Artificial Intelligence: Real Talk

Eli Davis: Welcome to Artificial Intelligence, real Talk.

This is Eli Davis, your host, and I have here Peter Swim who is a a conversational AI technologist living outside of Seattle, Washington.

He founded.

To Ville, LLC and specializes in conversational ai, early stage company planning and community building.

He has helped companies grow towards successful exits, curate diverse and exclusive communities, and support other acts of human kindness.

For fun, he also likes to produce and create strange music.

How you doing, Peter?

Peter Swimm: I am great.

I'm great.

Thanks for having me.

Eli Davis: you know, one of the things that, I am interested in and I want my audience to know because you know, once you get into the artificial intelligence bubble, you start to think that everybody knows what you know,

Peter Swimm: Yeah.

Yeah.

Eli Davis: so, so, so what we're gonna have to do is we're gonna have to ba basically break it down.

Could you tell us what you do?

Peter Swimm: Sure.

So, for the last 10 years I've worked as a product owner in making tools for building conversational AI things.

So think like chat bots you know, the robot voice on the end of the phone.

You know, all kinds of things that are kind of like ways for people to talk naturally.

IE not.

You know, like enter commands into a command line, but just like through normal interactions cause the computers to do work.

And that's like the simplest way to explain that industry, I think for lay people.

And what I do is I help people design and build products that are, one, do what they set out to do and do it in a way that is beneficial for both businesses and the people who use them.

Eli Davis: So, so, so from my research you use agents.

Peter Swimm: Yeah, so the, they used to come 'em chatbots and they used to call 'em all kinds of things.

And you know, it used to be press nine, but now it's agents.

And I think the difference between the chatbots of your, and what they're calling agents today is the fact that these chatbots are able to make decisions on how the problem was solved independently of humans.

And that's the pitch.

But the reality is very different.

Eli Davis: Okay, so, so, so that goes into one of the things I want to discuss.

Us what is the importance of human first design and ai?

Peter Swimm: So one computers work at our pleasure as humans, right?

And they're, you know, if you have a computer that is.

An agent with agency and it's making decisions who is responsible when the decision's wrong, right?

And who's responsible when the decision's like harmful.

And I think these are all like really important questions that kind of, that need and require human expertise to be available either as a form of like management of these agents or as a arbitrator of like decisions these agents make or whatever.

So, of.

There's this pressure, I think between like, people want agents because they think it costs too much to pay people, right?

And they wanna get rid of people out of the whole thing so they can make more money.

And then there's people who want agents because humans have too much work.

And there's some things that an agent could do better than a human.

Same way that there's some things a human can do better than an agent.

So.

Eli Davis: So, the challenges in assuring AI serves as a tool rather than an replacement.

So, you know, I teach at miles College and, you know, I offer I, I develop like a custom instruction chat vibe using open ai.

And I I try to get them to use that to do the work.

But what I find is they would rather have it replaced them.

You know what I mean?

So go.

So lean into that for us.

Peter Swimm: I mean, I. I, you know, on the bottom of my website I say no, the con content here was not made with gender of ai.

'cause I think like, one I like writing, you know, I, you know, I was an English major and I like, you know, I was hooked in the idea of like writing ideas and working on them.

That being said, I have a DHD, you know, it's hard for me to like, proofread and do my writing.

So I'm using Grammarly.

is like super effective for me because it, like, it helps me.

I don't see, like, when I have a missing adjective,

I just don't see it in my reviewing and it finds it and fixes it as awesome.

But I think like you have to learn how to cook before you can push the instant food button, because then you'll never know if it's good or not.

Right.

And I think like, especially like Zoomers and stuff like that, they don't.

And I think this also goes to the people who make the products.

They don't understand, like the thing that comes out of it is plausibly content.

It's plausibly, you know, things that make sense and are valuable, but is it good?

and you probably, someone reads a lot of like content across the spectrum of quality to put it charitably, right?

Like, and so like.

I think like there's that bland valley in the middle that AI is good at that, you know, people can turn in and they didn't learn anything but it like serves the function of the paper, you know, the assignment.

And I think like we are gonna have a reckoning moment as society where like there's a point of going to college or school, turning in assignments or to learn how to do a thing, you know?

Right.

And I think that's kinda like the challenge that we have right now.

Like, you know, we have a lot of, like, I talk to a lot of people in industry whose like younger siblings are in college and they're just like, but Google says it, but Google is not reliable anymore.

Right?

Because it has the AI thing at the top that doesn't know who played for the 87 Mets and event baseball players that don't exist and stuff.

There's no like difference in the presentation layer between bad quality and good quality, right?

So, I think there's a lot of interesting questions that I tell my clients and my businesses seriously consider before you engage in it because if you lose your respect of your users as something that you know is unreliable, they'll stop using you, you know, so.

Eli Davis: I think that is very important.

All right.

So, uh, so how does Toll Ville approach the balance between automation and human touch and business?

Peter Swimm: Yeah.

Yeah.

So, there's like two big problems that people come to for conversational ai.

One is,

Eli Davis: and lemme just inject real quick because that's what Toll V does.

Peter Swimm: That's right.

Eli Davis: ational ai.

Go ahead.

Peter Swimm: Yeah.

So that's our bread and butter.

And we get two kinds of customers.

you know, our big client is like externally facing.

So like, say for example, you're flying and your flight gets canceled and you get a text message from a agent that offers you a rebook, then you go into whole process.

And that's probably an agent, you know, that's probably not a person texting every single person at flight cancel that

An automated process.

And then the other kind of things are like internal scenarios.

Like, I worked for a client that is the largest employer in the United States and their healthcare is bigger than the Medicaid in 13 American states.

So, and they have all the problems for those people answered by 200 people in a call center.

In the south, and like when COVID happened, the call center had to shut down.

And so no one's picking up the phone.

So they had in a hurry.

They had to develop a system that at least would tell people, there's only 10 people on the phone right now.

We're working on getting more people up.

Here's the top answers that we can answer without even knowing who you are.

You know?

And those are kind of things that people are trying to build.

They're looking for like.

You know, there's people that wanna automate everything.

You know, they want, like I've had a person from that same company who was like, I wanna have a completely automated contact center in 45 days, or whatever.

This was half a decade ago.

It was even less plausible then.

But but the other thing is like, there's hundreds of the scenarios that are like.

Super trivial scenarios that take up a lot of time.

Like we did an analysis of the contact center and the average call length is two minutes, right?

Two and a half minutes and a minute of that was verification.

So like you just saying who you are and where you're calling from and your account number.

Right.

And so we found out that, okay, so we can do twice as amount of work if we put the bot on the front hand of the conversation and the human on the other half.

So when the human gets the call, little baseball card pops up with all the data they collected, and then you just go right to the helping versus the investigation.

You know?

So there's like.

If there's a spectrum of things that you can do to like, increase the performance of people without increasing the workload, and I think that's like when it's like a really, it's cooking, you know, that's like a good scenario.

I think that I try to encourage our clients to like look at.

Eli Davis: So, I just tried to I just applied for this fellowship and it's using, it is Fusion, artificial intelligence and AI together.

What I noticed when I was applying was that it wanted you to make sure that you use artificial intelligence.

My and you were just talking about the productivity I. What that made me to think of is, will the workload become more, because productivity will increase, will we be required to do more, to produce more and also higher quality at the same time.

Peter Swimm: You know, I think, I mean, that is a very good question to ask, and I think that's.

Anyone who's working in an organization that is adopting AI should be pushing for solutioning and arbitration.

Like, like if you think about it, like if a company is using automation to answer questions, the data has to come from somewhere.

So someone has to write the reference document.

So the employee manual has to exist.

Right.

And you can't.

At some point, there's a writer and sometimes, and at some point there's a programmer and there's no, like, there's no manner from heaven without that initial spark of human ingenuity.

So if I work for a company and companies have this like, you know, like the retail companies of the world, they have their call centers and all their email and they've kept copy of every single email they've gotten for 20 years.

So they have the data to create a map of the most likely scenarios and do things if I as an employee are responsible for 10% of that.

I'm not gonna get 10% of the revenue they save.

You know?

And I think like there's gonna be a greater awareness of the work that people do and the input that goes into it.

And there has to be some sort of reckoning of paying people, like, like open AI stole books.

End of story.

They went to the li, you know, they went to all the library systems and they downloaded books.

And they're arguing in court right now that like one book doesn't amount to anything.

It's like the corpus of the world.

That's the value.

And therefore they don't owe any money to anyone.

But the books wouldn't exist without someone to publish the books without paying the offers to write the books.

Right.

And so if you cut off that industry.

The models are not gonna approve anymore.

They're just gonna be where they are today and then stagnate.

And it kind of goes to like your students, like they're like, they don't learn how to write, then they can't write without open ai.

And so open AI now has people hooked on the $20 a month plan forever, right?

Because they can't do work without it.

And I think that's their.

Eli Davis: Plan.

Peter Swimm: their, that's the plan.

That's their ex. You don't, you, you don't burn 2 billion, you don't burn $2 billion a month without a revenue strategy.

Right.

And so, so we use these tools and, you know, for different goals.

And I think like there, there's two things that I have my I caution my clients is we increased in two years.

We went.

Super deep in quality, but it looks like that quality's plateaued.

And we're not at a hundred percent coverage yet.

We're at like 65% coverage.

Right.

And so you're gonna spend a lot of money being the Guinea pig for that last 30%.

Or you can like cut bait and invest in people and make things better today with what you got.

Right?

And so like I'm really like, you know, I don't work for open ai, I don't work for Microsoft, I don't work for Amazon.

I don't.

Have skin in the game, you know, I just want people to like have good processes and be aware of the decisions they're making and be able to measure the impact, you know.

Second.

Eli Davis: Yeah.

Yep.

All right.

Here we go.

All right.

So, the future of human AI collaboration, how do you envision the future of AI collaboration?

if you could possibly break it down to.

the people who like, like my fraternity brother, he had never even opened up the app section on his iPhone.

Peter Swimm: I guess the good news for people like that is the technology's coming for you even if you aren't like an early adopter.

Like they're putting it an Apple message and they're putting it everywhere, and there's gonna be a taste of it at some point everywhere.

The problem is like if you're not used to how it operates you may put reliance on something that is like, may not be like a hundred percent, you know, or it may be something like, for example, um, I mentioned my I'm very bad at text messaging.

And the text messaging on my iPhone used to be pretty good at where I would just type and it would just like, correct.

Right?

And it can't do it at all anymore for me, like I have to like very carefully type again because it's using AI to auto correct my words.

Incorrectly all the time because, you know, as opposed to like doing like whatever cheap version of autocorrect was 10 years ago, that worked for me.

It's using some sort of thing based on how people text, which is not how I text.

And so, it's there's a writer, Cory Doctor calls this in acidification.

So the product has been in, you know, Google's been inify all that's been in,

So.

So a lot of people will first encounter these technologies when they notice that something they relied on is much worse now.

Right?

And so those people kind have a problematic view.

'cause they won't understand why they think, oh no, am I getting old?

Can I not type anymore?

You know, that's my first thing.

And like, oh, I just can't type anymore.

But, and then I realized, oh yeah, it's, the autocorrect is much different for me.

And I think that's kind of like an important thing where, if AI is used correctly, you don't notice it, and it's just like better.

But if it's used incorrectly, you do notice it and it's frustrating.

And then there's other people who don't notice it and it's wrong.

And now you look stupid.

And so like you're turning a paper that doesn't make any sense to your teacher, and you can't explain why you wrote that because it's, you plagiarized it from a prompt.

Eli Davis: Yeah.

And it's and it's super vague.

Peter Swimm: Like I use I don't use AI tools for writing, but I do use AI tools for my outlines.

'cause I feel like that is helpful for me as a neurodiverse person to like just kind of talk through problems and make a bold list of what I want to talk about.

And then I write.

And I think that's super helpful for me.

Like, but I constantly have to tell the model to like, quote me verbatim, don't rewrite what I say because it will like write it like less spicy.

And like as a writer, you're trying to be, you're trying to be fun and spicy and interesting and, everything in the AI just kind of pushes stuff to the boring middle, right?

And so like, you don't get any of interesting insights and stuff.

You do get from things.

So, you know, it's really good for writing a business email, you know, like an email that has like a. Or anything that is just like very rote, like you need to email your landlord or whatever.

Yeah.

Like AI can write it for you.

That's beneath you.

You know, that's like not a good use of your time to like struggle with writing a proper email.

But if you're writing an email to your grandma, you know, you should write that.

Right.

You know?

Eli Davis: Right.

All right.

So, I'm a musician and I'm a writer too.

I also used artificial intelligence to, to see how I can I. Play it like I play my guitar, you know, you know, I, I know also that there's gonna be multiple iterations of these large language models and how you could write with them and all.

But I also know I want, you know, I started playing the guitar when I was 27 years old, and now, you know, I've been playing I'm 51 now, so I've been playing for a very long time.

But I also know where I started.

I. You know, and how I started with practice.

So, so what I'm trying to do is there's going to be some emergence that happens with artificial intelligence.

You know, what is going to emerge from this technology, especially when it comes to creativity, right?

I love I'm not an artist.

I can't draw, but I love to blow into artificial intelligence, you know, particularly chat GBT and see what I can create from my imagination now using this new tool, you know?

So, that just goes into the segment of now what kind of weird music do you make?

Peter Swimm: So yeah, I, so I like.

There's three kinds of music I like and I try to put it in the same music.

I like old video game music, so like old Nintendo music and stuff like that.

I like eighties new wave and electronic music and like industrial, I like heavy metal and I like surf.

So I try to like put them together in ways that don't make sense and yeah, I think like you, like, I play music because I like it.

Like all the stuff that AI is offered to do for me is like the parts I like.

So why would I want to do that?

You know what I mean?

Like, it's kinda like, if my hobby was portrait painting and then I just tell AI to make it like, 'cause it, the idea is not the idea, right?

It's the process of

Eli Davis: I.

Peter Swimm: the thing.

And like the process of like taking an hour of your day and like just learning how a song works.

And like, to me that's like.

I'm not trying to release a Top 40 song, you know, I'm not trying to like, make a product.

And I think the, that, that question of like process versus product is kind of like at the heart of using these things for creative endeavors like music.

Like I've used all the music generating tools and I've not been impressed, right?

Because this like, it's like, oh, I wanna hear a ska version of Marvin Gaye, you know, or whatever.

Like, interesting.

But it's like also like.

One, it's stolen.

You know, like I know that as artists, I know what it takes to make music and it, you know, it's difficult, but it's not impossible.

And like you can go get a guitar and you can put in your 10,000 hours and learn how thing works.

And I think that's a worthwhile human investment.

And I don't think AI takes that away.

What is cool applicants?

The ai, like I use AI to rip stems from records and I sample them.

So, you know, like I could take a baseline off a record and like automatically take out the drums and the vocals and I can isolate a bass note that I like the sound of and use that to make something new, right.

To me, I think that's like transformative and it's not derivative in the way that like, you know, putting Rihanna acapella over some be their computer made is, you know, and and I think like a lot of creative endeavors are like that.

Like, I. If I told the AI to make what is a guitar that's three 30 feet long?

Sound like, you know, like, that's interesting, right?

Because you can't physically do it and you're gonna have a computer like, just imagine what that is and make that happen versus like, you know, do a yacht rock version of you know, yacht rock, version of a journey song or something like that.

And it's just like, and I think there's a mistake that ideas are important as execution.

With these things like like people all the time, like you, you have a friend who always like, claims they invented something because they thought of an idea for something, but they didn't build it.

You know, they didn't make it so they don't get to be the inventor.

You know, like, and I think like there's something be said about like, take an idea to the end, then nurturing and all that is still kind of like an important human process.

And that's kind of like how I, you know.

I do my music because I like doing music and I use AI to write my business email because I hate writing business emails and it's good enough, right?

It's like not like, I don't think like business email format is like the number one most plagiar realizable thing in the world because it doesn't matter.

You know?

It's just like, it's just like a way to get a job done and and that's kind of like my holistic view of like.

And I've had a lot of friends who won't work with me because I have the word ai, my thing, and I, because I think there's a lot of people in the AI industry who has no respect for knowledge or IP or anything like that.

And I'm not one of them.

You know, like I, you know, I, all the logos and design my website, I pay the artists to write, you know, I don't play with, I don't play with you know, making studio.

MEMS and stuff like that because it's just like, it's boring, you know?

I like doing art and making a computer do art for me is not fun other than like, you know, a 10 minute thing where you play like,

You know, making a computer art, make a art sound or something.

It's funny, but it's not like, it's not like the process that I like to do,

Eli Davis: yeah.

I think I I feel you on that, and that's very interesting because.

I wouldn't want artificial intelligence to play the guitar for me.

Not at all.

You know?

Because I love the way the f the strings finger, the feel underneath my fingers.

I love the bounce.

I love the rhythm.

I love so much of the physical aspects of it.

For me, I think it's very interesting to see what this new tech technology can do.

And I like to mess around with it because, you know, as a, I'm a, I'm an educator and I I work in mostly marginalized situations and what I what I see is the potential to to be able to, to.

Get people to think about things in a new way with using artificial intelligence.

You know, especially for people who have, may not been able to have the kind of, I don't know your education, but you seem smart to me, you know, you know, so, but working in the inner city of Birmingham, Alabama.

And they haven't had access to that, that, to that.

You know, I think that using artificial intelligence as a tool to teach and also a tool to create to birth new ideas.

I thought it was real cool that Bill Gates, I was watching the video and Bill Gates was like, okay, this is intelligence.

Like, like this is a form of intelligence.

And now we are gonna be able to offer this form of intelligence to to me how I look at it to this community who have been just left behind, you know?

And if Go ahead.

Peter Swimm: No, go ahead.

Well, I was just gonna say that, you know, I think access to technology is like the.

The pro predominant problem of our age.

Right?

And if having access to gen AI levels, the playing field for people in like the market of the economy and stuff like that, that all means it should happen, right?

Like, I think you have to be very careful the execution of like, you know, I grew up, I'm, I grew up in the north side of Chicago and, you know, I was, I always tell people that I'm the last analog kid because like, I'm an elder millennial and like every, you know, I went to audio school for music and they taught me on tape machines, and the year after I graduated, they got.

Pro tools and computers and stuff.

So like, and like all the, you know, all the guys I came up working with are like the old guys who like in the tech industry's guys who worked on, you know, big magnetic tapes and, you know, programming and fortran and, you know, I think there's like access to things that just happens to people.

And you know, I grew up a poor white person, but I even still had the privilege of like, you know, I can clean up and put on glasses and I talk smart.

I didn't go to college, you know, I didn't, I. I, I too neuro spicy for college and I worked my way up, but like I wouldn't be able to pull that off if I didn't already look like a tech person

Eli Davis: I love that neural spicy.

Oh my goodness.

That is amazing.

I

Peter Swimm: Yeah.

but, and I think like technology is a way for people to like interface in the world that meets them more individualistically.

Like I could set up my computer the way I like it and help me do things and all that, and that's a super powerful.

Force multiplier for what I like and what I'm good at.

And I think everyone should have the opportunity to do that.

I think the difficult thing is like these open AI of the world and all that, they've taken, you know, a free and open internet and they pay all day, right?

So like open, if open AI is like all Reddit and Wikipedia and none of that money goes back to making Reddit, Wikipedia possible.

Those people are gonna start, like, stop making that content and they're gonna start giving away for free, and then no one's gonna have it.

Right.

And I think like, you know, you're probably around my age too, like where you remember before the internet and after the internet and before cell phones and after cell phones.

And there's a whole, you know, more than half of the Americans right now can't.

Imagine a world before the internet.

Like I, I had a friend, I when I used to live in New York City, I had a friend who doesn't use a cell phone, still doesn't, and like we were talking to a millennial and they couldn't get it.

They didn't understand like, well, how do you meet up with people?

It's like, well, I email them and we make plans and I meet them at that time.

You know, like, so there's all kinds of like dependence and technology that like people don't get, and.

I think there's a real risk of being overdependent on these AI technologies where the benefit of like having that force multiplier is unevenly applied and, you know, people become dependent on it in ways that other people with privilege aren't.

You know, like, like the whole agentic thing is being posited by tech CEOs who have access to the 1000 smartest human beings on the planet.

Right.

And so they think agents are, is that, you know, like everyone's gonna have access to 1000 smartest people on the planet.

It's like, no, you were the only person who gets to talk to the expensive people.

Everyone else gets the agent,

Eli Davis: Yeah.

Peter Swimm: you know, everyone else talks to the abstraction of expertise.

And that, I think that's where the big risk comes in there, you know, where it's just like, okay I get the Temo expertise and you get the Amazon Prime expertise, you know?

Eli Davis: Yeah, I I wrote an article I write articles and I publish on Medium, and I wrote an article about this idea of name brand, LLMs.

You know what I mean?

It's that it is just gonna be like the genes and I am around your age.

And when I went to school, well, you know, I don't know for sure, but from the content that you're saying I, I understand.

And and I'm from Milwaukee, Wisconsin.

What's upsta?

Peter Swimm: Yeah.

Yeah.

Eli Davis: My my sister, she she lives in Chicago.

So, well, I think that there is a, there's this moment I study black history, right?

As an educator, but particularly I'm interested in you know, the impact, the historical impact of enslavement and how that like bam and what that done and all this stuff.

Through that historical impact.

I think that this is really one of the first times that African American people or people who have been historically marginalized will be able to have.

Access to this fundamental and very new technology like never before.

I think because of cell phones and which gives you internet access and and you know, and of course, you know, people have computers and stuff.

I think that this is one of the first times that we have had.

This level of access to something that it will be as foundational or as revolutionary as the artificial intelligence.

There has been no other time in history in which we have had it.

So this is why I'm all in.

Let's see what you could do.

You know what I mean?

Peter Swimm: Yeah.

Yeah.

Yeah.

Eli Davis: I 100% understand where you coming from and making sure that human is in the loop.

You know what I mean?

That is so very important.

So, just to honor your time, Peter I just wanna say thank you for coming on amazing con conversation.

And we should probably do it again if you don't mind.

Peter Swimm: Oh yeah

Eli Davis: so, go ahead and just give the people where they can find you, where they can get you and what they can get when they got you.

Peter Swimm: sure.

Well, you can reach me on my website.

It's toy.com.

That's ITS toil, TIL.

BILL e.com.

And the one thing I wanna pitch is I do an office hours.

So every Wednesday morning on Twitch and LinkedIn and YouTube I just have like office hours, like a professor would where you can come and ask whatever questions you want and we just talk about the issues of the week.

So for people who are figuring out where they fit in this world, or wanna ask questions or maybe want to play with the stuff at work or at school, feel free to come on by and we could just talk about stuff.

Eli Davis: Sounds good.

All right, well, thank you very much.

And also thank you for sharing your creative expression and you know, one of my favorite cities in the world is Chicago, so, you know what I mean?

I spent a lot of time in Chicago.

So you just exemplify and exude that favorite city energy.

Peter Swimm: Well, I gotta say, like, I, I know we're at time, but Chicago I really credit for my approach to things, you know, where I grew up on North Side, Chicago and Devon Avenue, where every block is a different country of the world.

You know, like, you know, Indians and Koreans and Syrians and, diversity kind of prepared me like I don't know nothing and and that all kinds of people know something and it takes all kinds of people to do things.

So I think Chicago is really that kind of like, it's of all the major American cities, I think it, it has that kind of like variety of diversity.

I think that is really important for as we go into this kind of like blended society where everyone's in the same social networks and

Eli Davis: Yeah that's really interesting.

My father who have now passed away, God rest his soul, you know, but he's, he stayed around Divine Avenue on the north side of Chicago.

My sister stays on the west side, you know, so, so I'm familiar with that area.

And they have a amazing food over there.

Peter Swimm: Oh, yeah.

Yeah.

I grew up right.

I lived my whole childhood right by that little league that runs stadium.

A little miniature Wrigley field that they have there.

Eli Davis: I'm not, I don't know it like that.

Peter Swimm: Okay.

But so someone listening to your podcast because they'd be like, fill.

Alright, well, thank you for having me.

Eli Davis: Oh, man.

Peter thank you very much.

Everybody Peter Swin.