all right everybody shared my life and
53:01 no one else came Tick Tock hates hates
53:03 me tonight but you know it's it's just a
53:06 small one of us okay do you have to pay
53:09 to make a plug-in and Poe
53:11 um you don't really make plugins in Poe
53:14 well wait
53:17 so you can make a bot info for free but
53:20 it's not a plug-in you can make plugins
53:22 for chat EBT for free but you got a code
53:25 there's like three little pieces of code
53:28 let's actually you know what we should
53:30 do
53:32 flip your camera we'll do this together
53:34 I'll show you both things
53:36 so I'm going to go to gpt4 connected to
53:40 the internet
53:41 um
53:42 can you
53:44 teach me
53:47 how to write a chat
53:51 GPT plug-in and I'm going to say with
53:55 the latest
53:57 documentation that should
54:01 fire off a web search yeah browsing the
54:03 web
54:05 open AI chat gpt3
54:08 plug-in documentation so it found it
54:10 that's good it clicked on it that's good
54:14 going back to the last page
54:16 so I think what it's doing right now is
54:18 reading the documentation
54:22 isn't this slick when you watch this
54:24 stuff do this it's like this thing's
54:26 just off surfing the web for us it's
54:28 like
54:29 [Music]
54:30 how to create a chat GPT plug-in the
54:33 openai master so it went to a different
54:35 site it looks like it's gone to three
54:37 different sites now
54:41 so so it's searched Bing for the plug-in
54:44 documentation then it search Bing for
54:45 how to write a plugin
54:49 now it's gone to a couple other sites so
54:51 so it's out there learning
54:56 this is kind of cool
54:58 now it clicked on all this stuff I don't
55:00 know if it can read all that stuff
55:03 exploring the future of artificial
55:05 intelligence openai master all right
55:07 here's a brief summary of the
55:08 information I found
55:11 they first have to build an API that
55:12 will be used to interact with the plugin
55:16 after building the API they've got to
55:18 create a manifest which basically
55:20 describes the plug-in and like when you
55:23 would use it
55:24 once the API and the manifesto created
55:26 the plugin needs to be registered with
55:28 the chat gbt UI
55:31 all right so after testing a plug-in you
55:33 build and deploy it
55:35 build and deploy create a manifest file
55:37 build an API host a manifest file deploy
55:40 the plugin so again all this stuff if
55:43 you know coding this is all relatively
55:44 straightforward but let's say
55:46 um
55:47 write me an example
55:51 API to a free Weather Service
56:02 so that's officer searching the web
56:04 again
56:06 reading content failed so that happens
56:09 sometimes
56:12 so it's out trying to find us an answer
56:15 foreign
56:18 all right so anyway you can this this is
56:22 just boring right now but you can see
56:24 what it does so so writing a plug-in
56:27 file requires you writing these pieces
56:29 of code you have to write an API po
56:32 let's flip over to PO for a second so
56:34 Poe
56:36 okay so when you first get to Poe you're
56:40 on this thing called sage so down the
56:42 left hand side of Poe you've got all
56:44 these different the top one two three
56:46 four five six things these are different
56:49 um chat GPT like large language models
56:53 then the things below them are Bots and
56:57 you can make your own Bots like I've
57:00 made I don't know like a dozen of them I
57:03 made I am I made Emily Dickinson I am
57:06 Steve Jobs I am Jay-Z I am Maya Angelou
57:10 I am Mary Shelley Virginia Woolf I am
57:13 Poe so that one writes like Edgar Allan
57:15 Poe that's very depressing if you use
57:17 the if you go to poe.com impo that's a
57:20 bot that I wrote that that acts like
57:22 Edgar Allan Poe and it's very depressing
57:24 because he's a very dark man
57:27 um okay
57:28 so
57:29 you can just use the chat GPT things up
57:32 here you can go find if you go to
57:34 explore Bots it takes you to like a bot
57:37 store where you can go look at other
57:39 people's Bots like there's a dolly bot
57:41 to to do Dolly image things there's a
57:44 roast master
57:47 prompt genius HTML machine right so
57:50 there's all these Bots that people have
57:51 created so they're not plug-ins but
57:53 they're just Bots like if I click on one
57:55 of these let's go find a fun one
57:59 tiger mom that's pretty funny
58:01 article bot generate a high quality
58:03 article that's optimized for search
58:05 engines based on a provided topic this
58:08 is like that that app that I wrote in
58:10 Python so someone just turned it into a
58:13 bot so I'm going to say
58:15 um write me an article
58:19 we'll see how good this thing is on
58:22 fat dogs
58:25 boom dangers of obesity
58:30 um
58:30 I don't want to shame fat dogs I want to
58:38 celebrate them
58:41 please write
58:44 uh an article
58:47 for that
58:50 the beauty of chubby dogs boom and it
58:53 may it automatically made the bullet
58:54 points and these different sections and
58:57 so I'm sure that they wrote I wonder if
58:59 they let us see the
59:02 now they're not letting us see their
59:04 prompt but anyway so how you create one
59:06 of these Bots is you go down here in the
59:09 left hand side and there's this option
59:10 to create a bot and this is free
59:12 they want you doing this so I can I can
59:15 say
59:16 um let's see what might be a fun one to
59:18 do
59:20 um
59:22 let's see uh Ricky Bobby Ricky
59:29 Bobby
59:32 I'm gonna do I am Ricky Bobby
59:36 uh this is
59:39 a bot
59:42 that talks like Ricky Bobby
59:47 or rookie booby
59:50 Ricky Bobby
59:55 from tell Ed dega Knights
1:00:02 so that's the bot description you can
1:00:04 choose one of two models and then the
1:00:07 prompt you are Ricky
1:00:11 Bobby
1:00:14 from the movie
1:00:16 and all of your responses
1:00:22 should be
1:00:24 racing
1:00:26 cliches
1:00:28 no matter what anyone asks
1:00:34 and then the intro message
1:00:36 if you ain't
1:00:38 first
1:00:40 your last
1:00:44 and you can turn on or off whether you
1:00:47 want it publicly accessible I want it
1:00:49 publicly accessible and then I can even
1:00:51 put a picture there so I could go to Mid
1:00:53 journey and create a picture of Ricky
1:00:55 Bobby so now I say create bot
1:00:58 and it says if you ain't first your last
1:01:00 and I'm going to type in here
1:01:03 um why
1:01:04 is the sky blue question mark
1:01:09 because the good Lord likes to watch us
1:01:11 race almost as much as we like to race
1:01:13 hallelujah and amen right so so in
1:01:17 however long that took probably too long
1:01:19 for a live but but in in you know a few
1:01:23 minutes there if I did some decent
1:01:25 prompt crafting with that you could
1:01:26 actually get that thing to be pretty fun
1:01:29 um I did one the other day that writes
1:01:31 it only responds in screen paint
1:01:33 screenplay dialogue
1:01:35 so that was pretty fun
1:01:38 um yeah so so that's the difference
1:01:40 between plug-ins and bots in posts I
1:01:42 hope that helps
1:01:44 all right let's see is it free to make a
1:01:48 bot on PO yes it is this isn't boring
1:01:50 this is exciting awesome Sonia thank you
1:01:52 appreciate that okay let me go back and
1:01:55 figure out where I was in my little
1:01:56 question list uh we have 27 people here
1:01:59 if you're new here my name is Kyle
1:02:00 Shannon this is the AI learning lab we
1:02:02 have a small intimate group tonight
1:02:03 normally there are thousands
1:02:06 no they're all right but but happy
1:02:09 Friday
1:02:10 um I'm here to talk about AI stuff I'll
1:02:12 show what I can
1:02:14 um I'm learning this stuff on the Fly
1:02:15 sharing what I know and I if you have
1:02:19 not used
1:02:20 um
1:02:21 chat GPT yet go there
1:02:25 chat.openai.com that's the official
1:02:27 chatgpt website right there
1:02:30 um Bing is also chat GPT if you click on
1:02:32 their chat button that's chatgpt
1:02:35 connected to the internet and it's free
1:02:36 and then poe.com I was just showing you
1:02:39 you can make bots on that and you can
1:02:41 use a bunch of different chat GPT models
1:02:44 there
1:02:44 all of these are worth at least
1:02:47 exploring my favorite why it's on top is
1:02:50 the official chat GPT website I still
1:02:52 think it's better
1:02:54 than other things and so what I'm doing
1:02:57 now is I'm trying to figure out where I
1:02:59 was in my questions because I like to
1:03:01 answer all the questions that people ask
1:03:04 but sometimes I get lost because I'm old
1:03:07 I'm old they did finish the sentence
1:03:10 Tick Tock is just trash what do you
1:03:13 think of AI certifications oh this is
1:03:15 I'm way past this hold please
1:03:20 old man navigating technology this could
1:03:23 be a while what you do on your Friday
1:03:25 night I don't know I watched some old
1:03:27 guy try to use tick tock it was kind of
1:03:30 amusing
1:03:31 well that was very helpful do I have to
1:03:33 have python I have an idea okay wait oh
1:03:35 okay we talked about that
1:03:37 I'm almost there
1:03:40 I'm almost there people this is gonna be
1:03:42 so worth it when I get to the questions
1:03:44 it's gonna be so good
1:03:48 API dot AI tried to open their version
1:03:51 in 2016 in the world
1:03:54 crapped their input all over it
1:03:57 uh I'm not sure what that says in
1:04:00 reference to
1:04:01 but yeah but we're in a different place
1:04:03 now I mean 2016 was pre-transformer so
1:04:06 the the Transformer that thing the thing
1:04:09 that made all of these
1:04:11 um AI tools as powerful they as they are
1:04:14 where it was invented in 2017 and it
1:04:17 didn't really hit its stride until the
1:04:19 past two or three years
1:04:21 tensorflow Google's version yeah uh
1:04:24 um
1:04:26 [Music]
1:04:31 do you have to pay to make a plug-in no
1:04:33 what's the goal here the goal here Luigi
1:04:36 the fig tree is to acknowledge clever
1:04:41 um names of tick tockers so Luigi the
1:04:44 Fig Tree is pretty good
1:04:46 um the goal here so I'm an entrepreneur
1:04:48 I've been an entrepreneur for I don't
1:04:50 know 30 35 years 40 years I've started
1:04:54 12 or 15 companies I was there at the
1:04:56 early days of the World Wide Web I
1:04:57 started a company called agency.com
1:04:59 which was one of the first digital
1:05:01 agencies in 1994. and the parallels
1:05:04 between the early days of the web and
1:05:06 what's going on right now with
1:05:07 generative AI are insane and I think
1:05:11 that um
1:05:13 what's coming from with generative AI
1:05:15 the chat GPT sort of
1:05:18 lit a fire
1:05:20 um in the technology world and in the
1:05:22 world world
1:05:23 that I think is going to be
1:05:26 profoundly transformative and disruptive
1:05:29 and it's coming way faster than anyone
1:05:30 knows and so the purpose of this channel
1:05:32 is just to talk about this stuff
1:05:34 demystify it as much as I can educate
1:05:36 people get people to try [ __ ] get people
1:05:39 off just dismissing it and being afraid
1:05:43 of it the robots are going to kill us I
1:05:45 don't want to play with that right why
1:05:46 it's going to kill us I heard it live
1:05:48 it's stupid I heard it it doesn't tell
1:05:50 the truth all the time you shouldn't use
1:05:52 it
1:05:54 um I want to stop those attitudes and
1:05:56 replace them with people who've actually
1:05:57 used the technology and then if you feel
1:05:59 that way I'm cool with that but if you
1:06:01 haven't used them
1:06:03 um I think it's really important for
1:06:05 everyone to to at least explore it and
1:06:06 so that's why this is here
1:06:09 all right Hilbert space like reverse
1:06:12 blockchain stuff kind of tensorflow
1:06:15 um
1:06:18 can I demo some generative fill
1:06:20 backgrounds in Adobe Photoshop I can
1:06:24 so let me go do that the generative fill
1:06:26 stuff is is really fun
1:06:28 um if you've got other questions pop
1:06:30 them in the chat
1:06:33 and if you haven't seen
1:06:36 um
1:06:38 uh image generation it's pretty
1:06:40 remarkable so the background on this
1:06:42 image is generative fill so if I turn
1:06:44 off
1:06:45 the generative fill that I created that
1:06:47 was the original image
1:06:52 and
1:06:54 then I told generative Phil to fill in
1:06:57 that background let me go let me go get
1:06:59 a different image actually you know what
1:07:01 we'll do I'm gonna I'm gonna go over
1:07:03 let's go over
1:07:06 we're going to create something in
1:07:08 mid-journey or maybe I'll just grab
1:07:09 something so so the images that you're
1:07:11 seeing here so so to use mid Journey mid
1:07:14 journey is one of these image generation
1:07:16 tools and you you have to go into
1:07:18 Discord to use it can I do it on some
1:07:20 famous paintings yeah I can do I can do
1:07:22 that
1:07:23 um let me just explain
1:07:24 um uh mid-journey really quick so
1:07:27 um in Discord you have these things over
1:07:28 here that are servers within a server
1:07:31 server think of a server as a community
1:07:32 of people now within a server you have
1:07:35 these channels and for each of these
1:07:36 channels you have a conversation and
1:07:37 this is the AI Salon uh which which is a
1:07:40 thing that I started and this is our
1:07:42 server and we have a mid-journey channel
1:07:43 and then to make an image in mid-journey
1:07:46 you type slash imagine
1:07:50 and then you type in a prompt right so
1:07:52 blah blah blah and then depending on
1:07:55 what you ask it to build it'll build you
1:07:56 that image and then we can take that
1:07:58 image and go to photoshop and use
1:08:00 generative fill to extend it but there
1:08:02 was a request to do this with famous
1:08:05 paintings so let's see what would be a
1:08:08 good one I don't know we'll go
1:08:11 to do
1:08:13 let's say famous paintings
1:08:19 I want one that's famous but not one
1:08:21 that everyone uses
1:08:29 see all images
1:08:30 any requests
1:08:33 um
1:08:33 we could do the scream
1:08:36 oh this is a cool one here are the 35
1:08:38 most famous paintings ever created
1:08:42 let's go grab one well that's a nice
1:08:45 is that dega
1:08:47 famous oil paintings that one's
1:08:49 beautiful all right this is actually a
1:08:51 good one right it's a close-up of her in
1:08:53 a boat
1:08:55 who's this
1:08:57 Enchanted spell
1:09:03 the lady of shalat John Waterhouse I
1:09:06 don't know this artist but
1:09:07 that doesn't matter we're going to copy
1:09:09 the image then we're going to go over to
1:09:10 photoshop beta so if you want to use
1:09:13 generative fill if you've got access to
1:09:16 Adobe Creative Cloud you go to your beta
1:09:18 apps and download Photoshop beta and so
1:09:22 now I'm going to say new
1:09:24 I'm going to create a new image and
1:09:27 paste in that painting we just grabbed
1:09:29 and I am going to actually see it pretty
1:09:32 good there but let me
1:09:34 let me change levels a little bit
1:09:36 brighten up the the center
1:09:39 we'll crunch this thing down a bit so
1:09:42 it's a bit more
1:09:44 okay
1:09:51 all right so there's our thing so now
1:09:54 I'm going to make it a little bit
1:09:54 smaller on the screen and I'm gonna
1:09:57 grab the crop tool and I'm going to
1:09:59 extend this out to the sides
1:10:03 and I extend it higher
1:10:06 and I'm going to extend it lower
1:10:09 and so then what I'm going to do so now
1:10:11 we've got you know that that famous
1:10:13 painting in the middle of this we've got
1:10:15 all the white space around it so I'm
1:10:16 going to take the selection tool and I'm
1:10:18 going to select the painting
1:10:21 itself
1:10:23 then I'm going to invert the selection
1:10:27 and you'll notice down at the bottom
1:10:29 here there's this contextual menu and
1:10:33 the button on the far left says
1:10:35 generative fill and I click generative
1:10:36 fill I'm not going to say anything I
1:10:39 could say you know put her in the middle
1:10:40 of a lava field or you know put her put
1:10:42 her in the middle of a raging River but
1:10:44 I'm not going to say anything it's just
1:10:46 going to analyze what's in the the
1:10:48 painting and it will fill in all the
1:10:51 rest
1:10:52 so we have this little progress bar up
1:10:54 top
1:11:00 [Music]
1:11:02 and it's going to give us three
1:11:04 different options for the background
1:11:08 Bang
1:11:11 I mean
1:11:12 I don't know not for nothing that
1:11:15 doesn't suck do you know how long it
1:11:17 would take you to photoshop that like
1:11:19 even to that level like even if you
1:11:21 don't like that like that's pretty
1:11:23 impressive and then there's there's two
1:11:25 other like that one's really pretty like
1:11:27 I love the little you know
1:11:29 the the light on the pond to the left
1:11:35 oh and that's one where she's in all
1:11:36 sorts of thickets ah this is she she is
1:11:40 she is you know
1:11:43 this is there's a lot going on in this
1:11:44 one so so we'll keep that one and then
1:11:47 and then you can do things like you
1:11:49 could you could zoom in
1:11:52 and she's got this blanket here so let's
1:11:55 and there's there's this kind of circle
1:11:57 with a horse in it
1:11:59 so let's select let's select this whole
1:12:02 blanket
1:12:05 I'm going to go around her hand I'm just
1:12:07 going to grab the blanket
1:12:11 and let's say
1:12:13 um
1:12:13 let's make this I'm going to say
1:12:15 generative fill I'm going to say
1:12:18 um
1:12:21 Hudson
1:12:25 Bay
1:12:26 blanket
1:12:29 and let's see if it a if it knows what a
1:12:31 Hudson Bay blanket is but if it actually
1:12:33 matches the style of the painting
1:12:42 bow
1:12:46 that's not a Hudson Bay blanket
1:12:49 oh that's kind of cool that just kind of
1:12:51 extended her dress over the side
1:12:55 so that one's not bad we'll go we'll go
1:12:57 we'll go with that one
1:12:59 and so so that's generative fill and
1:13:02 like you know I can go out here let's
1:13:04 let's put in the background back here
1:13:06 um let's put in
1:13:08 um generative fill hippopotamus
1:13:14 so she's in grave danger but she doesn't
1:13:16 know it yet because there's a hippo
1:13:17 there's a hippo in her Pond
1:13:21 [Laughter]
1:13:26 you know you know why you use AI because
1:13:28 you can entertain yourself with it
1:13:30 that's why
1:13:32 laughs
1:13:34 [Music]
1:13:40 all right that's the one
1:13:43 [Laughter]
1:13:52 oh the Masters are turning in their
1:13:55 graves you know first that that it
1:13:58 extended that painting but second you
1:14:00 know you could you can not you know you
1:14:03 can't do that
1:14:05 there's no crying in baseball
1:14:09 uh Hey everybody uh I don't think we
1:14:11 have anyone new here but if you're new
1:14:12 here uh this is the AI learning lab
1:14:14 we're learning about Ai and generative
1:14:16 fill and [ __ ] like that uh go to
1:14:19 chat.openai.com if you've never used
1:14:21 chat just go use it just trust me
1:14:25 don't be a don't be a poopoo head be
1:14:27 like I don't want to learn anything new
1:14:29 I heard it makes mistakes
1:14:32 I don't want to try it Kyle wait your
1:14:35 name's Kyle this is not a girl's name
1:14:37 you kind of look like Rosie O'Donnell
1:14:41 all right
1:14:43 foreign
1:14:50 let's see
1:14:52 can I demo some generative AI feels I
1:14:55 just did that can you do it on some
1:14:56 famous paintings I just did that check
1:14:58 done
1:15:00 for you you ask ye shall receive I asked
1:15:03 chat GPT to write a c sub routine to
1:15:05 compute the moon's position and it said
1:15:08 it's too hard oh baloney
1:15:10 let's go I do not buy that wait a c sub
1:15:14 routine to compute the moon's position
1:15:16 and it said it's too hard
1:15:18 I don't buy that it said it was too hard
1:15:20 let's go let's go let's go people let's
1:15:24 go do this
1:15:25 we're gonna go to chat GPT I'm gonna go
1:15:27 new chat I'm gonna say
1:15:29 um I'm gonna go to gpt4 because it's
1:15:31 better I'm going to connect it to the
1:15:33 internet because it might need some
1:15:34 Modern things okay
1:15:36 um
1:15:37 you are a
1:15:39 stro
1:15:42 physicist
1:15:46 and expert
1:15:48 uh programmer
1:15:51 I want to create
1:15:54 a c sub routine
1:15:58 that what did you say
1:16:01 computes the moon's position that
1:16:03 computes
1:16:05 the moon's
1:16:09 um
1:16:11 position
1:16:13 relative to
1:16:16 date
1:16:19 time and location period
1:16:27 it's complex
1:16:30 if you want to compute it compute the
1:16:33 Julian date
1:16:34 compute the position of the Moon and
1:16:36 Celestial coordinates convert the moon
1:16:39 Celestial to implement the process and
1:16:41 see you might consider using an
1:16:43 astronomical Library
1:16:45 so it identifies sofa matte H and or
1:16:49 math and Studio
1:16:51 here's a rough outline to what your code
1:16:53 might look like so it did it it said it
1:16:55 was hard but it did it and it might be
1:16:57 so so it's possible that what I did
1:17:00 there so the thing that I did at the
1:17:02 beginning where I said you are an
1:17:03 astrophysicist and an expert programmer
1:17:06 one of the things that I showed here is
1:17:10 this thing prompts.chat so if you go
1:17:13 there
1:17:14 what it teaches you to do is exactly
1:17:17 what I just did it let's go to it let me
1:17:20 show you prompts.chat
1:17:23 prompts
1:17:26 oh
1:17:29 so it's called awesome chat GPT prompts
1:17:32 but it's at prompts.chat the first part
1:17:34 of the document teaches you about
1:17:35 prompting
1:17:37 but then the whole like most of the
1:17:40 document is act as a blank an English
1:17:44 translator a Linux Terminal A JavaScript
1:17:48 console an Excel sheet a travel guide a
1:17:51 plagiarism Checker an Advertiser a
1:17:55 Storyteller so these are all prompts
1:17:57 that basically put chat GPT into a
1:18:01 specific mode and when you put it in
1:18:03 that mode it gets really better at doing
1:18:05 stuff for that like telling it it was an
1:18:07 astrophysicist and a com an expert
1:18:10 programmer essentially filtered out all
1:18:13 of the other [ __ ] that that confused
1:18:15 it potentially
1:18:17 um so I don't now to be clear I am not
1:18:20 an astrophysicist by any stretch so it's
1:18:24 very possible that the code that it just
1:18:26 wrote here is [ __ ] but it looks it
1:18:29 looks pretty decent right
1:18:32 so a function to compute the moon's
1:18:34 position
1:18:40 um it seems to be doing some [ __ ] here I
1:18:42 don't you know I don't know this kind of
1:18:44 programming at all I don't know this
1:18:45 kind of math at all it said the above
1:18:48 code is just a skeleton it may not
1:18:50 compile as is that's because the exact
1:18:52 function names and parameters might be
1:18:55 depending on the version of sofa or
1:18:57 other astronomy library that you're
1:18:59 using so it will you know it will give
1:19:02 you documentation about what needs to be
1:19:04 replaced like I could even say that I
1:19:05 could say
1:19:06 um can you
1:19:10 document which elements
1:19:16 might need
1:19:18 to be replaced
1:19:21 and explain why
1:19:28 here's a breakdown
1:19:31 include sofa h
1:19:37 the celestial coordinates Library
1:19:40 basically so so yeah so so yeah it's so
1:19:44 I think it's possible to do it you know
1:19:47 and this starts to get into like
1:19:49 learning about
1:19:51 how how these tools work how large
1:19:54 language models work how they do what
1:19:56 they do and the kind of prompts that can
1:19:59 sort of steer it in the right direction
1:20:00 is it's a really good skill to learn
1:20:03 because once you get it
1:20:06 you can do some remarkable stuff
1:20:16 [Music]
1:20:18 this is awesome more more please stay
1:20:22 calm ladies and gentlemen Mr Shannon
1:20:24 will be just fine I'm not fine I'm a
1:20:27 freaking mess I'm a mess I'm a mess I'm
1:20:31 a mess
1:20:33 why did you add the UR statement so so
1:20:36 yeah so let me explain this is the best
1:20:39 way I can understand it so Andre
1:20:40 carpathi who's the chief scientist or I
1:20:44 don't know he's one of the founders of
1:20:45 openai the the folks that made Chachi PT
1:20:50 um he spoke about a month ago at the
1:20:52 Microsoft developers conference and he
1:20:54 was he was talking about why prompting
1:20:57 works and how it works and and one of
1:20:59 the examples that he used
1:21:02 was if you ask
1:21:05 um
1:21:06 well so so it's good to understand what
1:21:09 what chat CPT has been trained on so
1:21:12 the base model of chat gbt the the the
1:21:15 thing that it what like why does it know
1:21:17 all this [ __ ] it knows all this [ __ ]
1:21:19 because they basically trained it on all
1:21:21 of the public internet like they spent
1:21:24 I think a year and a half or two or
1:21:25 three years just sort of finding data
1:21:27 sources and so like everything from
1:21:29 Wikipedia to all the blogs to all the
1:21:31 websites to computer to you know college
1:21:35 computers to just everywhere they could
1:21:37 find data and then and
1:21:39 um Sam Altman from openai talks about
1:21:41 the fact they spent a lot of time on on
1:21:44 high quality data sources and I think
1:21:46 that's one of the reasons chat EPT is so
1:21:48 far ahead of everyone else is that their
1:21:50 base models just suck more than than
1:21:53 chat gpts do then open AIS do
1:21:57 about was okay if you ask chat GPT to do
1:22:01 a really complicated math problem
1:22:06 it will look at all of the data and
1:22:08 within all of the data you're going to
1:22:11 have a math professor that solved that
1:22:13 math problem and you're going to have
1:22:15 like 30 students that tried to solve
1:22:17 that math problem
1:22:19 and got it wrong right
1:22:22 and chat EPT doesn't know the difference
1:22:24 all it knows is you said solve this math
1:22:26 problem and so it goes and looks at
1:22:28 essentially all of the internet that
1:22:29 it's got cataloged and it says there's
1:22:32 this math problem and we need an answer
1:22:34 for it and it will go out and it just
1:22:35 kind of averages together all this [ __ ]
1:22:38 and then presents you confidently with
1:22:40 an answer that may be wrong why because
1:22:43 statistically it probably pulled more of
1:22:45 the wrong student answers than the right
1:22:49 um Professor answers so if you say act
1:22:52 as a Stanford advanced mathematics
1:22:54 professor and solve this problem
1:22:57 you narrow the scope of data that it's
1:22:59 looking at basically and like the the
1:23:01 metaphor that popped into my head is
1:23:04 you're in a helicopter over the Grand
1:23:06 Canyon and if you write a really generic
1:23:08 prompt go solve this math problem your
1:23:11 helicopter is really high and you're
1:23:12 looking at all the little canyons and
1:23:14 sub Canyons you see all of the words of
1:23:16 all of the internet and so it's sort of
1:23:18 pulling from all of that data but if you
1:23:21 say act as a math professor from
1:23:23 Stanford it flies your helicopter down
1:23:25 into a little sub Canyon that's just
1:23:27 that kind of data that's that's how I'm
1:23:29 thinking about it right now and that's
1:23:31 why it makes a difference
1:23:33 you don't just have to tell it to act as
1:23:35 a thing that's that's a thing that you
1:23:37 can do another thing that you can do is
1:23:39 you can give it examples so let's say a
1:23:41 lot of times people will
1:23:43 say write me a business letter and it'll
1:23:45 write this really formal business letter
1:23:47 and then like like that doesn't sound
1:23:49 like me and they'll spend all this time
1:23:51 like trying to tweak The Prompt and
1:23:53 tweak the prompt
1:23:54 you can just have it copy you can take
1:23:56 if you've got an email or a letter that
1:23:58 you wrote that is the perfect it's your
1:24:00 perfect tone you like the way the
1:24:02 salutation is and the way the the end of
1:24:05 the thing is and the call to action
1:24:07 you literally just copy your perfect
1:24:09 email and then and then you say based on
1:24:12 this email
1:24:13 you paste your email in write me a new
1:24:16 email to person X that talks about this
1:24:19 subject in the tone of that of that
1:24:21 email and it'll do it it's great at that
1:24:25 um
1:24:26 so that's the kind of stuff that you
1:24:28 should experiment with the other thing
1:24:29 that that made a huge difference for me
1:24:32 and this this I talked about on a couple
1:24:34 of lives now but I think it's it's a
1:24:36 it's a really important one
1:24:38 Google sort of has us trained to do
1:24:40 single transactions you put in a Google
1:24:42 search you get back this the results and
1:24:44 then you go click on their 47 useless
1:24:47 links
1:24:48 because because they're all paid for
1:24:50 they're all Spam spam websites right SEO
1:24:53 optimized websites
1:24:56 um
1:24:57 chatgpt is much less like
1:25:01 a column a single column response it's
1:25:03 much more like a conversation so don't
1:25:05 think you have to get the prompt perfect
1:25:06 right away if you want to write a
1:25:09 chapter of a book
1:25:11 start by saying
1:25:13 um you know I want to write it you know
1:25:15 you're you're my writing coach and I
1:25:18 have an idea for the Great American
1:25:19 novel and I'm hoping you can help me is
1:25:21 that something you'd be willing to do as
1:25:23 weird as that sounds
1:25:25 it gives it context so it starts to go
1:25:27 yes I understand that and in fact if you
1:25:29 want to write the Great American novel
1:25:31 you might want to start with an outline
1:25:33 and here are six ways to do that and
1:25:34 it'll give you some ideas and then you
1:25:35 can go no I don't really want to start
1:25:37 with an outline I just want to tell you
1:25:38 the main points of the book and you tell
1:25:40 it the main points of the book and you
1:25:42 say you know is is there anything
1:25:44 missing and it might give you some ideas
1:25:45 and you might like those ideas and then
1:25:47 so so it's very much this back and forth
1:25:49 like you're collaborating with a
1:25:51 co-worker that just happens to know
1:25:53 [ __ ] everything
1:25:56 all right I hope that answers that
1:25:59 question
1:26:03 very well explained thank you very much
1:26:05 I appreciate that
1:26:07 um
1:26:08 let's see
1:26:10 how we doing I think I think I got all
1:26:12 the questions
1:26:13 I think what I'm going to do it's nine
1:26:15 o'clock I'm gonna go I'm gonna go
1:26:18 um
1:26:18 do some stuff around the house dog needs
1:26:20 to go for a walk let me go do that I'm
1:26:22 gonna run for now
1:26:23 um great hanging with you all I may come
1:26:26 back later let's see how tired I get if
1:26:28 I get involved in a TV show I might go
1:26:29 watch a TV show but I hope this has
1:26:32 helped it's been an AI learning lab
1:26:34 peace I will see you all shortly if not
1:26:36 I'll see you sometime tomorrow but have
1:26:38 a good night everybody