Design First or Build First? – AI Field Day 6 Delegate Roundtable

I'm Steven fst here at AI Field Day, and we are giving you, uh, I guess a, a look inside the kind of conversations that the delegates have about the bigger issues of technology when it comes to ai. One of the biggest questions that has been arisen is basically, is this product useful? Is this product serving a need?

Is this product the result of design or is it just something that we have because we have it right? You we figured out how to do this. So we're darn gonna do this.

Unfortunately, in many cases, um, in society and, and humanity, we tend to do what we can do without asking first, should we do it? Or is this serving a need? And honestly, this stuff is so darn cool.

I think that that's, you know, to me it reminds me of, um, when the Macintosh came out and you had all these fonts and, and clip art and stuff, and everybody started making newsletters and banners and stuff because it was just so darn cool to do it. It was so pretty, you could print it off. And he pulled it up and he said, look what I did right?

That was completely useless. We now would laugh at that. That's kind of LLMs, right?

I mean, is it really useful to have the thing write a haiku about Hai chew candies? No, but it can do it. And so let's do it, right?

Is it really useful? All these things. So the, the question is, did anyone really ask for this?

I was just one quote, and I'm gonna shut up, I promise. You know, I you, I go back to Ford Henry Ford's quote, if I had asked people for what they wanted, we would've, they would've said faster horses. So, I mean, that says it all.

I I'll, um, I'll, I'll add to that another one, another kind of axiom that I use is when you're in this fast evolution or we, you, you're just in a startup, it's today's products are tomorrow's features and a lot of things that we see created are innovations, but they may not be a, they may not be a category, they may not be a big industry. There may be an innovation that belongs as part of something else. It may be a standalone company, whatever.

So in terms of investing in technologies and choosing the right LLM or what development tools use, it's got ai, even mine, that, that, that will probably be replaced in a year by something else that's better or been incorporated into somebody's product. Yeah. And uh, I was going to just take the approach that there's a lot of beneficial stuff coming out of ai, but I think an awful lot of it is being driven by the people who are making money off of it, selling high margin, either services or equipment.

And these, these are an awful lot of these people are saying, you really need these things to be able to go forward in life. And oh, by the way, buy some of our stuff. And it's, I, I think there's a genuine demand for it, but I also think there's demand generated by the people who potentially profit from it.

And I think that AI has amazing potential as well. Like, I think all of us do. But when we connect that to a brain computer interface, do we want that to be a read only a read write only, or a read write, execute capable system?

Um, and I think these are questions we will need to be answering. Uh, what is that need? Who, who does need that?

Uh, yes, the answer is yes, we need that. Um, the answer is yes, we should profit off of it. Uh, you know, 25, 30 years ago when the cell phones first came out and people walking around in an airport with a bag, a giant suitcase with a phone on it, you looked at that and you said, what a display of o ostentatious.

Right? And that's who ever wants to walk around with a phone on in a, a suitcase on their shoulder. And today we find the mobile devices invaluable and inseparable from us, and nobody will even go to the bathroom without their phone.

Right? The world changes and you know, it's get on board the train. That's the way it is.

So My question to everybody on a question, because, um, I can remember a couple of times I can remember Lennox World where if you wanted to see all about Linux and it was packed with all sorts of people there with all sorts of things, and most of 'em aren't around anymore. And then, um, the mware were, were when virtualization got rolling, VMworld was a place to go. And you'd see all sorts of stuff that sometimes didn't even have anything to do with virtualization, but that's where you go to see new things.

Where are we supposed to go right now to see all of these new products besides Tech Field Day? Like where can we go to a, where is there, or is there a central location to go to to find these new things? I lied.

Um, I think you mentioned it, it, the, the, the all things open folk are running an all things AI and, and it, it, they, their attendance was phenomenal. So it's a free conference. It's not an institutionalized like Lin, I remember sbo.

Yeah. But I, I think right now in Raleigh and March, they've got a all things AI and it, and you can attest to the all things open. There was a lot of AI conversations going on with IBM And the diversity of the opinion.

I think, Yeah, it's, it is, it, it will be an incredible, uh, in all trends, especially I'm speaking there, but, but it is, it is, uh, I think it's gonna be an incredible first large scale AI event. Um, Um, well, who asked for this and when? Well, as, as had been pointed out, I mean, star Trek premiered in the 1960s, right?

It's not like this AI thing is, is just suddenly along and no one's ever thought about, hey, how it might be useful and things like this. So, you know, what always happens is we get the bandwidth, we get the processing power, we get the storage, we get this, that, or whatever, get the internet, and finally we can do these things. Um, and, and the development of, of anything, you know, there are lots of products that come up that look really exciting.

Hey, the, the, the AI one I'm really excited about is AI generated emojis. I want an view on my laptop to do that. Well, not actually, right?

Things like this. So, so we kind of, we can have fun with this. Um, I do think there's been a lot of consideration for this.

We have to be careful about not, you know, for the, for the critics of it. Not just going back to the good old days, right? When we didn't have this, that, whatever, you go back to the fifties, we had polio.

Maybe they weren't the good old days after all, and things like this. So with all of this, we have to balance history. We have to see what's happened before.

There's a cycle to the computer industry, right? Over and over and over again. So for the newer people entering this, learn your history because you'll see it happen again.

Right here. 3 does have AI generated emoji, and apparently that was one of the most requested features, uh, by end users of Mac Os. And so that's the major new edition to Mac Os.

I'm just thinking back to the movie Oppenheimer from last summer, um, where Oppenheimer at, it's like two hours before the bomb's gonna go off and, uh, the general says, wait, there's a risk of blowing up the world. Well, it's almost zero. What, what do you want?

And the general goes, zero. And I'm very concerned, you know, I'm looking, I'm thinking back to genetic engineering where Dolly the sheep was cloned 25 years ago. And so far we've still said we will not clone humans as far as we know.

I'm not hearing that same of concern here, that we don't know what we're gonna release, but it's cool and we can do it really fast. And so, you know, I'm just, I I I realize it can't be zero, but it better be damn near zero before we try doing something things. And again, the reason I'm bringing up the genetic engineering is because many pharmaceutical companies are, I think we talked about this yesterday, are creating molecules faster than ever.

Dude, we gotta slow this down a bit. There is serious chances of harm. It's not gonna be the next ambient right?

Or back to X point this morning. So to, to answer a couple things that came up. So the first is, you know, where to go.

I think all things opens a great perspective. I think if you wanna also understand the larger zeitgeist of what's gonna happen, what's what we're all gonna be marketed to, like you mentioned, drugs, go to South by Southwest. Take in some of that interactive content, uh, branch out beyond just the, the purest technology in, you know, in it it's about looking at the application in market of what you might be experiencing next.

And so I think of it more multidisciplinary if science is something that explains and predicts, uh, if you think about where the, the, the very, uh, horrific origin story of a dive table from the Navy, people literally died to figure out what is appropriate for coming from a depth, not from that depth to going back up. So how we get that information, some of these simulation ideas are actually quite provocative in that what if we didn't have to do animal testing? What if we didn't have to put people in harm's way?

And what if there was a digital twin, a simulation that allowed us to do that? Um, but at the same time, think of all the other horrible things that could, uh, be done. Um, and that's the balance.

So I, I want to get it kind of refocused here though. Are we developing products, I mean, to Andy's shirt yesterday, are we designing first or are we designing last and you know, are we actually designing things that we want and then going to figure out how to build them? Or are we saying, whoa, this thing can make goofy pictures.

Let's use that. You know? Um, We, but Steven, it's, it's not a binary choice, right?

It's not. We're very, it's not, we're we're very good at two things. We're very good at moving fast and breaking things, and we're very good at bigger, better, faster, more.

And those are not incompatible with each other and design and building cool toys and are not incompatible and, and doing something and say, oh, that's cool. I can do something with this. Is, is valuable.

I, I think the question are of are we design first is an actually a valid question. That's not how we create products. It's not how we create companies, it's not how we innovate.

We take ideas and we create them and we explore. And sometimes it's very intentional and sometimes it's a post-it note, it's accidental. It depends On the Product that You're well, but, but my point is not everything is designed at the start.

So mo most things I, I would contend are not designed. And it's the serendipitous, unintended outcomes that really are the innovations in many cases. And, you know, that acts actually true.

Uh, if I can kind of jump in on that, if you look at some of the most successful products of the 21st century, certainly they were, they were not, that was not what they were supposed to be. I mean, everybody's talking about TikTok. TikTok was a karaoke app, right?

Um, Twitter, they certainly had no idea what Twitter was. In fact, I still don't think that they know what Twitter Is. That's Sure.

I know. You know, a lot of these things are just accidental and, and it's about how people use them more than it is how they intended to be used. That kind of negates your whole premise for this discussion.

Yeah. It's our premise, our Premise for that discussion. But the way you're, well, anyways, so I, I think something that's changed as well is, um, what is the intention for the product?

So these goofy video games, uh, are, uh, hang on a second. I got it. I'm sorry.

Sorry. Uh, dying. Um, so I think the, the goofy chats, filters or face filters have this, um, they look like they're goofy and they're not doing anything if they're free.

They are, we're training models to look at faces and to make adjustments to faces and to do things. So we're actually providing data for somebody else's model. So there's actually a use that we're not seeing, um, as a consumer of that application.

Right? Um, and I, I think all of it just goes back to like, yeah, there's gonna be a bunch of things. I think we've, we'll, we'll see cus companies that create chasing, um, they might see a bunch of whales that need to solve a particular problem as they move towards adopting ai and they have a clever way to solve that problem.

And they may chase that, and that's fine because they'll get absorbed into another company and the technology will get absorbed into something bigger as a platform as this whole thing matures. Did you say that whales are using ai? Yeah.

Okay. And looping this back around to the, the original premise, it's, it's like, no, we didn't necessarily ask for any of this. We're getting some of it whether we like it or not.

And the, it's, uh, I think since we are getting it, we do need to think about what we actually end up with and, and possibly make sure that we get some design put in along the way so we really do get what we want rather than getting something we really don't want, like Jay was describing, Bob, did you want in? Well, and we do have to ask what are the motivations of the people who are giving it to us, right? Mm-hmm.

So whether they're reusing the data, right, you know, it just says, was just mentioned and, and things like this. Um, Twitter became really a political battleground, right? Uh, continues to be that way.

So I mean, a lot of these tools, yeah, they're not what they originally intended to be. They evolved into something we didn't expect them to be, but we shouldn't just use them because they're cool, right? Ask, did we ask for them?

Um, why are we using them? Why are we still using 'em? People are leaving some of these services, right?

And so we should examine, uh, these types of questions. Well, and I, I agree with you, Bob, and one of the other unintended consequences that we didn't ask for definitely was that the content that we generate has become data fodder for these engines to build their, their context and information from. And I mean, that is certainly not something that we asked for.

That was not something that we understood we were getting into when we started generating content that was being consumed by these things. And when, when we created a hot or not app for college students, we certainly didn't know that it was gonna change the politics of the entire globe and set up a new advertising personalized, advertisement centric, um, mega corporation. Uh, but yet here's Meta and Facebook and, and Instagram and, and, and all that.

Uh, yeah, I I all of this is a, is a, it's a lot to think about the Consequences of any technology that we've produced over the last 2000 years when taken to extreme, uh, uh, can be very dangerous. The thing is, they have been taken to extremes because the thing about it is whoever does, whoever controls the narrative has the power. And at this point, That was the case for newspaper.

It was the case for Everything. If you look at the printing press, there was, there was, even in the Americas, they had, um, this was before there was a United States when the printing press was invented, and they were, they were printing out leaflets, which was their view of spam that were dis on the king, right? So like, this has happened, and that's what happens forever and ever.

Exactly. Whoever had, if we, you see that the, the technology is able to persuade people and educate people and give a worldview to people. You of course wanna control that.

And you want to use it for different purposes, which is why it is so critical to understand how they're using our content, how they're using our usage of an app, and um, and, and then what is becomes an app. So I just would, uh, like to remind everybody here that since we're all sort of like data scientists that remember that every disaster movie begins with a scientist being ignored. So, um, just, I constantly, you know, I'll be in a car someplace, um, and my Uber driver will tell me, this is my part-time job.

I'm really a something else. Is AI going to take my job? Right?

And we have to remember that there are a lot of people that have absolutely no idea of what this technology can do. And I, I think it's very important for us to consider that we are kind of the guardrail constructors and we need to keep shouting maybe from the top of the drive through, you know, window or something that folks, we need to listen to what other people are saying because this is this, I, I am quite frightened by someone using this without rails, you know, whatever, to climb the heights of a corporation, Go back to printing press, you know, how many people were put out to put out to pasture 'cause of the printing press. People were copying these manuscripts by hand in monasteries for centuries.

It to create like 20 books or 10, you know, a hundred books printing press comes along and now we can create thousand books, images of, of the same thing in, in a matter of days. Which is why you can they, which is why the, there's so much power in that because you could get thousands of pieces of information in front of people's Hands. Now it's millions of pieces of information for sure in a matter of seconds rather than days.

But the, but the same problem exists. People are going to have to transition to other roles. AI is gonna come along, it's gonna take a lot of the things that we take and do for granted day in, day out.

And we don't have to do 'em anymore. Yeah. We just have to move upscale.

That's fine. Yeah. Yeah.

But there's a lot more to that. I mean, we had the discussion yesterday about CEOs, right? Who, who from the survey saying, Hey, employees are all with me.

You know, we're gonna use this AI stuff and you're all gonna be fully employed and we're gonna be a great new corporation. Right? And then they turn around the Bruce chance they get and say, well, maybe not right.

Uh, I'll still be CEO but few hundred employees, we're gonna let you go 'cause we did all this sort of stuff. There are always nuances and you can't just give them carte blanche to do the latest new thing. Agreed.

Right. And truthfully, job retraining hasn't worked very well in the past. Uh, in most cases it's been retirement.

That's worked well to transition people, uh, uh, as a whole to transition a population from one, uh, industrial phase to another. Listen, I Was personally inspired from my prior job. I had 50 to 150 people working underneath me.

I was fired. 'cause I wouldn't do something that they wanted me to do. It took me years to understand how to be a consultant.

You can do this sort of stuff, but you have to take it on. You have to be able to transition, you have to move forward from wherever you were Your I agree. But it also depends on your education and your support system.

But this, this is a lot, lots of aspects. Don't Get me wrong. I saved up long and hard to get to that point Being stripped away.

Right. Anyway, there's a lot of depth. The the thing is too though that it's, it's okay that we move on.

The thing about especially a new, um, information delivery system, which is really what generative AI is at least, right? Is that you have, um, you have all, a lot of people who are very experienced that it's taking the, the crap job away from them. There should be a way that the more experienced people now are able to fill in the holes for the people that are coming behind them that don't have to do the crap jobs anymore.

Which is how all of us learned all the details. That won't be what happens. 'cause we live in a capitalist society.

But that's okay. But there's ways to make it so that the humans are doing the stuff we never could get around to. And the humans are training the people coming.

It's not, that part's not the part that scares me, part that scares me is the control of the narrative. Well, we do have to wrap. Uh, we are actually hitting the end here.

Um, I agree with you, Ray. People can change and people can adapt to new skills. I mean, obviously you have done it.

Um, as a population, it hasn't worked well. Um, as a population. We haven't done a good job of designing first and building second.

Um, as a population, we generally have done whatever we can do and let the technology drive us instead of us driving the technology. And I see that happening in AI as well. And so I think as, as was pointed out, it's, it's incumbent on technologists to try to help guide that narrative because it's coming.