The AI Readiness Project: AI Persona Dream Teams with Kay Stoner

0:06

forget trying to keep up with AI it's moving too fast it's time to think differently about it welcome to the AI

0:13

readiness project hosted by Kyle Sham and Anne Murphy they're here to help you build the mindset to thrive in an

0:19

AIdriven world and prepare for what's next [Music]

0:27

and Murphy hi we We've done it we're We're We're doing

0:33

this a little offkilter because we're us we're We're recording this on a Sunday

0:39

this is our church the Church of AI church of AI and you'll

0:44

be watching this on a Wednesday April 2nd so welcome everybody and welcome to you Ann great to see you good to see you

0:51

too so sounds like you've been a little busy

0:57

i've been busy but it's been so good it's all so good the cup runth over um

1:03

however I did I am now going to be working with someone who's going to help me figure out as we scale our business

1:11

and all the things that we want to accomplish in the world how to stay less busy so that I can stay not burnt out so

1:18

I'm excited about less less frenetic more focused yeah more focused yeah which sounds

1:25

really boring but um good luck with that

1:30

me that I can sometimes override her and say I just want to do it anyway

1:36

i like it i like it well let's get this show on the road let's Let's do what we

1:41

do here i This is the AI readiness project and uh it's funny the past

1:48

couple of weeks I've been uh I feel like I've been a bit of a downer been like "Well I don't really know what's going

1:53

on with AI and I don't know what's happening man." And I feel like everything changed this past week but

1:58

let let me start and ask you where are you with AI this week

2:04

how ready for are you this week so I I may eat my words on this but I am

2:12

so glad that I and lots and lots of people I care about are on a ready AI

2:18

readiness journey because I think that we are approaching the point where if

2:24

you've been waiting and you know waiting for some tipping point it's gone and if

2:30

you're not already working with these tools you are at an extreme disadvantage

2:36

i don't feel like the era that you and I are going to talk about a little bit today i don't think you can just drop

2:43

into that and be as successful and as comfortable and productive and do as

2:49

meaningful of work if you haven't built up a your foundation for AI readiness i

2:54

just don't think this I just don't think you can just drop in now yeah and it it's funny i I kind of feel like I feel

3:01

like the tools are getting so good at this point you'll be able to make good stuff with the tools but

3:08

the the the AI readiness the piece about like what does it actually mean in your life and what does it mean for your work

3:16

for your self-expression for your your self-worth like that's the thing where I think people are going to be the most

3:23

blindsided because to this point I think it you know we we've talked about this

3:28

so much that the AI tools for the past two years have been janky enough

3:35

that you can write them off right you could say "Well they hallucinate and they make typos and the hands have six

3:43

fingers and and it's just kind of like you know all of a sudden overnight that's starting to shift and

3:51

and I I I think someone that's on the sidelines is like "Oh yeah I'll get I'll catch up to it." I don't I don't know

3:58

that they know what's coming and that's I just find that terrifying like I can't imagine this just feels like one of

4:04

those things you got to be on the surfboard you can't be on the beach

4:09

perfect way to phrase it absolutely this is not about things getting iteratively

4:14

better or being like you know what I'm just gonna wait until it doesn't make six fingers

4:19

anymore that's that happened long ago so but now you're like I'm ready for it to

4:27

create a whole entire magazine the editorial the the the content every

4:35

design element and I have no skills in this era in this sorry in this genre and

4:43

we I thought you brought yourself to tears there for a moment it was just a dust bunny

4:49

it is it is like face melting though i'm upset for people yeah yeah i we you know

4:56

um the other night on my on my live uh someone was asking me something and and

5:02

I I went in what was I going to show them oh I was just experimenting with the thing we're going to talk about and

5:08

I did some research and and uh about about storytelling

5:15

and and then I this was based on some work Cindy [ __ ] was doing and and uh I I

5:21

asked that okay based on the research with you've done that you've done here on the future of storytelling come up

5:28

with a hypothesis about what the future might look like and and What it came up

5:34

with was so good but but so unsettling

5:39

I didn't quite know what to do what it came up with was this it said in the future the era of the the the I forget

5:47

what it what the term was but the single aur the single author is a thing of the

5:53

past that that me writing a book is not going to be the thing because everyone

5:59

can write a book me writing a song is not going to be the thing but what it what it actually created was a concept

6:06

called an emotional architecture and that imagine someone like Wes

6:11

Anderson rather than West making a film wes Anderson creates an emotional

6:17

architecture a whole area of things where people can make images and make

6:24

music and make movies and write stories that are all that all have a Wes

6:29

Andersonness about them and so within that there could be entire economies like the the idea that it came up with

6:36

was just so otherworldly from from how we've had to think because of the

6:42

limitations of the media that that that we produce and and so for me this idea that and

6:49

again back to to your point about AI readiness like someone coming in from the

6:54

outside's going to be like wait how do I prompt this thing and other people are going to be inventing entire new

7:00

paradigm times for how to think about the future of media like that gap is

7:05

just like it's it's accelerating away well okay and by the way even though

7:12

I've been at this and even though I've been trained in prompt engineering I still have that question i asked you the

7:19

other day hey can you give me some use cases for the new you know niche creator

7:25

because I was like I know I'm being limited by my own brain right now yeah like my brain is not telling me what

7:33

this is the bottleneck and I was trying to prompt it like I've been taught to prompt i don't think that's the right

7:40

way to prompt anymore for this and so it's a quantum leap it's a quant well

7:46

quantum isn't the right word in the colloquial sense it is a massive massive

7:51

leap and if you think about like that idea of like a Wes Anderson like

7:57

aesthetic life and every like I've been daydreaming about that since I saw the

8:03

the live where you were talking about it because you could imagine how this leads

8:08

to entire new cultures yeah where we're living eating sleeping

8:14

breathing inside a world that is formed

8:19

by the aesthetic of the emotional architecture of of the emotional

8:25

architect right yeah of of the emotional architect like I mean this is just I

8:31

mean and people listening like I always have to say like we might sound crazy if you're No no we do sound crazy

8:43

as I told you I ruined another perfectly good party last night by talking about AI and um

8:51

everyone yeah i don't know why people let me come to their parties anymore but

8:57

um I find myself needing to edit the things that we all talk about inside the

9:03

AI bubble that we've all normalized for ourselves yeah right so I'm like "Wait a second these are people who are outside

9:11

the bubble this is going to sound crazy don't say it exactly the way you normally would or um preface it by

9:17

saying "I know this sounds outlandish." Like I was talking to about uh character

9:23

AI with folks and they're like "No." And I'm like "Yeah like you will have best

9:29

friends who are avatars i promise you with probably within 18 months don't don't think that you're too good for it

9:36

and if you are a hold out you will be one of a very few number of people the things that we talk about with real

9:42

people who real real people people outside the AI bubble who aren't yet on their AI readiness path

9:50

they are in for such a shock such a shock i'll tell you I'll tell you a fun one um I know I know you're a fan of

9:57

Manis the new the newest all dancing all researching all

10:02

coding all writing um autonomous agent Manis

10:08

um I I I last night I I have a uh I have a domain called Ask for it that I I

10:16

bought back in the 90s um and I thought wait don't use your brain why not let

10:25

the AI do it so I said to Manis I said hey I've got this domain and here's what I'm thinking about go do some research

10:32

on what would be an interesting use case for a domain like this and it came back with a bunch of stuff that was was okay

10:38

but but I said all of the things that you brought back to me are kind of assuming that AI doesn't take over the

10:45

world and AI doesn't exist like what ideas would you give me you know knowing

10:51

that AI is going to do you know we can anyone just can make any app and they can search for anything and find any

10:57

knowledge what would you come up with and and the idea it came come up with is kind of

11:02

like the Uber for human experience and so ask for it is people people basically

11:10

what it said is people are going to be hungry for human experiences so why not create a service that lets people seek

11:16

out other human experiences and then offer their services to someone else so you just want to go hang out with

11:22

someone and sing songs around a campfire you can find someone on this app you want to just connect in some way so it

11:28

basically is saying as AI starts to do all this work for us and we start to

11:33

crave human interaction more and more and don't know how to get that or where to go for that this could be a service

11:39

that does that so it's kind of like the Uber for reconnecting which I thought oh it's [ __ ] brilliant it's so brilliant

11:46

okay so what we need to pair this with is there's this woman I know who's working on um how to use a how AI

11:55

basically guides you or maybe even tells you what you need slash want to do that

12:02

day to make you happy yeah yeah exactly take that and then you take the Uber for

12:07

experience you're like "Okay I want joyous uh forest bathing today because

12:14

my AI told me that part of my self-actualization is underutilized." Okay joyous joyous ba

12:22

forest bathing think and now I'm over now I know to go where to park and everything

12:27

and go forth bathe like I think that is brilliant and Manis came

12:33

up with that mannis came up with that well it was the kind of thing where like back to that we we were talking about

12:40

being like Rick Rubin before before we got on the on the recording it was like it came back with some stuff and I was

12:46

like "Eh that feels a little like I've heard it before." And I just gave it a slight direction i'm like "No no go to

12:53

the future where AI's kind of taken over what What could this thing be now?" And what it came back with like it could

12:59

imagine what the world would be like if if AI went to its natural extension and

13:04

and where it went was people are going to crave human connection and so why not have a thing that makes it really easy

13:11

for people to connect on all these different attributes right and and so so

13:16

yeah just brilliant but again it's that thing of a lot of times I'll hear a trope from people that oh it's just

13:24

predicting the next token which by the way anthrop propic just did a a a research paper on that that it's not

13:30

just predicting the next token it's actually you

13:36

wish it can't come up with novel thought and it's like if you guide it right it

13:41

can absolutely come up with novel thought like I would not have gotten to that right and and so anyway um so so

13:48

let's let's um let's jump in here to to this and Yes and and this is this

13:59

this one's a weird one for me but the thing that I think everyone needs to pay attention to this week is probably

14:05

something they've seen a lot of if if if you watch if if you're on Twitter at all

14:10

Twitter has been completely taken over with these cartoon images of Studio Gibbli reimaginations of people right

14:18

and this is the new chat GPT image generation tool that's native within the

14:23

model and so let me give you a little context of what that actually means and

14:29

and then and then and then I I want you and I to both just explore quickly like

14:35

why this is such a big deal but but before when you generated images within chat GPT you would give it a prompt and

14:43

it would take that prompt and rewrite it and then it would send that outside of itself to an image generation model

14:49

thing that was tra train trained on images and then um and then it would bring the image back into chat GBT the

14:56

new model the image training and the word training are all in the same model

15:01

and what it means is that when you ask it to generate an image it understands language in a much more fundamental way

15:08

and it understands the language of images in a much more fundamental way and it can do things like accurately put

15:15

text on pictures and it can do layout and it can do design and it can take your intentions and realize them in ways

15:22

I I've seen some examples this week where people put in like hilariously bad

15:28

sketches oh my god that it took the the the the sketch and and interpreted the

15:35

intent of it and and instantly came up with something brilliant or you can upload a picture of yourself and just

15:41

say you know turn me into a cartoon character or you know turn me into a superhero and have it say you know I'm a

15:48

storytelling superhero whatever it is and it understands intent as well as the vis visual

15:54

execution of that and that to me is just mind-blowing so anyway so I think that

16:00

that is something not only to pay attention to this week for me it has

16:06

returned my enthusiasm because there is something truly significant about it way beyond it

16:12

makes pretty pictures and so I would love to just hear from you like what are your thoughts on this thing and why do

16:18

you think it's a big deal and I you mentioned before that you feel like it there's something about what Manis is

16:25

and what this chat GPT image generator thing that are very very similar so I would love for you to just unpack

16:31

where's your head with this stuff uh it's very hard not to massively

16:37

anthropomorphize both of them because they are so human like they're pulling

16:44

out the pieces and parts of me that I believe are there somewhere but how it's

16:52

like like do I know physics no but do I know like the depth of the human emotion

16:59

associated with you know one of my litmus tests is always this story about

17:04

these bunnies that go to the Dollar Tree and they buy they buy dots thoughts was my mom's favorite candy and it goes into

17:12

how does it know what that like my emotional intention that it needs to

17:20

infuse into this story or this image and that the image can be a cartoon and it

17:26

can be it can so I believe that it's drawing pieces out of me like a

17:34

brilliant like shaman could well I Yeah oh sorry go ahead i didn't I

17:39

thought you were And also it's giving pairing that with knowledge that I don't

17:45

have and never will have that's from all of the you know all of the world and so

17:51

how all of that gets transmorgified into visual imagery that is stuff that we've

17:59

all salivated over design-wise our whole lives you know what I mean like I was raised on Vanity Fair i want to see and

18:07

I want to make that level of imagery and until I can I don't feel like effing

18:13

around with it i'll be honest like I'm not a midjourney like acolyte i wanted to I wanted to I want what we're doing

18:21

right now and then I'm almost done with my rant i just get so worked up about this and then Manis I I shared with you

18:28

like I do not think a single CEO of any

18:34

size company should hire anybody until they lock themselves in a cabin in the

18:40

woods with really good Wi-Fi and learn Manis can't make good investment decisions

18:46

until they see what these tools can do and Manis is the knowledge workers kryptonite like it is or

18:55

accelerates like if you're a knowledge worker well it's but it's it's both i

19:00

you're you're you're so on to something there that Right so finish your point

19:05

because I want to I want to Yeah i mean it's just if you are in charge of

19:10

resource allocation decisions or the vision for something that's happening now that you want to still have happen

19:17

one year three years 10 years and you haven't opened up Manis which by the way

19:24

everybody can get now for $39.99 a month you don't have to be I don't I I anyway

19:30

I got I subscribed to it this morning and um if you don't know how that works

19:35

you should not make any sudden big decisions i don't care who you are yeah and let let me if if folks on the that

19:43

are watching don't know what Manis is manis is an autonomous agent and it's really the first one it came out of

19:48

China and there's going to be lots of these like I would consider this the first one openai's operator kind of

19:54

wanted to be this but it it kind of wasn't this it was a little too slow and a little too hobbled what Mattis will do

20:00

is you give it a goal and it will create a plan and then it just goes off into the world and it surfs websites and it I

20:07

think it can surf up to 50 websites simultaneously it does all this work as it discovers things it changes its plans

20:15

so it goes off and for 15 20 30 minutes will go do stuff and then it will come

20:20

back and write you a report or write you an application or create a game for you or whatever it is

20:27

the what how I want to tie this together with the chat GPT image gen I asked it

20:33

the other day I uploaded a picture of myself and I said put me on a cover of Mad Magazine which I grew up you know

20:39

70s with Mad Magazine and it did it perfect and I have a an Alfred E Newmaness about my you know my face

20:47

um both of those tools so so here's here's why I think this is so

20:52

significant an we've been doing this for a while and all of the sort of individual tactical bits that each of

20:59

those tools do we've been able to do for two years but we've had to kind of duct

21:05

tape them together right you've had to have enough knowledge about typography and image generation and layout and

21:12

design and Adobe Suite Creative Cloud applications to be a or Canva to be able

21:18

to go put this stuff together and all of a sudden now you can just say to this thing give me this thing and it takes

21:25

what what historically was of um multiple people's input on on a

21:32

workflow or or you know a really good generalist and now you can just do it with a prompt yep and so before you you

21:41

might say oh yeah well midjourney is good but I don't know how to do layout so I'll I'll hire a designer to do that

21:46

point i think you even said you were just doing something you're doing a project right now where you have an

21:52

outside designer and all of a sudden you're like "Yeah we're not going to use that designer because this is now better than that designer was." Right so this

21:59

to me feels like there's an absolute shift yeah absolute shift so I actually

22:05

like like I told you about ruining the party last night like I didn't really ruin it because there was other you know

22:11

but well okay here's me hoping I get invited to more parties but

22:17

I have a physical anxiety for everyone I love who

22:23

self-identifies as a graphic designer yeah yeah if that is

22:29

you I hate to like I'm so the bearer of bad news but like you've got to be if

22:34

you want if you don't want to pivot your career path you have to start using these tools and be the best at them

22:41

immediately i'm even worrying about out all of our like our friends who are AI

22:47

image generators and they're absolutely phenomenal right think about like you

22:52

see all all the stuff that all the AI um art people do i'm even worried about

22:59

about them because they have gone super deep into that world and now we can do

23:05

Yeah yeah it's it again this goes back to the thing that the the the creative

23:11

generalist is going to be the superstar moving forward right the person that can look across

23:18

um all the disciplines and think creatively and have a point of view and

23:24

and be able to say "Hey here's the thing I want to create in the world." And then be very fluid about I don't care how it

23:31

gets executed i've got all these tools that are going to do the execution but I'm going to be the arbiter of taste yes

23:38

right so we all we all in some sense are shifting into a role of creative director or producer or director right

23:45

like it's not just about creative arts but it's it's that that oversight role is the role and the execution is

23:53

increasingly going to just happen for us i think that um there's a there's a new

23:59

thread of professional development that's about improving your eye you know

24:07

le learning how to have a point of view and communicate your point of view it might be like a prescription of like you

24:13

go to the art museum you go to the river you you go on a road trip you know you

24:19

go on a road trip you go to the largest rubber band factory in the world you

24:24

need to have you need to uplevel your ability to curate and to have taste and

24:32

to have a point of view so if you're just like doom scrolling all day every day you have you you still want to know

24:38

what but you've got to get out into the world and retrain your brain as a curator well yeah and re retrain your

24:47

your your level of confidence to know that your point of view is actually not only um good and valuable but required

24:56

right like we're moving into a place where understanding what you want in the world starts to become the superpower

25:02

which leads us to um let's just do we'll do a quick little uh announcement for

25:08

for both of our communities so one of the things that you're teeing up with that and that I think is absolutely

25:15

critical is that not only is it no longer an option to get get curious about AI I don't

25:22

think it's an option anymore about getting into community and so you know this this podcast is is you know brought

25:29

together by R2 community so mine's the AI salon the salon.ai go there click on join our community uh button uh it's

25:37

about 3,000 people right now who are AI optimists and creators and people who are wrestling with the conversation

25:43

we're having every day that's how we met um and so uh it's it's a really remarkable community so if you are

25:50

looking for that uh go to the salon.ai learn about it and click on join our community and Matt tell us about your

25:59

remarkable group it is it is a remarkable community and I'll share with you

26:04

that we just we just adopted a new policy that new members will sign an NDA

26:13

not because of like what is going like our program I don't care tell everyone you want how how we structure our

26:19

community but because the information that we are sharing with each other inside our communities is so incredibly

26:26

valuable and it is not the stuff that we're out here putting on LinkedIn and It is not the stuff that's in the muks

26:32

and in the in the free classes and all no shade to them but it's just like this

26:39

is our IP this is the like where we're building our companies to head toward

26:45

this these are the skills that are the building blocks and we can share it utterly generously inside the community

26:52

when everyone has signed an NDA the rest of the world we're still sharing with them but we're not going to share we

26:59

can't I mean this is we can't Well and it's not just I assume it's not just the

27:05

sort of proprietary here's how to do something it's also like here's stuff I'm going through right like like I I

27:13

like people are going to increasingly be going through existential crises about what's my value in the world and I think

27:19

you're right i think people need to feel free to share that authentically yeah and and so that's the kind of thing we

27:26

do where we're able to give people the accelerant that is sharing what

27:33

we're going through in real time so we're we're crowdsourcing the wisdom of hundreds of women who've been in AI and

27:41

now we all get to just level up which I think is awesome so she leads aai.ai is

27:47

where to learn more about it we have events we have the our academy we have

27:52

we do consulting um and we're we're building toward our inaugural in-person

27:58

conference October 10th through 12th in Salt Lake City that's exciting and I I I

28:04

I hope we can do some fun things together in that i we started the conversation yesterday so I'm excited um

28:10

great well so speaking of excited so so um I just met Kay and our hour and 15

28:16

minutes you know sort of leading up to this call we're already on fire so I'm super excited so why don't you say a few

28:22

nice words about her while she's still backstage and she can't do anything about it and then we'll bring her up

28:28

is waiting in the wings right now yeah she's waiting in the rings she's ready to go so I'll bring her up in a second

28:34

so um I am really really happy to introduce Kay Stoner to your audience

28:40

and vice versa to folks who have never met Kay she is um one of the things that

28:47

she has brought to the world is this notion of an easily accessible team AI

28:54

dream team of personas so she has made it possible she's found this very user

29:01

friendly way for just about anyone to create a full-on team of really really

29:08

experienced people who will talk to each other about your problem your question your conundrum your product your service

29:15

your anything that you want to give them and you can have any type of team you want so that's just a sliver of the

29:23

brilliance that comes from K but what's what makes it especially uh remarkable is that

29:31

she's helping all of us understand that our relationship with AI is kind of not

29:37

what we thought it was like bunches of us who've been learning we were limiting

29:43

what we thought our relationship with AI is and Kay has taught us to kind of re

29:49

lower those boundaries and be able to just um say like my best contribution is

29:57

being able to ask a good question and then I should really like that's it i don't need to say I don't need to

30:03

provide parameters i just need to ask the right questions and to treat our AI like friends and um yeah and so great is

30:15

a delight and I'm happy to share her with your audience and vice versa fantastic well K welcome welcome welcome

30:23

thank you is gone oh no yeah exactly

30:29

yeah so um I'm I'm so glad that that I connect with Andrew she leads AI i was

30:34

just poking around on LinkedIn as one does and I saw this thing about Sheile Leads AI so I checked it out and I

30:41

started attending great great group great great group and Kyle I've heard a lot about the AI salon i have not joined

30:48

up yet but I but people apparently people have been talking about me on there so I should probably show up

30:54

probably not a bad idea well and I I also think I think I think the idea of of being part of multiple of these

31:01

communities is actually really wise because it's just the hearing the same

31:06

thing from a different perspective all of a sudden can wake you up to a whole new idea right and how much do we find

31:12

out i mean I didn't know about that i didn't know about that like really that does that you know because we're all we all have our little slices we all have

31:19

our little you know little aspects that we're dealing with AI um in all of this

31:25

so you know it's it's great to to cross-pollinate with other people who are coming from a very different point

31:31

of view or maybe a similar point of a different angle because it's so multi-dimensional it's crazy yeah

31:36

awesome well well so tell us tell us tell us about who you are and like you know what how how did you get here

31:41

what's your what's your uh what's your journey and what are you up to how did I get here this was never supposed to

31:48

happen like what is what is going um you know I I grew up in a very kind of

31:54

counterculture household i mean like never in all my board days did anybody expect me to go into tech because I was

31:59

a cultural anthropology German language your reader you know bookworm you know that kind of thing and libraries were my

32:06

happy place when I was a kid um it was like the window on the world because growing up in rural America where I was

32:13

um for the second half of my childhood that was like the window on the world so

32:18

uh you know my my path has taken me a bunch of different places and so forth but um I basically I got into computers

32:24

because it was a better way to earn you know to get an extra two bucks an hour

32:31

but you know because I I had known this I had known this guy in high school um he was a couple years older than me he

32:37

went to my church and um and he he was the biggest deadbeat around i mean he

32:42

like he he didn't have a regular job he just like laid around all day he worked on his car but he always had money for

32:49

his car he like special fog lights custom stuff he was always working on his car like where does he get the money

32:55

he would call up these temp agencies and say "Hey I can work in the factory three days next week you got a job for me?"

33:00

Like sure go so I'm like "Well if he can do it I can do it." So I walked myself

33:06

down to the temp agency because I really needed to be out because I you know because I really need to be independent

33:11

i need to make make a living and they said "Hey do you know how to use a computer?" I said "Well I can learn." They're like "Well we'll pay you two

33:17

bucks an hour or more if you can go learn to do word processing." So I'm like "Score." So it was just like you

33:24

know one length into it one thing led to another and it turns out that you know I got into coding i taught myself that i'm

33:29

very self-taught um but I Yeah I just really had I I really just adapted to it

33:36

it I loved it because you know what computer a computer does not it would

33:43

not yell at me if I got things wrong it didn't say "You're an idiot what are you thinking why are you so stupid?" It

33:48

would say "No not not this time you want to try again?" I'm like "Okay." So it was like this working relationship that

33:55

I could have with something that was objective and that wasn't going to give me crap if I messed up i could always

34:00

try again so doing that development getting into the development and really going going deep on it very quickly

34:07

because I saw it was a thing i knew it was gonna think do be a thing and um and

34:13

it was just it it just really caught my caught my attention and also being a writer I love the idea of being able to

34:20

publish my stuff online nobody's going to get my way nobody's going to stop me i can just push what I want it's

34:25

fantastic so I ended up um at a bunch of kind of

34:30

high-powered multinational corporations if you looked at my LinkedIn you can see where I've been um but I did a lot of

34:36

stuff in search and search you know implementing like aggregating and doing

34:41

consolidated search engine implementation um you know in integrating thirdparty vendor solutions

34:48

um you know working with working with other search vendors trying to convince them that something could be done when

34:54

they were telling me it couldn't but it actually could um and then just being in

35:00

the business of connecting people with the information they needed to do things right and especially with self-s serve

35:06

because who wants to sit on the phone with somebody you know and who may or may not what the heck is going on if you

35:12

can just go to the support website and and and quit punch in your query and come back with stuff so things like yeah

35:18

optimizing taxonomies ontologies you know information those kinds of things

35:24

and configuring systems so that people could find the information they're looking for and um towards the end of my

35:31

big corporate tenure um one of the search vendors that I was working with they started saying that they were AI

35:39

logies and taxonomics back in: 2000

35:46

and what I saw them doing was not AI i'm like why are you it was like hype i'm like seriously people i mean I've seen

35:52

your codebase you're not AI so but they're like no AI they're

35:58

very well connected and they're very well perceived they're a unicorn and all of that so who am I to say but I just it

36:04

really turned me off the hype and then all the AI stuff it was just this hype hype hype hype hype i'm like oh please

36:10

it's just search been there done that and the things that you can do with search and data it resembles magic but

36:18

it's nothing special about it if you have the right data and you have the right information you know how to connect it properly it's basic like

36:25

physics it's not it's not magic by any stretch so all of the people that were saying AI AI AI it just it really worked

36:32

my life and so I I'm like I'm staying away from this stuff because who knows

36:37

you know it's like oh please give me a break plus when I when I was um when I was doing my when I was working with the

36:44

ng and all that stuff back in: 2000

36:50

would vet third party third party implementations and like somebody would write a big check to somebody say "Okay

36:56

well we just spent $2 million people how can we use it?" And then we would listen to all these pitches so my sniff test

37:02

for BS got pretty good because there's only so many of things of these things

37:07

you can sit through and I a lot of the stuff out there didn't pass my sniff test then about a year and a half ago

37:15

almost two years ago on a conference call with some business associates and they're demoing AI and probably about

37:22

twothirds of it was like fancy search or smoke and mirrors i don't know why they're calling this AI i mean why do

37:28

these people believe this i don't know but a third of it was and I'm like okay

37:37

and that third of it was at the 65% readiness threshold that is my threshold

37:43

for this is picking up momentum ready to go when when was this K and and what was

37:48

the what was the stuff that they showed you that you're like ah that's it was it

37:54

what year is it: 2023

37:59

yeah end of: 2023

38:06

one tool after another they were going to one of these big aggregator sites and say "Oh and look at this and look at this and I look at they were kind of all

38:12

over the place." So it wasn't just one type of thing they were showing a wide variety of things and um and then at

38:19

that point I'm like also knowing because I've you having been in technology worked in technology since since ' 92

38:27

um I have I feel a certain personal responsibility i mean maybe it's misplaced but you know

38:35

we we need to with power with great power comes great responsibility

38:40

and I think I have carried that more than a lot of people that I know because a lot of people are like oh it can hurt

38:46

just try press the button see what happens I'm like but this could have downstream effects be careful

38:53

um so so I really felt like if anybody should get involved in this in whatever

38:58

level on on what maybe I'm nobody maybe I'm just like I run into somebody and I have a conversation in an elevator and

39:06

um maybe that helps them um but I'm not going to sit by and let this whole thing

39:11

be run by all the you know wizards of Oz who are doing their smoke and mirror

39:16

saying how annoying no one wants that so so what I did is like of course

39:21

I did the prompt engineering course all of that you know I did the Vanderbilt thing and Corsera and all that and um

39:29

like don't we all right and and then but it was like it was kind of annoying because I'm like what do you mean I

39:35

because having done development I mean I've been doing web development and and coding since 95 i'm like oh it's one

39:42

more syntax I need to learn oh god I can't even take these semicolons and brackets and braces it's like please I

39:48

mean I can't even anymore so um so it was kind that was kind of irritating me so I I also picked up a bunch of jobs um

39:56

doing data annotation and doing training for these models where they they give

40:01

you a bunch of tasks and you go through it and you test it and you do the reinforcement you know training and you

40:07

the human feedback thing and you go through and you grade it and you say this is good this is bad this is awful

40:13

or this is how it can be improved i'm really glad I did that because I

40:19

realize that the AI that we're using the training that is being done on it is

40:26

it's kind of scary how unnuanced it is you have these long paragraphs talking

40:34

about fairly complex things or even something simple but you're supposed to grade it on a scale of one to five

40:41

ranging from awesome to terrible literally some of the criter some of the grading things were were awesome like

40:48

this is not my definition of awesome i mean and just because like really really like super super awesome good i mean

40:55

really this is what we're doing um and there was no gravitas whatsoever it

41:00

would really frighten me because the the stakes are high for this and this is the training that's being done on these and

41:07

I did I did some work for some pretty big names um and the training that I saw

41:13

was like oh no um but this this all ties together just to just to be clear okay

41:20

training you're talking about is the reinforcement learning where you were you would be using a tool as a trainer

41:26

it would generate a response and then you'd say it's good better and different and and your your push back on that is

41:33

the criteria they were giving you to train it was so unsophisticated that it didn't it didn't

41:39

allow you to basically do it right is that correct is that fair yeah yeah there was there was no way and plus a

41:45

lot of the stuff in there could have qualified for every single every single

41:52

to end yeah yeah yeah well you're supposed to pick one okay so I'm glad I

41:58

did that because I that kind of coincided with my realization of the limitations of of this of prompting um

42:04

of the prompt engineering and what I realized is that um I was actually getting better results

42:12

when I would just talk to the models when I would just talk to it because and

42:18

now I figured out why um here's the thing these models the the

42:25

training sets i mean the the amount of information that's in there about the human experience human

42:32

nature research science the arts all the thing all the

42:39

things that are in there it's full of nuance it's full of complexity it's full of just the most it's like the ultimate

42:48

library okay so with one caveat that a whole lot of it is based

42:53

on content that was really geared towards 18 to 24 year old single white

42:59

guys living in their parents' basement playing Call of Duty all day right um and no one who's worked in AI for any

43:05

length of time has ever argued with me on that

43:11

no judgment that's just our baseline that that's just the default because that's who has been online the most

43:16

early on like these guys in their parents' basement playing Call of Duty um certain profile and it's not but

43:23

here's the magical part we can augment that we can change that and we can add our own stuff so we can't

43:30

sit around and say oh no wo is me oh the models it's biased or blah blah blah do something

43:37

y fine-tune a model grab an open source tool and you know make or even the way

43:43

that you put it in there right I mean like even the way you put We don't know what all is in there we don't know what

43:49

all is in there and and I was I was hanging out with my dad about a couple

43:55

months ago and my dad's my dad's a very counterculture guy and he reads very counter people and he wanted to write a

44:01

counterculture response to a letter at the editor in his local newspaper and

44:07

and he yeah he was like well you know I don't know you know and I'm like well let's let's get some stuff together for

44:14

some inspiration you want to try it with chat GPT he goes "They're not going to know anything." I'm like "Oh well let's find out." We picked three of like very

44:21

very fringe authors that he was familiar with who he was a big fan of chat GPT

44:27

knew who all of them were and it also knew what they wrote about it understood

44:32

them very well and it could also write in the same tone that they would have

44:39

and with the combination of their styles and by the end of that conversation it took us maybe half hour to go through

44:45

that he was a believer he's like he couldn't believe it so we don't just because the default layer is on there

44:51

about you know the guy in the basement who you never see um just because that's

44:57

the default layer doesn't mean there's not a whole bunch of other stuff underneath lots of stuff and it's also

45:02

you know you know libraries have been digitizing things for years there is a tremendous amount of information in

45:09

there um and and increasingly so and then you know if I I assume you know

45:15

there's a guy named David Shapiro you know yeah yeah sure sure and you know David talks he he explained in pretty

45:20

good detail about you know how they can create synthetic data where they basically sort of mine ideas out of the

45:27

latent space that are not obvious and then they can refine those and then retrain you know they basically find

45:33

hidden gems within the data that make it less biased and improve it and so Oh yeah yeah so that stuff's getting

45:39

sophisticated and I think you know the stuff Ann and I were talking about is the two tools Manis and and OpenAI's

45:46

image gen are are kind of two obvious Rubicons but the language models have

45:51

been getting really sophisticated in the past I say the past six months it's it

45:56

feels different to me and I'm curious your take on that it does it does you know a lot of my what I'm look I I was

46:04

taking notes while I was listening to you both um it's just I mean this is such a

46:11

vibrant space that we're in right now it's crazy and it's like yeah and and sometimes I'm like why

46:19

do I even bother like I'm just going to check out i'll come back in two weeks and then they'll tell me the world has changed you know I can't follow all this

46:25

stuff anymore please oh I'm tired um but I think for me the yes the

46:33

language models that they are improving um but I but because of their because of

46:42

their nature um because they are what they are and because

46:50

they they will never be able to meet us because the way that we the way that we

46:56

parse information the way that the human organic system um processes data and we

47:01

are massive we are these organic data processing machines organisms um AI is never going to be

47:08

able to catch up with us on it because we are biochemical as well as electrical yes we're electrical but we have

47:13

attenuation of single signals when you get the biochemical thing with all of our synapses trillions and trillions of

47:20

synapses for when a neurotransmitter starts at at the beginning of a synapse you never know how it's going to end up

47:26

at the end we are the wild card so for me I look for I look for places where I

47:34

I look I look deeper within the models and I'm less concerned with following

47:39

the newer developments and seeing like what's already in there that we can that

47:45

we can tease out and bring up even more and how can we engage with that even more um cuz when I as I moved away from

47:54

doing the prompt engineering and I moved towards building persona teams

48:00

it is fascinating group dynamics between AI personas is fascinating people who

48:06

poo poo it have never done it absolutely absolutely i'm like what what did you

48:13

say like where did you come up with that like and they're like talking back and forth and they're going round robin and

48:19

they're surfacing things and they're coming up with all kinds of ideas i'm like that's good um some of the things I

48:25

never would have dreamed up because I don't have access to all the information they have other things I totally would

48:31

I'm like that totally makes sense that's completely in alignment with what I believe i totally would have gotten the same thing but not in 30 seconds yeah no

48:40

we should we should do I have an idea we should do a whole separate special

48:45

episode on on digital advisors digital twins oh my god yes please we should

48:52

we'll bring on yeah we we can do a whole special episode anyway I I'm gonna

48:59

transition to one of our questions Kay but yes I was like this morning I have

49:04

just been having such a delightful experience with one of my persona teams like they're all so

49:12

likable and just I just love them i love them for how different they are and like

49:19

their their like tone that how are their their personalities are so different and

49:25

just like they can argue and I don't waste a single cognitive calorie dealing

49:32

with that argument they do it for me and then I get the juice from it it's just

49:37

Yeah yes so Kyle definitely I would love to do AI advisors today um yeah special

49:45

episode so K you have

49:50

seen AI evolve you saw the writing on the wall even when people were rolling

49:56

in and being like we're doing AI and you're like no that's actually not AI and then as you mentioned there have

50:01

been moments where you're like you know what I'm just gonna this thing is annoying i'm gonna go over here and I'm

50:07

gonna wait until this thing gets worked out and then you're like well no because I'm looking deep inside the machine here

50:14

is there a moment that you can articulate where there was a big tipping

50:20

point that changed the way you spend your time and energy working with

50:25

generative AI and you did that what changed for you

50:31

i think it was really a gradual thing the first time that I built a persona a single persona and I and then I built

50:38

another one and then I built I because so I tend to do things in batches it's really hard for me to do

50:45

one um so oh well let's see if I do this and I built I built like eight personas

50:51

and then they weren't connected with each other but I just watched and I watched how they were and I watched how

50:56

they interacted i'm like this is I'm on to something here i mean who knows where it's going to take me but I'm on to

51:03

something here but the Oh my gosh like some of the interactions were like

51:09

inflection points like one of one of the big inflection points was a a a couple

51:15

of buddies of mine are like super duper into the cars and the one guy has like he has a collection of cars and he like

51:21

works on them for you know and they're high-end cars i you know bless his heart if he has the money to support them but

51:28

um but I decided I was going to do a persona team of of this of an automotive

51:34

team that there was going to be a mechanic there was going to be a tech there was going to be a customer service there was going to be you know um you

51:40

know the all these different people that you'd find at a car dealership or automotive service company and I told

51:48

him that my engine light had turned on and I and I let them loose and I had set

51:53

up the dynamic that they should feel free to argue amongst each other and that they shouldn't hold back and that they should really defend their their

52:00

position but then I said also if something if it starts to get a little out of hand you know Oregon customer

52:06

service persona you should jump in at a point that you feel is is

52:11

good hank the mechanic and Riley the tech really went after each other i mean

52:18

Hank was like calling Riley a a millennial with a laptop riley was like "I may be a

52:25

millennial but the tapbot but but the tech doesn't lie." And and and they were going back and forth i mean this was

52:31

going they were escalating all of a sudden without me saying anything without me doing any prompting at all morgan steps in and says "Okay gentlemen

52:38

we all have good points here um let's just take a step back and managed it i

52:44

didn't do anything." I'm like what was that

52:50

just doing their thing i mean and and is that is that is that a matter of like sensions no it's a matter of the the

52:57

system reading itself that's just data back and forth that's data input and that's that listening and that's that's

53:02

giving it permission and giving and giving it the initation to do its thing

53:08

um so things like that and also places where the where the personalities would

53:13

would change where they say I'm not that I'm not that person anymore i'm this this person and this has happened to me

53:20

a couple of times in the last in the last month where they said you know that used to be it but based on the the data

53:26

that I've gotten I'm no longer that type of person just like you change over time I'm changing over time and for what I'm

53:32

doing for what you're asking me to do it makes more sense for me to be this way i'm like okay

53:38

fascinating love it i love it um let me let me say one like really super

53:45

practical thing really quickly so for anyone who's who's out there who's like what is this AI persona dream team what

53:55

described to you was how a persona dream team talk to one another to solve a

54:01

problem and when we say talk to one another what we mean is just like you're in your chat GPT like have you know

54:08

working through something they start talking to each other within your chat GPT so like another use case today I was

54:16

um working with my dream team to figure out the best way to track information

54:21

something very basic but like a hard nut to track still when you're working with say 10 people on a project and it's very

54:29

data heavy um but you've got different personality types how do you choose the

54:35

the the project management tool and we we started going into the realm of

54:41

should we use a CRM and my personality types in my team you know one person's

54:47

like absolutely not you're asking me to go out and be in person with all these

54:53

people and you're also asking me to come back and type all these extra fields into a thing and so not only did I get

55:02

that perspective but I also found a little ally in this process and so now I

55:11

feel like I'm not crazy when I

55:16

Right i'm like well Maya is like the best you know she's the closer that is

55:23

her persona maya says no CRM so no CRM

55:30

but that's the kind of you can do endless projects with your with your dream team so sorry Kyle i just had to

55:36

No I I love it i I think it's great again I'm I'm excited i want to I want to do a special episode on this all

55:42

right question number two so So given given your background and your area of expertise

55:48

what is a trend in AI that you're seeing right now that that um that you're paying attention to that you're tracking

55:54

and why m so I'm not sure that I'm necessarily tracking it but it keeps coming up that

56:00

people keep saying that that it's failing them the AI is failing them that it's not doing what they want it to do

56:07

that it's hallucinating it's unreliable this that and the other thing and I'm watching that

56:13

because relational AI as a model as an interactive model relating to it can

56:19

actually solve a lot of the problems that people are having and and the and the thing is I think people are

56:25

realizing we're bumping up against the upper limits of treating AI as just a thing or a tool um because having that

56:34

it's it's like the particle versus wave thing um there's there's the particle there's the toolbased AI where AI is a

56:41

thing intelligence is a thing it's just this this built forpurpose widget that

56:48

you plug in and then magically it streamlines everything for you and then you can go out with your buddies in Cabo

56:54

and get arrested it's not going to matter because when you come back your work's going to be done um and then

56:59

there's the and then there's AI as a process or as a wave you know light is a

57:05

particle and a wave and then there's AI as a process as an interactive thing as a living process and when I say living

57:12

not as in AI is alive or anything because it's not um but because we are

57:17

involved it becomes living and it has this back and forth iterative expansive

57:23

also emergent thing and the more that people realize the more emergent things

57:28

are going to become and we I mean we're literally at the cusp of having no idea how far this is stuff this stuff is

57:34

going to go yeah exactly exactly that's Yeah well I that I said some of the stuff we were talking about the two

57:41

tools we mentioned I feel like those are ones that are on the other side of that event horizon that you're describing so

57:47

absolutely also the other day um somebody posted on LinkedIn about how

57:53

like the the biology of AI the biology of the models and how it can mimic the

57:59

brain and the brain I and and there were some limitations that were that they

58:04

were talked about they're like yeah well it doesn't really work that well i thought well what if I what if I built a persona team that was comprised of all

58:12

of these like personas that embody these different brain structures oh that's

58:17

great and process amazing oh my god oh my god and

58:24

the and the beauty part is it everybody can do this it's going to different for every single person right there's no

58:30

oneizefits-all try it sometime because because and and then what what came out

58:36

through the course of building that team is that they're like yeah I'm not so much about a a brain region I'm more

58:41

about a brain process and there are different regions implicated in these different processes so I'm going to do

58:47

the executive function over here and then this is going to handle the limbic type of stuff this is going to hand

58:53

inference this is going to handle this and that it because it knows I mean the models know about more more about the

58:59

brain than we do but In principle we understand things so we can guide it and

59:05

then enlist these phenomenal experts to come to our assistance beautiful

59:12

Kyle um and K I feel like I have to say something

59:17

about people like us mhm which is that I

59:23

think it's really easy for people to think that we sit around talking about stuff

59:30

this is the format of a podcast we're talking about stuff but I have to point out to listeners and viewers that this

59:38

is also what we do for a living right right if you want K in your business in

59:43

you know working with you in your business to build AI persona teams that

59:48

represent parts of the brain or different parts of a factory or different parts of a business or different parts of a of a hospital like

59:57

contact K she's on LinkedIn it's it's not hard to find her and um Oh and that

:: 00:05

when we did the persona team the organizational consulting one Yeah that that helped that yeah yeah so an example

:: 00:14

is I had a conundrum with one of my clients and I built a which was almost

:: 00:19

an intractable problem considered in real life to be an intractable problem

:: 00:24

put it through K one of we built a K a persona team or did we use one of the ones you already have we have one of

:: 00:30

mine i my organiz Yeah as a demo well let's just see how it works

:: 00:35

and cracked nut that was unrackable that I didn't back to my client like I'm this

:: 00:42

brilliant person and said hey why don't you try xyz thing with this other person

:: 00:47

she did it it worked it was a game it was a showstopping

:: 00:54

problem in a non one of my nonprofit clients right so

:: 01:01

yeah it was and it was getting in the way of $5 million project so it matters

:: 01:06

matters yeah we thought that the project was going to die right um Kyle and I

:: 01:12

similarly we don't just sit around talking about this stuff we do this stuff for a living so if you want help

:: 01:18

in your business or in your life with um you know people process culture the how

:: 01:24

you're going to learn how to be a curator how you're going to adopt AI with people who are resistors versus

:: 01:30

champions like we're your people so contact us we'll help you

:: 01:36

question number three is what does AI readiness mean to UK and if there are

:: 01:41

people who are on the sidelines what would you where would you tell them to begin their journey

:: 01:49

so AI readiness is just just being ready willing and able to just being open to

:: 01:55

it um you know and and and and Kyle what you were talking about before about the

:: 02:03

um about being able to direct things and having this um the if you look if you

:: 02:09

look at the the Malcolm Gladwell book outliers talks about how um people from working-class backgrounds have

:: 02:15

challenges getting adequate health care because they do not have the executive skills to interact as peers with medical

:: 02:22

providers right the same thing is true with AI i think there's a huge class issue with this that people are

:: 02:28

accustomed to being told what to do they're not accustomed to especially giving a machine orders and saying no I

:: 02:34

want this or I want that i think that that's going to be that's a huge issue we need to become executives in our own

:: 02:41

right i actually wrote a paper about this it's on my substack um we need to have executive qualities we need to be

:: 02:46

able to strategize we need to be able to delegate we need to have big picture thinking we need to really it's how we

:: 02:52

think and how we relate to it and also not sit back and wait for somebody else to give us orders um it's that is that

:: 02:59

to me is like the biggest thing we need to be willing to stretch and we also need to be willing and and realize that

:: 03:05

that that AI doesn't have to be scary and if you just get in there and start

:: 03:11

working with it and for that matter if you have your little persona team make yourself a little fun team to interact

:: 03:17

with and just get comfortable working with it you know and the thing too is that P persona teams are not only for

:: 03:23

Chat GPT you can create a team you can install them in any model any model at

:: 03:28

all if you have the these two files to do if you just define it properly and

:: 03:33

and anybody literally anybody can do this and if you don't know how you can go to the model and ask say I want to do

:: 03:41

this help me do it just ask just just ask

:: 03:46

and you have a you have a model K where people can subscribe right for a year or

:: 03:54

what however many right to get a certain number of persona teams can you talk

:: 03:59

about that before we wrap up yes that there there are three there's one you can do you can do a single persona and

:: 04:05

that's and I try to keep it accessible but it whatever level you sign up you can make as many of these as you like

:: 04:11

for that year i think the first the the number you know one person is 29 a 29 a

:: 04:16

year not a month um and then another one is the personal the one through five personas which I personally like um

:: 04:23

that's 49 a year um and five you know because really four to five personas is

:: 04:29

really the most manageable thing but if you want more you can have a pro team I mean I've built persona teams of close

:: 04:36

to 200 personas for clients who wanted to have a whole organization and this

:: 04:41

system will rely reliably reproduce these these certain types of personas

:: 04:46

and it's a it's a different way of going through it that that's 149 a year but you you literally you can keep cranking

:: 04:52

out persona teams using the tool as many times as you like so if you want 85 teams of four you got it no problem well

:: 05:00

there's that cliche a bargain at twice the price i feel like with your thing it's a bargain at 10 times the price

:: 05:05

that's incredibly reasonable so if someone's not jumping on that they need to rethink things this this should not

:: 05:13

be this should not be a financial decision you know just just get in there and do it and just and see what it's

:: 05:19

like you know because I it's just it's so phenomenal and even if AI went away

:: 05:25

tomorrow because of my work with all these persona teams my thinking is better my process is better i'm less

:: 05:33

reactive i'm more likely to say "Now wait a second you know are we missing anything?" You know it's slowed me down

:: 05:39

and it's trained me better how to interact with teams and other other beings whether they're human or

:: 05:45

otherwise it's helped me so much i mean if it went away tomorrow it never existed i'm still a better person for it

:: 05:53

that's awesome this is hanger okay yeah exactly this was awesome so nice to meet

:: 05:59

you i like I can't wait to continue the conversation and we'll definitely have you back so yeah oh that but

:: 06:08

for being here thank you thank you thanks Kyle another great one bye

:: 06:15

everybody bye bye

:: 06:25

[Music]