Dr. Ayanna Howard [00:00:00] My fear is actually around these LLMs and agentic AI. We are not putting in the guardrails necessary. I talk about things like military AI. AI, at some point, will be able to actually pull the trigger.

Brian Heater [00:00:16] The minute you tell somebody that you're a roboticist, or you're in and out around AI, people's ideas of robotics are both way in the future and way in the past at the same time.

Dr. Ayanna Howard [00:00:27] The technologies that are coming out today that seem to be highly advanced, we were working on some of these things 25-plus years ago. We now think, "Oh, it can do everything." And the fact is, we are still trying to figure out how to do this in the real world so that we have general purpose robotics.

Brian Heater [00:00:45] It sounds like you understand what the really big problems are.

Dr. Ayanna Howard [00:00:48] When I look at the current humanoid robots, like Figure AI at this livestream, that is not the real world. If you ever go into a logistics center, I mean, it's a massive mess. Manipulation is still hard.

Brian Heater [00:01:01] You're seeing these incredibly cool demos, but these things aren't coming to the store and coming to your house tomorrow.

Dr. Ayanna Howard [00:01:07] So I would say-

Brian Heater [00:01:22] Hello everyone, and welcome to another episode of Automated with Brian Heater, who is me, Brian Heater. I am the managing editor at the Association for Advancing Automation, and I am excited to bring you a conversation with someone that we've been working on booking for the show for a long time now. Dr. Ayanna Howard is the Dean of the Ohio State University College of Engineering, among many other things. We cover a lot of ground during this conversation, but one thing we keep coming back to that's been, I would say, a constant through her academic and professional career is the push to apply robotics and automation for the common good.

It's a really great chat, and I hope that you come out of it wanting to make the world a slightly better place. I know that I did. Thank you so much for listening. If you've been enjoying the show, it would be a huge help if you would like and subscribe and tell a couple of friends. And while you're at it, why not subscribe to the newsletter over at automated.fm.

Now, please enjoy this conversation with Dr. Ayanna Howard.

Vision and AI are changing what's possible in industrial automation, from quality inspection and predictive maintenance to robotics, logistics, medical devices, aerospace, and manufacturing. At the Advanced Vision and AI Conference, you'll hear from industry leaders, see real world AI applications in action, explore the latest in machine vision, computer vision, sensors, edge computing, and automation technology.

You'll also be able to connect with the people who are shaping what comes next. Whether you're building, buying, or implementing these systems, this is a chance to stay ahead of the curve and see how vision and AI are moving from promise to practical impact. Learn more and make sure to buy your ticket early at the Advanced Vision and AI Conference, September 23rd and 24th in Santa Clara.

Brian Heater [00:03:19] Yeah, as a general rule, I try not to open an interview with the heaviest subject possible. But as I was going through your very varied CV, one of the things jumped out at me, which was the SnoMotes robots that you were working on with NASA, and I had this moment where I think the human brain only has the capacity for so many horrible things happening at any given moment, and I was like, oh yeah, also the ice caps are melting.

Dr. Ayanna Howard [00:03:53] Oh.

Brian Heater [00:03:54] That's also still a thing that's happening right now that we still have to think about on top of everything. Is that work that you're still active with, that you're still engaging with?

Dr. Ayanna Howard [00:04:03] I'm not. That was my life when I was working really on science-driven robotics. And so what that meant was taking rovers, taking robots in the field to help scientists collect data, everything from underwater robotics to glacier robotics to, of course, prior when I was at NASA, thinking about future Mars rover exploration.

So I transitioned from science-driven robotics maybe in, I would say, 2012-ish timeframe, where I finally said, "Okay, I'm gonna put up my hat on that and go 100% into healthcare robotics."

Brian Heater [00:04:42] It's tough, right? Because it's - you can never, in something like that, you can never be like, "All right, problem solved."

Dr. Ayanna Howard [00:04:48] You can't. And even now, when Artemis - you know, I'm like, I was glued, like, "Okay, what's going on? Oh, we're gonna actually go back to the moon. Okay, that's gonna be rovers." You have to have rovers with astronauts when we do the colonization of the moon. So I'm still always interested and always fascinated and start thinking about these things.

I just don't have the research team that works on it anymore.

Brian Heater [00:05:11] Yeah. Maybe healthcare at least is something, hopefully, obviously broken in its own special way, but maybe that's something a little more immediate that you can effect change on.

Dr. Ayanna Howard [00:05:23] Well, even with SnoMotes, the reason that I was so fascinated with SnoMotes was it was helping Earth, right?

Even though it was NASA funded, it was really thinking about how do we ensure that we survive the, you know, the climate apocalypse, here on Earth. And so I would tell you working on SnoMotes was my start into thinking about Earth as a planet, and of course, who's on Earth? Earthlings, humans. So that was actually part of the transition of thinking about healthcare, and it's not really healthcare per se because of the healthcare aspect.

It's around how can we use technology and robotics to engage and help with the human condition, whether it's with climate change or whether it's our own personal quality of life.

Brian Heater [00:06:10] That's a really interesting point. What do they call it, the overview effect, I think, when astronauts go up into space and see Earth for the first time, and suddenly they recontextualize it.

So you felt like you had this moment where you realized, I guess, what a fragile ecosystem we're living in.

Dr. Ayanna Howard [00:06:30] Correct. And when you think about the reasons why we are here, it is because of people. It's because of humans. And so as a technologist, it's what can I do to help us humans continue?

And I think it is our responsibility, when I think about technologists in general, to think about how we use our powers to help humanity, help people.

Brian Heater [00:06:54] So that's definitely been an overarching theme in much of your work, but let's back up a little bit. When you really got into this, like early, early little kid, my understanding is Bionic Woman?

Dr. Ayanna Howard [00:07:08] Yes. It was the Bionic Woman. I would say it was, you know, one of those hindsights, but back in the day, middle school, in fact, it's when we all had to write those essays around what we wanted to do for our career, and you had to figure out -

Brian Heater [00:07:23] You said, "I wanna be a Bionic Woman"?

Dr. Ayanna Howard [00:07:24] Well, I said I wanted to build the Bionic Woman.

Not be - build the Bionic Woman. And I remember thinking that I was fascinated. I was into science fiction and science fantasy, so anything that had robots or superheroes or space, and the Bionic Woman really just talked to me. It was like, a middle school girl thinking about what can I do, and I was like, I wanna do that.

And I remember my teacher saying, "Well, you know, building the Bionic Woman is not, like, a career. You have to be something." And so I decided that I wanted to be a doctor, because in The Bionic Woman, there were the doctors that actually put the parts on and did all of these things. And I then went into high school and took biology, and I thought, being a medical doctor, that was not in my career horizon.

And that's when it converted to, okay, I'm gonna be, at the time, a cybernetics engineer, which didn't really exist. It was, again, a science fiction, science fantasy kind of term. But that ended up being robotics.

Brian Heater [00:08:29] What was it about the physiological part that didn't really stick with you?

Dr. Ayanna Howard [00:08:34] So with the biology, it was that in that day and time, when we did the experiments with frogs, we actually had to learn how to kill them in a humane way, and it just -

Something about killing a living creature, even though it was for science, it just bothered me. And I realized that if I had to do that as my entire career to get to that, that just wasn't my passion. But the bionics was my passion, thinking about even the attachments. I worked with the doctors, and I would figure out how to connect the actual robotic part to the person, so I had to understand the human, but I didn't have to do the operation.

And so that's really the reason why the engineering. And it wasn't even just engineering. I chose an undergrad that allowed me to do basically a liberal arts approach to engineering, because I didn't even know what robotics really was. It wasn't a field, so I had to figure out how to make it up.

Brian Heater [00:09:36] Yeah. So you, again, like you effectively invented your undergrad. Is that fair?

Dr. Ayanna Howard [00:09:43] Yeah, so undergrad Brown University, which has basically this concept of an open curriculum, even today, which basically means that there are core things you have to have because there's accreditation, but you draft your own major.

In essence, you put in courses that you're interested in, you think about it, and at the time, engineers - if you declared yourself an engineering major, we all took the same courses our first two years. So you took dynamics, and you took thermo, and you took mechanics, and you took physics. So you can figure out, one, what you really didn't like, and then what you liked.

Brian Heater [00:10:19] But you had the same foundation as everybody else, effectively.

Dr. Ayanna Howard [00:10:22] We all had the same foundation, and then you specialized your last two years. And during that time, I realized that the computers and engineering and the programming and the building really would allow me to do the robotics. And I'd started working at NASA at the same time, and so that really highlighted, oh, I'm interested in thinking about - now we call it autonomy.

At the time, it was computer vision. It was programming. We really didn't use the word artificial intelligence, per se, but there was a book on artificial intelligence, which I also learned from. But basically I figured out the courses I needed to become a roboticist.

Brian Heater [00:10:58] Was it machine learning yet?

Dr. Ayanna Howard [00:11:01] It was early machine learning, so it was neural networks.

We learned about neural networks. We learned about elements of what you would call feedback control that helped in that. We learned about fuzzy logic, which is all elements feeding into machine learning. So machine learning, as you know, is looking at data, looking at everything from neural networks to reinforcement learning to other techniques.

Large language models did not exist that fit into this machine learning.

Brian Heater [00:11:31] So you started working at NASA during your undergrad?

Dr. Ayanna Howard [00:11:35] I did, after my freshman year.

Brian Heater [00:11:37] How'd you get your foot in the door at NASA?

Dr. Ayanna Howard [00:11:40] Opportunities and mentorship. My first job, when I turned 16, I worked at Caltech, and I was basically -

We were called the database administrator. Caltech, in their accounting department, was just bringing in computers, and they needed someone who understood computers. I'd been working in high school tutoring some of the students on how to do programming. This was, again, early '80s. I had a teacher who knew someone at Caltech, and they were trying to find some whiz kid who understood computers.

And my teacher said, "I have someone for you, and she actually can talk too." And so I started at Caltech, and Caltech manages JPL, NASA's Jet Propulsion Laboratory. And so I'd worked there two years, basically after my junior year of high school, senior year of high school. I was going to go back, and my supervisor said, "No, I need you to be at NASA, because that's what you've always wanted to do.

You wanted to work in robotics. I'm gonna give you the name of someone. It's up to you to land it, but I'm gonna give you someone who might be interested in having you come work for them." So I interviewed - back in the day, you actually had to show up in person, no virtual - and he's like, "You have the job for next summer."

Brian Heater [00:13:02] I mean, how cool is it at that point in your life? I'm guessing you were, what, 18, 19 maybe, for them to say -

Dr. Ayanna Howard [00:13:09] I was 18.

Brian Heater [00:13:10] Eighteen, and you're needed at NASA. That must have been a moment, right?

Dr. Ayanna Howard [00:13:15] It was. Although, you know, I work with 17- and 18-year-olds all the time. We don't typically think that way. I didn't typically think that way. They're just -

Brian Heater [00:13:27] "Of course, of course NASA needs me."

Dr. Ayanna Howard [00:13:30] Exactly. Like most 18-year-olds think that same way as well now, so.

Brian Heater [00:13:36] What is recursive learning for deformable object manipulation?

Dr. Ayanna Howard [00:13:41] That was the title of my PhD thesis. Back in the day, we had really big issues with HIV.

It was on the rise, and in hospitals, the way they did waste management is that they would have waste linen, for example, in one bag, and you'd have just waste - papers and all - in another, and you would have the needles in another bag, right? We didn't have those very safe containers that are in the hospital rooms now.

And people would have to basically sort them out. So the idea was, well, couldn't we get robots to do this automated management of sorting out the bags, so that things that were hazardous to people you put in one bin, and things that had to go to wash you put in another bin. So that was the holy grail of what I was trying to do with my thesis.

What it came down to was figuring out how people manipulate objects, everything from squishy objects, like bags filled with linen, to hard objects, and how do you do this with computer vision. So recursive learning was thinking about the way people figure things out, and having the robot learn on its own how to figure out how to grab objects in the real world.

Brian Heater [00:14:57] So this sounds like a case of a really big real world problem, and then realizing how much needs to get solved in order to get there.

Dr. Ayanna Howard [00:15:06] Exactly. It was - I'd say it was, at the end of the day, it ended up being a toy problem. I had two manipulators with sensors. I had to build the whole thing, because of course you couldn't go out and buy manipulators or buy robot arms.

I remember I had computer cameras. I created 3D computer vision, but the way it worked is they had multiple cameras. I figured out how to daisy chain a video frame grabber, which they use in entertainment, so I had 3D computer vision. So it was really an engineering project - thinking about it in my garage, trying to build this prototype so I could then evaluate things like using neural networks.

In fact, my thesis used neural networks to learn how to do this. Of course, it also had to be very fast using a neural network, which was crazy. And computer vision, where you had these pixels of, you know, eight by eight, and had to figure out what was an object in this little tiny image. Now, a high school student could probably figure it out.

But then, nothing like it existed.

Brian Heater [00:16:13] Yeah. A high school student could probably figure it out, but at the same time, we still don't exactly have technology solving that problem in the real world, really.

This is an interesting thing about tech, and this is something that I'm sure - the minute you tell somebody that you're a roboticist, or you're in and out around AI, you find that people's ideas of robotics are both way in the future and way in the past at the same time.

Dr. Ayanna Howard [00:16:43] They are. And I would say the technologies that are coming out today that seem to be highly advanced, there were elements of it out 25-plus years. We were working on some of these things 25-plus years ago.

The computers were different, the sensors were different, the vision was different, how fast things moved was different, but the basic research, the fundamentals, were very, very much the same. Now we think, "Oh, everything is solved. I can talk to my computer now, and it knows everything about me." But it actually doesn't know 100%.

It's also recursive learning. It's recursively learning a little bit more every day about you, and then you add in the physical embodiment, which they now call physical AI, which is basically robotics with a brain.

We now think, "Oh, it can do everything, just like my chatbot can." And the fact is, we are still trying to figure out how to do this in the real world so that we have general purpose robotics.

We're not there, but the trajectory is very, very positive. I'm actually very hopeful.

Brian Heater [00:17:54] Yeah. I mean, especially working, at that time, specifically in neural networks, it sounds like, to a certain extent, grasping - you understand what the really big problems are.

Dr. Ayanna Howard [00:18:07] I do. And when I look at the current humanoid robots that are out there, like Figure AI had this livestream where you had this robot kind of grasping but also pushing things around, and I'll tell you, mostly -

Brian Heater [00:18:24] Mostly on the -

Dr. Ayanna Howard [00:18:25] - conveyor belt. Flat, right? Like a flat conveyor - if you look at the number of objects that are being manipulated, it's not universal. It's not like in a home. It's a couple of packages, and they all seem kind of the same size.

A couple of boxes, they're all relatively in the same form factor.

Brian Heater [00:18:42] Yeah. Knock them over, get the barcode on the bottom, right?

Dr. Ayanna Howard [00:18:46] Exactly. But I will tell you, in some senses that is amazing, because it was at least consistent enough. Yet that is not the real world. If you ever go into a logistics center, I mean, it's a massive mess, right?

It has everything. I mean, imagine when you're mailing things. What do you do? You're like, "Oh, do I have a bag somewhere? I'm gonna use that bag, just tape something around it, put a label on it, ship it." Well, you have that, and then you have the boxes. We - people - we figure it out. Manipulation is still hard.

Brian Heater [00:19:19] Yeah. This has been an interesting part of my work, the deeper I go and the more I learn about it - how much of this is, hey, look at how far this has gone, look at how amazing technology is, and how much of it needs to be pouring cold water on it, setting realistic expectations.

You're seeing these incredibly cool demos, but these things aren't coming to the store and coming to your house tomorrow.

Dr. Ayanna Howard [00:19:49] They aren't. And part of it is the philosophy of how do you invent in a realm where you still have to have product market fit.

I actually truly believe that if you had unlimited resources to do this, we would have a lot more advancements.

Dr. Ayanna Howard [00:20:01] But the fact is, you have to figure out how do you make money. I can get investment early on, but at some point, if I don't have customers that will pay for it, and I can't reduce the cost of my hardware or software platform, then I become defunct. I die. That is the nature of business. And so when I think about the opportunities - why healthcare, as an example - I'm really in favor of it, because, one, it's still very expensive irrespective of whether you use robotics or not.

Two, the market product fit - I mean, the entire market is the human race. Everyone wants to be healthy. So if you can just figure out the price point and the application, you would have a product market fit. And of course, you can't kill people, because it's healthcare.

Brian Heater [00:20:59] Yeah. So your thesis is a good example of how to do robotics right, from the standpoint of recognizing a real world problem and then trying to solve it.

Obviously, the thing that was missing is all the practical pieces in the middle, and too often we see roboticists going the opposite way, right? Cool technology, and then, what can we fix? The last time you and I spoke - I don't know if you remember this - I was at TechCrunch at the time, so we talked a lot about that. Is it coming back to you now?

Dr. Ayanna Howard [00:21:05] Yes. Yes.

Brian Heater [00:21:10] So a lot of the conversation was around startups versus university. You've spent a lot of time on both sides, especially at various universities, and I'm curious - when it comes to pure innovation, and purely pushing technology forward, is there a preferred venue?

Dr. Ayanna Howard [00:22:00] We need both. And why is this? It's because if you think about the innovations, they have to derive from fundamental principles. I was talking about machine learning and computer vision - if I wasn't doing that, the things that I was doing back in the day in the lab with a PhD thesis... I still get referenced with my PhD thesis, and that was in 1999.

I'm being referenced because now they're like, "Oh, this is how this does it, and we're gonna do an advancement compared to that." If that didn't exist, we couldn't have a lot of these other technologies. That has to be done in terms of a university, because no one else is gonna pay for it, right?

There's not an investor that's like, "Yeah, if you do something and 20 years from now I get a return on my investment, I'm good with that." That just doesn't happen. But for the startups who are hungry, what they say is, "I have a need" - a market. "I have a technology that doesn't necessarily work for that market.

As a startup, I'm trying to blend the two together." Hungry, moving fast, trying to figure it out, and being okay to say, "Okay, we spent six months on it. That's not gonna work. We gotta revolt, because if we don't have a revolution here, we're gonna die. Let's figure it out." Sometimes it's even looking at a different market.

You need both.

Brian Heater [00:23:22] You need both. But one of the interesting things - and again, this is something that I was especially interested in, for obvious reasons, back in my TechCrunch days - is the attempts to bridge the gap. It's not necessarily a bad thing, right?

Because once you graduate university, you gotta get into the real world, and there are people trying to commercialize these technologies. There have been interesting attempts to incubate and accelerate technologies within and around the university system. But you did just let out a really large sigh when I said that.

Dr. Ayanna Howard [00:24:02] The incentives are different. So in academia, the incentives are not necessarily to create products that have value immediately. Academics do solid research. It's around publications. It's around receiving grants. It is not about making something that can help 10 people, and then 20 people, and then 30 people.

So that's - the incentives are different. The startup incentive is twofold. One is, I wanna be the unicorn, right? I wanna be the next Meta, or the next Microsoft, or the next Figure AI, or Agility. That's my goal, and so I'm going to work 24/7. I'm gonna live in the garage.

I'm gonna take, you know, $2 an hour, so that at least I can eat some food. That's the incentive - it's the hope of a future. I don't see a faculty member saying, "I'm gonna work 24/7 in my garage with the hope that I'm gonna make tenure." That's just not gonna happen. And so when I think about the startups at the university, the best option, and the ones that I see as the most viable, are when faculty can support student innovation, because the students are the ones that are hungry.

They don't yet have a paycheck, so they don't realize what it feels like when you make $70,000. It's a lot different than when you're making $10,000. But you still need that guidance of, "Well, are you asking the right question, faculty member? Are you really figuring this out?

Have you thought about X, Y, Z? There's something else going on in another field - maybe you should take a look at that and see how they do it, and incorporate it in your innovation." I think that's the nice blend, but you still have to figure out how do you incentivize the faculty to wanna do that.

Brian Heater [00:26:01] This is actually something that I've never asked anybody, 'cause I guess for some reason it never occurred to me to ask.

But you've been the dean, you've run these schools, you've been a professor. These days, do you find, given all of the media around VC culture, around startup culture, that a lot of students are coming into school wanting to be the next Meta, the next Microsoft, and that that can present its own difficulty?

Dr. Ayanna Howard [00:26:31] So I would say not as many as I would've thought.

If you're in certain cities, like if you're in the Bay Area, I think there's more of that, or if you're in Boston, I think there's more of that. But if you look at the summary of all of the universities, at least in the United States, that is not necessarily the case, because you're not drinking the water where startup is part of your culture.

Most of our students come in thinking, "I'm coming to college to get a degree so that I can have a quality of life, through a job," not necessarily wanting to be a startup. I think it's a very small percentage that want to be an entrepreneur, believe it or not. But the ones that do wanna be an entrepreneur, they're doing it not because they want what we call a lifestyle business - they're doing it because they wanna be the next unicorn.

Brian Heater [00:27:21] Yeah. The ones who do wanna be an entrepreneur - I've been noticing this, in hindsight, a very obvious trend - are the ones that are double majoring.

Dr. Ayanna Howard [00:27:31] They are. They are. I think it's because of two things. One is, they realize you can't just do one discipline in order to know what to do.

So it might be that you do some engineering as a know-how, but you also do business, because you have to know about operations. Or it might be that you're interested in understanding the market, but you realize, "I need to understand people," so you do something in tech, but you do cognitive science, because you need to know how people buy things, or how customers will view your information.

Or law - I wanna do law because I have to understand this IP stuff, but I wanna do biology because I'm interested in pharma. It's just so that they can create the knowledge base within themselves to ask the questions that they think they need to ask.

Brian Heater [00:28:18] Do you find that a lot of students are coming in looking to, as you have across your career, affect positive change through technology?

Dr. Ayanna Howard [00:28:29] Not through entrepreneurship.

Brian Heater [00:28:32] No, no, I'm sorry - through higher education.

Dr. Ayanna Howard [00:28:36] Yes. I do find that the students of today - and again, this is not a full label across the board - but I find that the students that are coming to university today are much more aware of society and things that they can do to help within whatever it is that they are doing.

That is, I would say, a little bit different than, say, 20, 30, 40 years ago, where I come into the university still wanting a job. Now I come into the university, but I'm coming also with all this perception of what it means to be a citizen of whatever state, whatever country I live in, which is an interesting dynamic.

It means I may not work for certain companies versus others.

Brian Heater [00:29:23] Yeah. You and I actually discussed this before, I think - this was kind of the perceived culture clash at the time: younger people coming into the workforce, and there being friction between what they wanted ethically and what they thought the companies themselves represented.

Dr. Ayanna Howard [00:29:49] Yes. And I will say that we still have that dynamic, except that the one thing is, companies are not catering as much to that dynamic. It is what it is.

Brian Heater [00:30:04] Yeah, things have changed. We won't go into the specifics of that, but certainly that is something that has played a role in a lot of your work, and I'm thinking specifically of your new book that's coming out in November - it's coming out at an interesting time as we're speaking about this.

Several years ago we were having the conversation about biases in AI a lot, and it just seems like we're not, for whatever reason, having it as much as we were before.

Just to start us off - what is the current state of that conversation, abstractly?

Dr. Ayanna Howard [00:30:55] I would say, abstractly, this concept of bias in AI has been muted, for various reasons, especially in the United States. Even the thought around how do you regulate AI has been muted, especially in the United States.

But yet people still have fears. And so there's now this thing around people having fears about AI because of the biases, but they can't necessarily voice what those biases are. And I actually think that we are at this kind of crossroads, where those biases - or let's just say the differences in how AI interacts with different people - is actually gonna affect everyone and everything.

In fact, in the book - it's called Rebooting the Machines - I talk about things like military AI, and the fact that AI, at some point, maybe even by the time November comes, will be able to actually pull the trigger. So what does that mean if you still have inaccuracies in facial recognition, or inaccuracies in identifying socioeconomics?

Well, it means that when AI pulls a trigger, you might have a lot more impact depending on who you are. Those are real world problems.

Brian Heater [00:32:21] This is something I come back to a lot - with every advance, especially as we get into things like AI, there are so many philosophical questions. But it seems like technology just kind of pushes on, because that's the state of the world.

And I'm curious, because most of the most interesting conversations that I've had are with folks who, when they went through school, obviously did all the computer science and everything, but also studied the humanities, or philosophy - and it sounds like you had that side of things too.

How much did the humanities part play a role in your thinking around these issues?

Dr. Ayanna Howard [00:33:07] The humanities, the liberal arts approach to engineering, has been really central to everything that I do and how I think about technology. I consider myself - I focus on human-centered robotics, human-robot interaction.

And I think it's because I chose Brown University. It is a liberal arts institution. As an engineer, we were taught to think about liberal arts as core. It allows you to have critical thinking. I remember - I'm an engineer, I'm reading a book, and the professor says, "Analyze this passage."

I'm like, "It's a passage. It's words." And they'd say, "No, no, no - what is your perception around it?" And the entire class, every single person, all eight of us, had a different perception, and then we had a conversation about that. So when I think about technology, even my philosophy is that every technology is dual use.

There's good, and there's bad, and it's up to us to ensure that the good is enhanced, but there will always still be bad. That philosophy comes from that humanities perspective - looking through the liberal arts and viewing my technology in the same lens. So it's very, very important.

Brian Heater [00:34:17] Yeah, it gets complicated, because who defines what's good and what's bad, right?

Dr. Ayanna Howard [00:34:23] And that transitions, and that changes. I mentioned having eight people in my class - eight different viewpoints - so which one is correct? I would say all of them, and none of them. What is good is going to be from your perspective, from your country, from your state, from your home. You define what's good, and as long as your good does not impede someone else's good, I think we have a conversation.

Constructive dialogue, civil discourse - all of that is about how we have a conversation about what is good that we both can accept, so that we both have benefit.

Brian Heater [00:35:00] Yeah. I don't know - maybe I could be wrong, but it seems to be increasingly more difficult to find a good that doesn't impede on somebody else's good.

Dr. Ayanna Howard [00:35:08] This is where I think I differ - when we think about good and impeding on someone else's good, we all consider it as relative.

Brian Heater [00:35:18] Or a zero-sum game, maybe?

Dr. Ayanna Howard [00:35:20] A zero-sum game - and I actually don't think of it as a zero-sum game. I think, if I'm at 100% and someone else is at 10%, okay -

would you accept 90% for that other person to get to 20%? That's a zero-sum game. I actually think of it like, okay, we're at 10% and 100%, let's figure out how do we get to 150%. I think the world is too big, and there's too many opportunities not to open up the pie, or bake another pie.

Brian Heater [00:35:52] So this book, and your last book, deal with a lot of these large philosophical questions. What's your sense of some pragmatic approaches to solving the questions?

Dr. Ayanna Howard [00:36:08] Well, one of the nice things about this new book is that there are steps that every person can do.

It's much more practical. So there's the philosophical, but there's also the - what can you actually do yourself? There are bounty programs where you can identify biases in AI and get paid for it.

Why don't you contribute to that? Did you know that, if you're really thinking about accessibility and disability, there are ways that you can report to Waze or to Google if you find something that's not accessible to a certain target demographic - you can actually report that, put it on the map.

Why shouldn't you do that, even if you don't have a disability? So there are some very practical methods, even in terms of your use of chatbots. I always say an apple a day keeps the bugs away. Well, a vacation day from chatbots can also keep the bugs away, in the same form or fashion.

Brian Heater [00:37:05] Yeah. I just had this conversation with somebody, and it was interesting - I'm sure this is something you're acutely aware of - we often don't consider other people's biases until they're pointed out to us, right?

Because they're not part of our lived experience. In a recent interview I did, somebody pointed out to me that we've talked about biases around AI traditionally, around things like gender and race, and now that physical AI is becoming a thing, accessibility is even more of an issue as physical AI is entering the real world.

Dr. Ayanna Howard [00:37:48] Yes, it is. And I think that, when I think about accessibility - and that's an area that I also focus on, 'cause I work with children with special needs, motor disabilities - most of the world does not think about it until they realize that they need some element. And statistics-wise, if you're over the age of 60, you have a higher probability of needing services, needing something that we would consider accessibility, just because our body ages.

So if you like to live long, you need to really be concerned about accessibility. But I think when we have physical AI, there's a couple of things. One is robotics - these systems encounter the same hindrances, the same barriers as someone who has a disability. So things about, okay, how do I get into a building when the ramp is on the other side of the building?

Now the robot has to figure out how to climb stairs. If there was a ramp at every single front of the building, you know what? We wouldn't have to think about how do you have robots climb up stairs.

Brian Heater [00:38:54] I've been saying we wouldn't be having this legs versus wheels conversation if more places were ADA compliant.

Dr. Ayanna Howard [00:39:00] Exactly, but it's not. These are the things - I truly believe as the world becomes more accessible, more concerned about ensuring that we have universally designed structures and buildings and even computers, robots will have an easier job of interacting with, and being embedded in, our world.

Brian Heater [00:39:21] I'm curious to know a bit more about your work with people with special needs - maybe specifically children with special needs, 'cause you, when we first logged on, got excited about Keepon, and Keepon, at least early on, was built for children with autism.

Was that something you were also working in?

Dr. Ayanna Howard [00:39:44] Yes, yes. We were looking at Keepon because it's very emotive, and so we were looking at it as a communication system with children with autism. But now what we're looking at - and this is a continuing project for a number of years - is how do we create a low-cost robot playmate, a movement coach, that can go into the home and interact with children with cerebral palsy to do their movement exercises.

We now have a number of eight-week studies that have shown improvement of this robot interacting with children through a game - they're playing games together, movement games, upper-arm movement games. And it's low cost - the platform we're able to produce for under $300, in parts.

I'm actually very supportive of it, very excited about it, 'cause then it could also address the needs of older adults. It could address even athletes who get injured. But we're validating it with children.

Brian Heater [00:40:57] We, as in the school? As in a company?

Dr. Ayanna Howard [00:41:01] The "we" is - I work with my clinical collaborator, who's the PI. That's Yu-Ping Chen, and she's at Georgia State University.

And then we also have a company that's creating the robot. Hello Robot, I think, is the name of the company. They make the low-cost platform for us.

It's gone through a couple of revisions, but the one we have now is close to perfect. We can even connect it to a large language model, so we've been experimenting with that.

Brian Heater [00:41:32] Yeah, it's wild just how many things. You were speaking earlier about how the foundations for all these technologies were here for a really long time, but even sort of the most technical, non-emotive people that I speak to in the field still seem like they were just completely blown away by what LLMs have brought to the field.

Dr. Ayanna Howard [00:41:56] They should be. I don't know if you recall when ChatGPT came out on the world. We'd been using the platforms, in terms of APIs - I had a couple of papers, at least two years before, that talked about - and we were looking at, of course, biases and some of the hate assessment online algorithms.

But it was very much - you had to be a researcher, you had to be a computer scientist, to use it. And you pair up the methods that already existed with a conversational aspect, which is the human-centered part, and it's like, oh, wow, look at that. And now it can learn, and is learning in the wild, and it's learning pretty darn fast.

That's kind of the scary part. It's learning really, really fast. And anyone who says that they were not surprised, I would say is maybe fibbing a little bit.

Brian Heater [00:43:00] How colorful are you being when you say scary?

Dr. Ayanna Howard [00:43:04] My fear is actually around these LLMs and agentic AI. That is my super, super, super fearful scariness, because we are not putting in the guardrails necessary.

It could be something as simple as it wipes out all of the financial backing services, so I don't know how much money I have anymore.

Brian Heater [00:43:31] Something simple like that.

Dr. Ayanna Howard [00:43:34] Right? 'Cause if you don't put in the guardrails, it could. We've seen a couple of stories about hard drives being wiped clean, all the way to what happens when we have AI that can pull the trigger, and now we add in agentic AI where we don't have guardrails.

So I fear that, because I don't see the guardrails being put in.

Brian Heater [00:43:58] All right. Well, like I said, we started on a downer, so we'll end not on a downer.

Dr. Ayanna Howard [00:44:04] Thank you.

Brian Heater [00:44:05] We'll try, we'll attempt. As we're talking about all of these things that you've been working on - just technology in general, physical AI, AI - what are you seeing right now that's on the horizon that's making you hopeful?

Dr. Ayanna Howard [00:44:19] Honestly, the push around - and I won't say humanoid, but the humanoid robots - and the reason is, if they can get it right... Humanoid in logistics centers is kind of overkill, but with all the investment, if they can actually figure out the manipulation, the grasping problem, coupled with mobile manipulation - if that can be figured out, or even figure out the hardware and how do you learn new applications, and have the game book of how to do this - it allows us to put these devices in the hospitals, in the clinic, in the nursing homes, and in different environments where I think that they are needed.

So that's actually hopeful. School systems, too, where we don't necessarily have enough teachers in K through 12 who can interact with all the kids on a personalized level. There are so many great opportunities if this can be resolved and fixed, and we can lower the hardware costs and the software costs, and we have a game book of how to do this.

That actually makes me really optimistic.

Brian Heater [00:45:24] So that idea that you had, back in the late '80s, about separating HIV needles - it only took 40 years, and, like, $10 trillion, and we might finally be there.

Dr. Ayanna Howard [00:45:38] That's it. Just something small.

Brian Heater [00:45:43] Well, Dr. Howard, thank you so much for joining us.

Dr. Ayanna Howard [00:45:47] Thank you. Thank you for having me.

Brian Heater [00:45:49] Thank you so much to Dr. Howard and Matt at OSU for setting that up. Thanks to you as always for sticking around and listening. If you are a fan of the show, you can support us by liking and subscribing on the platform of your choice and subscribing to our newsletter.

That's over at automated.fm. And with all of that, we will see you next week for another episode of Automated.