Latest / Automated with Brian Heater / Matthew Johnson-Roberson on Why Physical AI Still Has a Missing Piece

Transcript

Matthew Johnson-Roberson on Why Physical AI Still Has a Missing Piece · Automated with Brian Heater

• 0:00
  SPEAKER_00: We don't live in a world where you look out your window and you see robots every day.
• 0:04
  SPEAKER_01: We're open to the possibility certainly that there will be a similar unexpected breakthrough in physical AI, but we certainly can't count on it happening.
• 0:13
  SPEAKER_00: Now for robots, it's like, well, what is the next thing I'm predicting, right?
• 0:17
  SPEAKER_00: Is there something that's that sort of atomic, that small, that I can do and then parallelized and put into a lot of GPUs and training games.
• 0:25
  SPEAKER_00: Matt, are you a vibecoder?
• 0:27
  SPEAKER_00: 100%.
• 0:28
  SPEAKER_00: Not even am I a vibe coder.
• 0:30
  SPEAKER_00: My whole desktop, if I were to show it to you, is a hundred windows of coordinated agents writing software for me.
• 0:37
  SPEAKER_00: I haven't directly written a line of code in three months.
• 0:41
  SPEAKER_00: We are trying to use AI agents to do all of the stuff that sucks.
• 0:45
  SPEAKER_00: I am ten times faster.
• 0:47
  SPEAKER_00: My ability to make decisions, think clearly.
• 0:50
  SPEAKER_00: The biggest advantage that higher ed has over doing RD in any other context is that hello everyone and welcome to another episode of Automated.
• 1:05
  SPEAKER_01: My name is Brian Heater.
• 1:06
  SPEAKER_01: I am the managing editor of the Association for Advancing Automation.
• 1:10
  SPEAKER_01: This week we are catching up with Dean Matthew Johnson Roberson.
• 1:14
  SPEAKER_01: The last time we spoke, he was the director of the Robotics Institute at Cargie Mellon.
• 1:20
  SPEAKER_01: He has since left that job and is helping to build a brand new college of connected computing at Vanderbilt University in Nashville.
• 1:29
  SPEAKER_01: He's also not ever too far from Startup Land.
• 1:32
  SPEAKER_01: He co-founded the Last Meta delivery firm Refraction AI and more recently the healthcare startup app Patients.app.
• 1:40
  SPEAKER_01: Thanks to Matthew.
• 1:42
  SPEAKER_01: Thanks to you as always for listening to the show.
• 1:44
  SPEAKER_01: If you're watching this before June 24th, 2026, please come out and see the third ever automated live on stage with Ali Agha of Fields AI at Automate in Chicago.
• 1:57
  SPEAKER_01: And with that, please like and subscribe and enjoy this conversation with Matthew Johnson Roberson.
• 2:03
  SPEAKER_01: So I had a realization uh about 10 minutes before this conversation started because you know I I associate you or have associated you with CMU.
• 2:13
  SPEAKER_01: Obviously, the last time we spoke, you were at CMU.
• 2:14
  SPEAKER_01: You're you were there for a while.
• 2:16
  SPEAKER_01: Um and uh I was on a business trip earlier this week.
• 2:21
  SPEAKER_01: I for 27 hours I was in Nashville.
• 2:25
  SPEAKER_01: There you go.
• 2:26
  SPEAKER_01: And and now and now we're on the call.
• 2:29
  SPEAKER_01: And and and and and and it dawned on me, oh, like I could have you know stayed a little bit longer and we could have actually done this in person.
• 2:35
  SPEAKER_00: Yeah, it's funny.
• 2:36
  SPEAKER_00: I have a whole setup here.
• 2:37
  SPEAKER_00: We it would have been great to have you in the office.
• 2:39
  SPEAKER_00: But uh it's great to speak to you virtually.
• 2:41
  SPEAKER_01: Yeah.
• 2:42
  SPEAKER_01: Uh I and one of the one of the things I uh I one of the questions I asked, I I was over at uh uh do you know Dexary?
• 2:48
  SPEAKER_01: Are you familiar with them?
• 2:50
  SPEAKER_01: Yeah.
• 2:50
  SPEAKER_01: So uh they're uh London-based, um they're uh inventory company.
• 2:55
  SPEAKER_01: They built what they say is the tallest AMR in the world, which I have no reason to debate because it's six feet tall.
• 3:02
  SPEAKER_01: Um and and they just opened up uh a giant warehouse, which if you haven't had a chance to I'm happy to connect to you with them if you haven't had a chance to go check it out in person.
• 3:12
  SPEAKER_01: They they opened up a large like testing facility in in uh in in Nashville.
• 3:17
  SPEAKER_01: And one of the things I asked him is, you know, what is the what's the talent pool like there?
• 3:21
  SPEAKER_01: Obviously you're you're coming from uh Detroit or I guess Ann Arbor and then um and and then Pittsburgh, really deep pools there in terms of like automation and robotics.
• 3:33
  SPEAKER_01: Uh so what what is the what's the community like over in Nashville?
• 3:39
  SPEAKER_00: You know, it's great.
• 3:40
  SPEAKER_00: It's certainly different than what we had in those other two pools, but there's a pretty vibrant uh tech scene here, you know.
• 3:46
  SPEAKER_00: So we've had some big corporations move recently, and I'm uh like I was in the last two places at a university here at Vanderbilt, and uh I will say um that's the through line.
• 3:58
  SPEAKER_00: Uh if you meet sort of kids that are getting into college today, uh they're the best talent pool um that I've ever seen.
• 4:06
  SPEAKER_00: Certainly I don't think I would get into college now, uh, given how uh challenging and how accomplished um sort of undergrads are these days.
• 4:14
  SPEAKER_00: But no, it's a great question.
• 4:16
  SPEAKER_00: You know, Nashville's been one of these places where healthcare has obviously been the dominant technology for a long time, much like uh robotics became sort of dominant in Pittsburgh.
• 4:24
  SPEAKER_00: And so we've been thinking about what does it mean to have um a tech-focused ecosystem, or at least a part of one here in Nashville.
• 4:32
  SPEAKER_00: And so um I think that's the question before us, and that's the hope is that what I'm trying to build here with um the university can be the um sort of the accelerant on what is already um a growing and sort of more vibrant uh AI and um computing sort of hub.
• 4:51
  SPEAKER_01: That's interesting as well, because um I'm now realizing that healthcare has also been kind of a through line for you, Pittsburgh.
• 4:58
  SPEAKER_01: Again, obviously, you know, uh you've got Pitt over there, you've got world-class healthcare facilities, and then you like your current startup, you actually have a healthcare startup right now.
• 5:10
  SPEAKER_00: That's exactly right.
• 5:11
  SPEAKER_00: So uh we have uh a startup we call patients.app, but it's an AI um healthcare startup designed to help uh clinicians basically review lengthy medical records and make medical decisions faster and better.
• 5:25
  SPEAKER_00: I mean, I guess the commonality is that um uh my wife is an anesthesiologist, and so uh, you know, finding employment for both of us in the same city means it has to be strong in healthcare and strong in tech.
• 5:38
  SPEAKER_00: And um, either fortunately or unfortunately, the number of opportunities to do that is is pretty narrow.
• 5:44
  SPEAKER_00: And so I kind of bounce between places with world-class uh healthcare facilities and growing and ever improving uh technology scenes.
• 5:52
  SPEAKER_01: Is so is she connected to the startup in some way?
• 5:56
  SPEAKER_00: Yeah, she serves as a medical advisor.
• 5:58
  SPEAKER_00: She's helping us out there.
• 5:59
  SPEAKER_00: Um and then we're trying to do um, you know, really the the sort of genesis with this startup is actually uh sort of watching her workflow.
• 6:07
  SPEAKER_00: So she'd come home uh and then have to do a bunch of charting in an uh an electronic health record.
• 6:12
  SPEAKER_00: And as opposed to hanging out with me and my daughter, uh, she'd be sitting there finishing charts from uh that day.
• 6:18
  SPEAKER_00: And I thought, oh, there has to be a way that AI can make this better.
• 6:21
  SPEAKER_00: Uh so very selfishly, uh, that was why we decided to build this uh startup.
• 6:25
  SPEAKER_00: At least my uh motivation for building a startup was I'd like to reclaim some of that time to spend with my family as opposed to uh sitting there entering stuff, watching her enter stuff in uh in EHR.
• 6:36
  SPEAKER_01: You you had that infomercial moment, there has to be a better way.
• 6:39
  SPEAKER_00: That's right.
• 6:40
  SPEAKER_00: Exactly.
• 6:41
  SPEAKER_00: I I mean it was to watch, um, and I don't know if anybody uh if you have any physicians in your life, but to watch what it is for them to have to use an interface that looks like it is from 1992 and click through about 100 to 150 screens just to close out one patient, I thought, you know, this can't be the future.
• 6:59
  SPEAKER_01: Yeah, I'm I'm hoping, you know, it's uh I I obviously I follow um automation really closely and and I over the past uh actually I you know the last time I th I I think really when you and I really first started speaking was really right around the pandemic and we were seeing you know the ways in which um the the the pandemic was transforming automation and accelerating a lot of things.
• 7:22
  SPEAKER_01: Um I I would like to believe that uh that the pandemic has had a hopefully somewhat positive effect on I guess I would say the American healthcare system.
• 7:34
  SPEAKER_01: That's a sticky subject right now.
• 7:36
  SPEAKER_01: But um we've seen a little bit of automation there.
• 7:39
  SPEAKER_01: You know, there there are a handful of uh beyond robot surgery, there are maybe like a handful of interesting companies in the space right now.
• 7:46
  SPEAKER_01: But um ha has there been progress, technological progress?
• 7:52
  SPEAKER_00: Certainly some, right?
• 7:53
  SPEAKER_00: So I mean if you if you ask me, I think that the biggest thing uh at least of the last couple of years coming out of the pandemic in terms of transformative technology in the healthcare space was sort of AI scribes, right?
• 8:04
  SPEAKER_00: So that is an area that grew really quickly, um, has taken uh a ton of land and now is sort of um becoming the standard of practice for almost all the major um academic medical centers and increasingly in in in small facilities as well.
• 8:20
  SPEAKER_00: Um and so that I think is gonna help, but you know, one of the challenges we realize is that even if you can get um the entire so the the way this technology works, right, is that they your doctor will whip out your phone or their phone and they'll put it down and say, Are you okay with me recording?
• 8:35
  SPEAKER_00: They'll record, and then everything you say to the doctor gets transcribed and hopefully turned into um something they can use as a note to kind of register this encounter.
• 8:43
  SPEAKER_00: And and we used to do this with humans, um, you know, traditional scribes, and now there's these AI versions of them.
• 8:49
  SPEAKER_00: Um but one of the things that I think um you still see is that much of the healthcare system is filled with inefficiencies.
• 8:57
  SPEAKER_00: Uh that it's not that we don't have technological solutions for, but it is just a much knottier problem in that um faxes are still the dominant way we exchange information between uh healthcare facilities, right?
• 9:09
  SPEAKER_00: And that's crazy to think about.
• 9:10
  SPEAKER_00: It's not as if we haven't built better technology than the fax machine, but um regulation and risk aversion and concerns about privacy means that fax machines still dominate um the way we move records around.
• 9:24
  SPEAKER_00: So I think there's a lot of headroom uh for technological growth uh in the healthcare space.
• 9:29
  SPEAKER_00: And in fact, um one of my biggest lessons coming out of the pandemic was that um, you know, building a hardware company is just hard.
• 9:36
  SPEAKER_00: It is a grind.
• 9:37
  SPEAKER_00: And I thought, uh, whatever the next thing I do is, it's gotta be something where we can flip a couple switches on AWS and then you know have it take off, as opposed to uh grind it out um building uh sort of generation after generation in a factory.
• 9:52
  SPEAKER_01: Yeah, uh sure.
• 9:53
  SPEAKER_01: I mean, hard obviously hardware is hard.
• 9:55
  SPEAKER_01: Obviously, robotics are are incredibly hard, but you know, you are as familiar with anyone as that the with the kind of like 99.99999% problem because you come from autonomous driving.
• 10:07
  SPEAKER_01: You know, you you had your startup in space, and then again you come from uh like Ford, you know, Detroit, you come from that world as well.
• 10:15
  SPEAKER_01: And I and I assume that healthcare, especially when we're talking about AI in healthcare, has a very similar 99.99999% accuracy issue.
• 10:25
  SPEAKER_00: Exactly.
• 10:25
  SPEAKER_00: I mean, this is the reason why it felt like an interesting place for us to do some experimentation next.
• 10:30
  SPEAKER_00: Um, you know, coming out of self-driving, one of the things that became really clear is that you, A, to your point, need to have a very, very reliable system that handles edge cases out to sort of mini decimal places.
• 10:41
  SPEAKER_00: But secondly, like I think a lot of the lessons learned for me, and I think that we've learned as a community in the self-driving car space is that in kind of safety critical technologies, there's both the technological challenges, but there's also a lot about acceptance, about safety, about working in a heavily regulated industry.
• 11:01
  SPEAKER_00: And that knowledge I think is really important for anybody that wants to innovate in any of these spaces.
• 11:06
  SPEAKER_00: I'll I'll include transportation in that in the same way I would include healthcare, is that, you know, when people's lives are on the line, it's really, really important to kind of bring an ethos of responsibility.
• 11:16
  SPEAKER_00: And, you know, people that work in sort of these legacy industries, in many ways, have come up with that as the way that they think about uh everything.
• 11:25
  SPEAKER_00: So it's not just transportation, you know, you can look at mining, you can look at oil and gas.
• 11:31
  SPEAKER_00: These are all industries that have, you know, factory work.
• 11:33
  SPEAKER_00: These are all industries where, you know, uh you put on your high-viz vest, you go through a very detailed um, you know, safety onboarding just to take a factory tour.
• 11:43
  SPEAKER_00: And that level of sort of rigor is obviously something that I think is uh really important and um certainly really important when you're dealing with a with a safety critical industry.
• 11:54
  SPEAKER_01: Aaron Powell Yeah, legacy, and maybe maybe I'm uh overstating this, but you know, the word legacy or the phrase legacy industry kind of strikes me as being a situation in which they might perhaps be uh almost hesitant or reticent to adopt a new technology.
• 12:12
  SPEAKER_00: Yeah, you know, I I think if I think about self-driving, it was a great example where it really forced uh the auto industry to think deeply about well, what degree of our workforce, what degree of innovation is necessary for us to remain relevant as a company.
• 12:28
  SPEAKER_00: And I think that that actually pushed, even for consumer cars that aren't really focused on self-driving features, it pushed those platforms and those companies forward, you know, um, uh probably a decade faster than they would have um gotten up to speed otherwise.
• 12:41
  SPEAKER_00: And I think that that's um a great lesson.
• 12:44
  SPEAKER_00: You know, healthcare is such an interesting one because uh there are lots of things we could do to make our healthcare system better.
• 12:51
  SPEAKER_00: Some of those are technological, many of them are our policy and sort of much more naughty and complicated.
• 12:56
  SPEAKER_00: But I think at the end of the day, like all these systems have a lot of commonality, and I'll put education in the same bucket as well.
• 13:04
  SPEAKER_00: They involve a lot of people.
• 13:05
  SPEAKER_00: So if you're building cars, that involves a ton of people.
• 13:07
  SPEAKER_00: You have a massive workforce, the warehouses have a ton of people in them, even today, even with the levels of automation we're seeing in car assembly and manufacturing.
• 13:14
  SPEAKER_00: Uh, in healthcare is probably the greatest example where um that is the sector of the economy that's still hiring uh while other sectors have slowed down.
• 13:23
  SPEAKER_00: And that's really because the delivery of healthcare is still a very human business.
• 13:26
  SPEAKER_00: And we can help with technology and we can do other things in robotic surgery, but there are eight, 10, 15, 20 people in the room when you're doing a surgery.
• 13:33
  SPEAKER_00: There are people bringing stuff in, bringing things out, uh monitoring different subsystems, all kinds of stuff.
• 13:38
  SPEAKER_00: And so it's just a very labor-intensive business.
• 13:40
  SPEAKER_00: And the same has been true for education.
• 13:42
  SPEAKER_00: So um, you know, we've gotten all kinds of things to help us teach better and uh educational tools, but my biggest bottom line expense at Vanderbilt here is still people.
• 13:54
  SPEAKER_00: And uh all these efficiencies have made my people better, uh more effective, my my faculty, you know, um better researchers, but we're still using people um uh all the time.
• 14:05
  SPEAKER_00: And so I think that that that's sort of the through line.
• 14:07
  SPEAKER_00: Like there's a lot of value in in what we can do with technology in these different areas, but I think it's it's still incredibly dependent on lots of humans.
• 14:14
  SPEAKER_00: And so I think that's been a big rate limiter for um change.
• 14:19
  SPEAKER_01: Yeah, you touched on something interesting in there uh as far as uh the auto industry and self-driving cars.
• 14:26
  SPEAKER_01: Um and I I'm I'm curious to get your take on this.
• 14:29
  SPEAKER_01: Uh as far as whether, you know, each uh uh certainly with like the large automakers, um whether or not they had specific success in the self-driving space, whether their efforts still uh moved them forward individually.
• 14:48
  unknown: Yeah.
• 14:49
  SPEAKER_00: I I mean this is just my opinion.
• 14:50
  SPEAKER_00: I I don't know, but but what I would say is that you know, uh we looked at who they started hiring um during those periods when they sort of got really started to get really worried about sort of uh the degree to which self-driving was going to eat into their core business.
• 15:04
  SPEAKER_00: And they brought a lot of people in who I think had a very um, I guess for the auto industry, sort of radical mindset and you know, thought a lot more about sort of rapid iteration, thought a lot about how technology can be an accelerant.
• 15:18
  SPEAKER_00: And we've seen also sort of internationally, right, like just an incredible amount of pressure on um uh sort of Western automakers from just the sort of rapid expansion and advancing of Chinese automakers, right?
• 15:31
  SPEAKER_00: Like you just go to China now and you get into any car and it feels like you're 10 years in the future from where um what you can buy off the line here.
• 15:38
  SPEAKER_00: And I think all those pressures have forced, I think to some degree, the auto industry to think a lot about well, you know, what what type of people do we need to hire?
• 15:47
  SPEAKER_00: What type of executives, what type of group leaders, what type of engineers?
• 15:52
  SPEAKER_00: And I think in the era of AI, you just uh I'm hoping that that kind of all comes together in a way that pushes um that industry forward with the same velocity and the kind of same um I don't know what you want to call it, but uh urgency that that the self-drive, like the first sort of uh boom of self-driving cars in the last 10 years did.
• 16:16
  SPEAKER_00: I I think it still remains to be seen, but you can go and I think look at any major Chinese automaker and you can see the impact of self-driving on the entirety of those that vehicle stack, right?
• 16:27
  SPEAKER_00: And everything from you know uh payment systems to personalization to the entertainment to the navigation to even the self-driving features, right?
• 16:37
  SPEAKER_00: And it's just it's very it's very impressive.
• 16:39
  SPEAKER_01: Yeah, so I I I've gotten a uh much better insight into manufacturing than than I had in in previous positions.
• 16:46
  SPEAKER_01: And you know, something that I've learned, it's now you know very obvious in hindsight, is uh when introducing new technology into um well, both manufacturing and logistics, uh there is uh hesitancy and there is pushback.
• 17:02
  SPEAKER_01: And um basically they want you to disrupt things with minimal disruption, right?
• 17:08
  SPEAKER_01: I mean they you know they m minimal downtime, you know.
• 17:11
  SPEAKER_01: Yeah, they uh a factory wants to continue to be able to produce things while introducing a new technology.
• 17:16
  SPEAKER_01: And I'm wondering if if there's an analog there for all these other industries that we're talking about.
• 17:22
  SPEAKER_00: Yeah, I mean it's a it's a great question.
• 17:25
  SPEAKER_00: Like, you know, you can one of the I guess the best examples I had of this is you know I got to tour some of you know Amazon's fulfillment centers, and you know, they were on their seventh or eighth generation of AMRs that were you know driving around.
• 17:41
  SPEAKER_00: But they were able to the Kiva ones, yeah.
• 17:43
  SPEAKER_00: They're able to show me sort of like the whole history of like how they had gone.
• 17:46
  SPEAKER_00: And to me, it was the greatest distillation of like when a robotic system is clearly really useful and it drives um the economics for a company, that you can bring a very disciplined and sort of iterative approach, but that like you can keep the factory floor humming, but bring out you know what are eight or nine generations of a vehicle and have them swap in and and kind of continuously improve.
• 18:09
  SPEAKER_00: And so I think that like companies that are doing this really well have kind of brought that kind of continuous iterative design principle to the physical world.
• 18:17
  SPEAKER_00: And um that I think is just is really what's necessary to both kind of continue to innovate, but also like you know, keep the lights on, right?
• 18:25
  SPEAKER_00: Because it's not an example where you can kind of shut everything down, you know, turn fulfillment center off or factory off or a warehouse off for six months to upgrade it.
• 18:33
  SPEAKER_00: Like that's that's really not uh viable when you're talking about the amount of capital they have sunk into these things.
• 18:39
  SPEAKER_00: And so again, I think anybody working in the hardware space, um there's just a real power to being able to have something run, learn what you can from it, and then upgrade it um, you know, as as as it continues to proceed.
• 18:53
  SPEAKER_01: Yeah, I I know that all these things tend to be sort of like cyclical and and they ebb and flow, but it strikes me lately, especially with the um I I I I do a newsletter and I've I've been interviewing a lot of um a lot of founders and a lot of CEOs, and man, it seems like last mile delivery companies can be getting a lot of press again, right?
• 19:13
  SPEAKER_01: I mean, you know, I just interviewed uh uh you know Ali from from Serve and C the CEO of of Coco.
• 19:20
  SPEAKER_01: Um you know, you were doing that or trying to do that a long time ago.
• 19:24
  SPEAKER_01: Were in a certain sense, we're was was refraction just a little too early?
• 19:30
  SPEAKER_00: I think so.
• 19:30
  SPEAKER_00: I don't know.
• 19:31
  SPEAKER_00: I mean it's a great question, right?
• 19:32
  SPEAKER_00: You try to do the postmortem and figure out what you know what you would do differently.
• 19:36
  SPEAKER_00: Um I think it just comes down to like there's a really hard problem um in scaling.
• 19:42
  SPEAKER_00: And so I guess like you let's use the last mile companies as a good example.
• 19:46
  SPEAKER_00: Like even if you look at companies that are incredibly well capitalized in the full-size self-driving space, like take a Waymo, right?
• 19:53
  SPEAKER_00: Like think about how slow and deliberate their rollout is and how long it takes them.
• 19:57
  SPEAKER_00: So they're actually driving around here in Nashville now, but they're not offering rides.
• 20:00
  SPEAKER_00: Um, and they've been driving around at least for the last six months, maybe year, um, I guess just doing mapping, presumably uh in preparation for a launch here.
• 20:09
  SPEAKER_00: Um but that's a lot of investment and time, resources, and people to uh maybe field, I don't know, what they're gonna have here, let's say a couple hundred cars.
• 20:19
  SPEAKER_01: And so um, you know, I think we're and they're and they're I mean, back up and say like that they are a Google company, so they have the resources to do that.
• 20:28
  SPEAKER_00: Essentially almost infinity resources relative to anybody else playing in this space.
• 20:32
  SPEAKER_00: And even there, it's like, you know, they're talking about bringing on, I see probably 50 or 60 cars floating around here, but that's still a pretty modest number when you think about like the size of uh the Ride Hail business.
• 20:44
  SPEAKER_00: And so that and then you know, look at the last mile um delivery space.
• 20:47
  SPEAKER_00: It's like um we just haven't seen or figured out exactly what the formula is, I think, to scale to tens of thousands, right?
• 20:57
  SPEAKER_00: And I, you know, I'll give a shout out to Starship.
• 20:59
  SPEAKER_00: I mean, Starship was certainly the furthest along of any of these, and they were pretty widely deployed on campuses.
• 21:04
  SPEAKER_00: Um I don't know if they still are, but but that to me was an example.
• 21:09
  SPEAKER_00: Like even there, we were talking, you know, thousands of robots, not you know, hundreds of thousands.
• 21:13
  SPEAKER_00: And probably in like an average day here, there'll be uh, I don't know, I got 10 things delivered this morning, right?
• 21:18
  SPEAKER_00: I got these mics delivered this morning.
• 21:19
  SPEAKER_00: So I don't know, Amazon came to my house three times, right?
• 21:22
  SPEAKER_00: Yeah.
• 21:23
  SPEAKER_00: You know?
• 21:24
  SPEAKER_00: And so there's probably, you know, hundreds, if not uh millions of deliveries happening.
• 21:30
  SPEAKER_00: And um the question becomes like, well, how do we get to that level of scale?
• 21:35
  SPEAKER_00: And I I'll keep going back to like the, you know, when I saw the level of complexity and sort of um the rollout of those Kiva robots and these Amazon fulfillment centers, right?
• 21:45
  SPEAKER_00: That's like a 15 year process for them to get to the point where they have, you know, whatever a million of them driving around.
• 21:50
  SPEAKER_00: And I just don't know what the what the arc is going to be for last mile.
• 21:54
  SPEAKER_00: I mean, I'm still a big believer, like there's no reason to think we can't do this.
• 21:58
  SPEAKER_00: It's just um It always feels like it's uh a couple years away.
• 22:03
  SPEAKER_01: Yeah, yeah.
• 22:04
  SPEAKER_01: There's so many there's so many unknown unknowns when it comes to that.
• 22:08
  SPEAKER_01: And especially when it comes to like urban spaces and and and just and interacting with humans, right?
• 22:16
  SPEAKER_01: I mean that's the biggest unknown of all.
• 22:18
  SPEAKER_01: Like, you know, and and I'm you know, I'm I'm I'm who was it?
• 22:21
  SPEAKER_01: Was it was it serve, I think, that's like trying to operate in like downtown Chicago?
• 22:25
  SPEAKER_01: Like it's a very densely populated area, like it's really hard to plan for that.
• 22:31
  SPEAKER_00: Yeah.
• 22:31
  SPEAKER_00: I think it's one of the big challenges, right?
• 22:33
  SPEAKER_00: Like, I think uh, you know, well we just you know, and this is the thing I think about a lot.
• 22:41
  SPEAKER_00: Like, we don't live in a world where you look out your window and you see robots every day.
• 22:46
  SPEAKER_00: And I don't know how long it's gonna take until we get there.
• 22:49
  SPEAKER_00: I think the the biggest example of this now is that in many cities, right, you look out your your window and you see uh Waymo or uh Zooks or uh, you know, pick your favorite um full-size self-driving car company.
• 23:01
  SPEAKER_00: And I think that that's like a big step forward.
• 23:05
  SPEAKER_00: So I'm hoping that that is part of a larger arc where like we will see robots moving around doing useful things in the next couple of years.
• 23:13
  SPEAKER_00: Um but again, it's one of these things where um until that takes hold, a lot of what we're seeing is sort of experimentation.
• 23:20
  SPEAKER_00: And um I just hope that one of these lands and hits and that then we, you know, we really get to see the scale that kind of we've been promised.
• 23:29
  SPEAKER_01: Yeah, I I I it really because you you said something similar to that um one of the first times that you and I spoke it, and and that has stuck with me, and especially lately, because one of the key things that we've been talking about, we've had we've we're now using the term physical AI.
• 23:45
  SPEAKER_01: I think we all sort of like settled on that as as the phrase du jour.
• 23:50
  SPEAKER_01: Um and and as I'm interviewing all these these physical AI companies, you know, our our kind of recent pet topic is is the flywheel issue.
• 23:58
  SPEAKER_01: And and that I think speaks that really well.
• 24:01
  SPEAKER_01: Is we're we're in this weird space right now where um we have the problem of we don't have enough robots deployed to collect the physical physic enough physical data to train them well, but we also don't have good enough physical data to deploy that many robots.
• 24:19
  SPEAKER_00: Right.
• 24:20
  SPEAKER_00: Yeah, look, you know, I think this is a good example of like where there's this big gap um in in what we're promising and what we're gonna be able to deliver on the physical AI axis, right?
• 24:29
  SPEAKER_00: Um I guess there's two I have two schools of thought on this.
• 24:34
  SPEAKER_00: I guess the first is that it is incredibly promising as a narrative and even as hopefully a reality that we would be able to move away from the world where we need to, for lack of a better term, hand design every algorithm for every person we have in robotics, right?
• 24:50
  SPEAKER_00: And so we just say, okay, well, here's the thing, let's get a team of engineers on it, let's work through it, let's uh, okay, that doesn't work.
• 24:56
  SPEAKER_00: Uh, try this, try that, try that, right?
• 25:00
  SPEAKER_00: And so uh, you know, the promise of this idea that, oh, well, we'll get enough training data and then we'll have a general purpose intelligence for physical systems that will allow us to solve many problems is incredibly promising.
• 25:12
  SPEAKER_00: And I think that's become more promising that like we've kind of gotten there and been able to achieve that with language models, right?
• 25:20
  SPEAKER_00: Certainly not all the way there, but but the capabilities of these language models relative to like what we thought they were going to be capable of or what we kind of were explicitly training them to do has been incredible, right?
• 25:31
  SPEAKER_00: So that's sort of been game-changing in many ways.
• 25:34
  SPEAKER_00: And that's the promise for physical AI companies, but but you kind of highlight, I think, where the big gap is is that we got onto some flywheel with language models where it just became we had enough data, we had enough of an understanding of how to train over that data with an objective function that expressed something that we could optimize against and get better at.
• 25:58
  SPEAKER_00: And you can do that.
• 25:58
  SPEAKER_01: I would have the major caveat there that that that data was all pre-existing because like they had libraries and books and internet.
• 26:05
  SPEAKER_00: Agreed.
• 26:06
  SPEAKER_00: And and look, you know, uh it's not that that's the same, but like one of the things that I struggle with when I hear people say, well, we don't have enough robot data, is I could take a robot, put it on the street today, walk it around for six hours or 12 hours or 24 hours, and have terabytes, you know, hundreds of terabytes of video data, of encoder data.
• 26:30
  SPEAKER_00: And so I think that like I would put the problem more specifically on we don't have an objective function that we can optimize over that tells me like how to use that data efficiently or effectively, right?
• 26:44
  SPEAKER_00: So like it's not the volume of data in my mind, but it is like, well, what's the utility of that data and how do I build something that learns on it and learns on it well, right?
• 26:53
  SPEAKER_00: And and that to me, like no one uh, you know, whatever, I try to follow what everybody's doing, and you know, and no one has answered that question to my satisfaction, where they say, oh, here's the objective function, and we just need to go gather a bunch of data to optimize that objective function.
• 27:11
  SPEAKER_00: It's not clear what that objective function is for the physical world.
• 27:13
  SPEAKER_00: And it's also not clear that like even if we had it, that it can be sufficiently optimized against.
• 27:18
  SPEAKER_00: So that's sort of the big open question in my mind there.
• 27:23
  SPEAKER_01: Can you expand on that a little bit?
• 27:25
  SPEAKER_01: When when you say objective function, what what what do you I mean, obviously it's hard to give an example because we don't have to do that?
• 27:31
  SPEAKER_00: No, no, but uh I'll try to give it make it really concrete, right?
• 27:34
  SPEAKER_00: So like uh not to distill or not to oversimplify like the last decade of incredible work in in large language models, but yeah, uh essentially you can predict the next word that's gonna come.
• 27:49
  SPEAKER_00: And that's a very simple objective function to express and to optimize over.
• 27:52
  SPEAKER_00: And it turns out if you do that over like an internet size of data, that you get things that are really smart, right?
• 27:57
  SPEAKER_00: So like that's like a very simple distillation of like kind of a lot of the work that happened over the last decade.
• 28:03
  SPEAKER_00: Now for robots, it's like, well, what is the next thing I'm predicting, right?
• 28:08
  SPEAKER_00: Is there something that's that sort of atomic, that small that I can do and then parallelize and put into a lot of GPUs and train against?
• 28:16
  SPEAKER_00: And yeah, like you know, there are examples of what we're doing.
• 28:18
  SPEAKER_00: Can I predict where this arm should be, where this leg should be, joint angles, can I predict the next frame coming out of the camera?
• 28:24
  SPEAKER_00: Can I predict X or Y or Z?
• 28:26
  SPEAKER_00: Um But I promise you, none of those are as atomic or as easy to represent or as easy to test against as is the next word right?
• 28:36
  SPEAKER_00: And so we just haven't come up with that thing.
• 28:40
  SPEAKER_00: And so it's not clear to me that more data is gonna get us out of that problem.
• 28:44
  SPEAKER_00: So until we have that thing, and that thing, whatever that is, whatever the next best word problem is for robotics, until we have that, I'm not sure that all of the you know trees that we're setting on fire uh to burn lots of uh GPU hours to train models is is gonna bear fruit.
• 29:04
  SPEAKER_00: Um but you know, uh I've been wrong many times before in my career, and I look forward to being proven wrong in this case as well.
• 29:10
  SPEAKER_01: Yeah, well, the the uh the burning planet conversation is a good one, but probably um one for a a different podcast.
• 29:20
  SPEAKER_01: But uh yeah, I that that that's interesting because the next word problem, I mean, that's um it's it's universal and it's consistent, right?
• 29:29
  SPEAKER_01: And it's something that that you need with every single kind of large language application.
• 29:34
  SPEAKER_00: Exactly.
• 29:35
  SPEAKER_00: Well, and and you know, uh despite all the complexity and everything we're doing, if you go to like you know, the frontier model companies, you go to Google, you go to Anthropic, they spend much of their resources just improving the models on that specific problem.
• 29:48
  SPEAKER_00: Yeah.
• 29:48
  SPEAKER_00: And then they do post-training and lots of other cool stuff after that.
• 29:51
  SPEAKER_00: But like what they spend uh, you know, whatever,$400 million training last week on is trying to get a better next best word predictor uh for lack of a better uh it's like a it's like a sol it's like a solved problem that's still big refined.
• 30:04
  SPEAKER_00: Exactly.
• 30:05
  SPEAKER_00: And it it is they are continuing to bear fruit in in making the models better at doing that.
• 30:11
  SPEAKER_00: And it feels like to me that we're still casting around for what that problem is in robotics.
• 30:17
  SPEAKER_00: And until we put our fingers on it, we may be going down sort of rabbit holes that don't actually end up getting us sort of our universal robotic brains that we're hoping for.
• 30:26
  SPEAKER_00: Now, if we stumble on that and we get the same kind of unlock that the language models did, right?
• 30:32
  SPEAKER_00: Like, you know, game changing.
• 30:33
  SPEAKER_00: And so I think it's still a worthy place to spend lots of time, resources, and have lots of smart people chasing after.
• 30:39
  SPEAKER_00: Um, it just feels pretty like non-linear to me, right?
• 30:43
  SPEAKER_00: If we find it and then we can spend billions of dollars training against it.
• 30:47
  SPEAKER_00: And if we don't find it, we can spend billions of dollars on training and get models that don't really do anything terribly useful.
• 30:53
  SPEAKER_01: Yeah, I I love uh, you know, I listen, I was a uh uh creative writing major at UC Santa Cruz, right?
• 31:00
  SPEAKER_01: So like a lot of this stuff is like is is sort of it's it's foreign to me, or at least I'm coming at it from a very, very different angle.
• 31:07
  SPEAKER_01: But I always appreciate, especially in with things like like um LLMs, um the surprise, you know, even the people who are working on them, like you know, the the whole black box thing and and and the fact that they like, oh, like these work a lot better than any of us expected them to.
• 31:25
  SPEAKER_01: And and that to me is like you know, a great example of that Arthur C.
• 31:29
  SPEAKER_01: Clark quote that everybody loves about you know, uh technology uh sufficiently advanced technology is close to magic.
• 31:37
  SPEAKER_00: Yeah.
• 31:38
  SPEAKER_00: No, I uh agreed.
• 31:39
  SPEAKER_00: I mean, uh I use LMs every day to do all kinds of things now, and I continually am delighted and surprised by like the degree to which they seem to A, be very intelligent in whatever way you want to define that, but really probably more importantly than that, like very capable on tasks that like in my mind I am pretty confident they have never been trained on, right?
• 32:07
  SPEAKER_00: Like I'm coming up with problems that like are just in like weird parts of the space that like I know people aren't really there's just not a lot of training data on the internet, right?
• 32:15
  SPEAKER_00: Like that's the sort of nature of of university research.
• 32:17
  SPEAKER_00: We're trying to do something new, and their ability to solve problems that I know are unsolved in the literature is really exciting.
• 32:24
  SPEAKER_00: And I guess that that is like what kind of fills me with like renewed optimism that even uh even if we haven't settled on what the solution is for physical AI yet, we're just we're we're in such a better world when it comes to tools for exploring what that could and should be.
• 32:41
  SPEAKER_00: And so stuff that used to take me, you know, a team of graduate students, and I'd put all of them on it, and they would do spend all their time implementing stuff and spend very little time thinking, that is like totally flipped on its head.
• 32:53
  SPEAKER_00: So now like you can have coding agents implement all the stuff you want, and we can spend all the time thinking about well, what's the next thing to try?
• 33:02
  SPEAKER_00: Is this actually working?
• 33:03
  SPEAKER_00: Why isn't this working?
• 33:05
  SPEAKER_00: And even trying, even getting to spend like the mental energy or the time to focus on that, I do think is gonna bear long-term benefits for research, for innovation.
• 33:18
  SPEAKER_00: Just um I don't see how it can't.
• 33:21
  SPEAKER_01: Yeah, so I guess if I have to put a really kind of fine point on it, then it would be um that uh you you you know you're we have to we're we're open to the possibility certainly that there would there will be a similar unexpected breakthrough in physical AI, but we certainly can't count on it happening.
• 33:41
  SPEAKER_00: Yeah.
• 33:42
  SPEAKER_00: I I think that is the exact distillation, right?
• 33:44
  SPEAKER_00: Like if we're expecting that it's gonna follow the exact same path as language models, I think we're gonna be disappointed because there's already enough data and indicators that it's on a different trajectory.
• 33:54
  SPEAKER_00: Now, not to say that it can't follow some nonlinear path that we haven't seen yet, but um the idea that it's gonna be easy uh and it's just gonna be another thing that we do, like what we did with language models, I think is is rapidly becoming not the case.
• 34:12
  SPEAKER_01: Yeah, I got I got the sense, you know, again, when when the first couple times we spoke, because it was, you know, pretty, I think pretty soon after you uh were sort of transitioning from startup world to to getting back into uh fully back into academics, that yeah, it was sort of like, all right, I'm never doing that again.
• 34:28
  SPEAKER_01: Like that was and and and and and as I'm hearing you describe this sort of decision to like get into the app world, to get into the AWS world, it's sort of like I'm getting into that again, but I'm getting into it in a you know, kind of like like I'm easing myself back into it.
• 34:44
  SPEAKER_00: Yeah.
• 34:45
  SPEAKER_00: Well, look, I will say two things that like have really stood out to me.
• 34:48
  SPEAKER_00: So like the first is that like I love robotics research, I love building physical things and I love testing them.
• 34:56
  SPEAKER_00: I think that the lesson in my mind is that that is super fun in the academic research space because the stakes are very low and it is all imagination and excitement and not terribly stressful.
• 35:13
  SPEAKER_00: Um to the degree that any startup can be not stressful, which I think is like uh basically um like none of them, but to the degree that it can be less, scaling a hardware business is like uh is just is is two types of stress at the same time.
• 35:28
  SPEAKER_00: Um and so I guess the way I've been kind of bifurcating my brain lately is that because I think that there's this problem we have yet to still solve, which is like let's figure out what the objective function is for the physical world, uh, for training AI models, like that seems like a super interesting and very exciting question to try to answer with a bunch of smart people and a whiteboard in an office.
• 35:52
  SPEAKER_00: It seems like an incredibly stressful and incredibly challenging question to answer on the fly uh when investors have given you lots of money and they would really like to see a return on their investment in the short to medium term.
• 36:07
  SPEAKER_00: And so I'm really excited about trying to take technology that's here, particularly like things like language models and figure out how they can really greatly advance um all the industries that we know.
• 36:20
  SPEAKER_00: So, not just not just healthcare, like we're working on the startup, but also here at the university.
• 36:24
  SPEAKER_00: So, like here in the College of Connected Computing, we are trying to use AI agents to do all of the stuff that sucks about being at a university as a professor or a student or a staff member.
• 36:34
  SPEAKER_00: And so we're trying to automate all of that.
• 36:36
  SPEAKER_00: We're building a new college from scratch, which means we have all this kind of blue sky to just make everything new.
• 36:41
  SPEAKER_00: And that's been great and very fun and exciting.
• 36:43
  SPEAKER_00: But on the research front, it's like, I am, I am like, we are deep in trying to just think through these really hard problems, which is like, well, if we want to train an AI model in the physical world, what are the ways that you gather data to do that?
• 36:55
  SPEAKER_00: What are the ways that you look at that data and try to understand is the model successful?
• 36:59
  SPEAKER_00: Is it working well?
• 37:00
  SPEAKER_00: And how do you do that without it being incredibly expensive, both from a time perspective and also from a risk perspective, right?
• 37:06
  SPEAKER_00: And so I like that kind of bifurcation where I think uh startups are great for scaling.
• 37:11
  SPEAKER_00: I think startups are harder to answer sort of open-ended uh questions that could take a decade.
• 37:17
  SPEAKER_00: And so to any founder, like my only advice to any founder is if you are trying to solve a problem that you don't know the solution to already, like, or no one knows the solution to, um that's a really that's a that's a naughty one.
• 37:30
  SPEAKER_00: Like if it's a decade off, startup is probably not the right vehicle to pursue it, would be my would be my pitch.
• 37:35
  SPEAKER_01: But you know, so you're saying there isn't the same inherent stress with publish and perish as there is with learning?
• 37:42
  SPEAKER_00: There is, but it's nowhere near look, we're it's it's small potatoes over here on the academic side of the phone.
• 37:48
  SPEAKER_01: Also, tenure, I feel like tenure probably makes a bit of a lot of people.
• 37:50
  SPEAKER_00: Tenure is an incredible, it's an incredible mechanism.
• 37:53
  SPEAKER_00: We have a system that is designed to allow you to like, I mean, the the the biggest advantage that higher ed has over doing RD in any other context is that like you can come in and basically say, I'm gonna work for 30 years on this problem.
• 38:08
  SPEAKER_00: And yeah, like you got to publish along the way and you got to get tenure and all this stuff.
• 38:12
  SPEAKER_00: But like when I'm hiring people, I'm hiring new faculty here, and they tell me I got a 40-year problem I'm working on, I'm like, great, you're never gonna run out of stuff, right?
• 38:21
  SPEAKER_00: You're gonna come here, you're gonna teach students.
• 38:24
  SPEAKER_00: That's a great uh economic mechanism to allow us to kind of stay, keep the lights on, and you can just think deeply about your problem, and there'll be new students every year that'll wanna think through that problem with you.
• 38:33
  SPEAKER_00: And 40 years from now, maybe you crack it, maybe you don't, but there's just few other places in the world where you can spend like at the outset, say, oh, over the next 40 years, I'd like to solve X, right?
• 38:45
  SPEAKER_00: Uh Yeah.
• 38:46
  SPEAKER_01: Well, okay, so here's a hypothetical analog to a startup.
• 38:51
  SPEAKER_01: Let's say that you go from being, you know, a dean at um, you know, the most respected, or certainly one of the most respected robotic schools in in the world, to uh starting up an entirely new school.
• 39:06
  SPEAKER_00: Yeah.
• 39:08
  SPEAKER_01: Isn't that is that is that not similar to the strata of a startup?
• 39:12
  SPEAKER_00: It really is, and it suggests that like despite any of my previous learnings, like I don't I don't learn from from my past and I just want to keep you know trying the same thing over and over again.
• 39:20
  SPEAKER_00: No, look, I love the early stages of a startup.
• 39:24
  SPEAKER_00: I love the early stages of building anything.
• 39:26
  SPEAKER_00: I think that is where, you know, I think there's a lot of specializational labor, right?
• 39:30
  SPEAKER_00: I think where I have the most value to add as a human being while I'm on this planet is that like I think I can help people uh that are trying to take an idea from like nothing and get it going, right?
• 39:40
  SPEAKER_00: And then um you'd have to, you know, uh ask other people.
• 39:43
  SPEAKER_00: But 10 years down the road, I'm not sure how good I am at being helpful or certainly not as helpful.
• 39:48
  SPEAKER_01: But I think when like when when there's asking you to do it, so there must be asking me to do it.
• 39:52
  SPEAKER_00: I must be doing something all right.
• 39:54
  SPEAKER_00: Um no, but exactly.
• 39:55
  SPEAKER_00: And so like in my mind, it is this balance between trying to say, okay, well, what works about what we're doing now or what works already that's out there, but what can we throw out and what can we start over with?
• 40:06
  SPEAKER_00: And I think bringing that mindset to higher ed, and particularly at a place here like Vanderbilt, where they've been so open to that, has just been amazing.
• 40:12
  SPEAKER_00: Like, you know, higher ed is not known for its flexibility nor its uh sort of uh dynamic ability to change rapidly.
• 40:19
  SPEAKER_00: And I've just been continually impressed with the degree to which they're letting me take risks, let me try new things.
• 40:26
  SPEAKER_00: We're just we're saying, hey, like, you know, everything's on the table when it comes to reconsidering what it means to um go to college, to teach in college, um, to get a PhD, to get an undergrad degree, to to learn, right?
• 40:39
  SPEAKER_00: And I think that that's you know, it's been a challenging time for higher ed in the last uh couple of years, I'm sure you're aware.
• 40:45
  SPEAKER_00: But I think if there's ever a time for us to consider like what should we be changing, what should we be keeping constant?
• 40:51
  SPEAKER_00: And then for computing as well, for particularly for computer science education, right?
• 40:55
  SPEAKER_00: Like things since since we last spoke, the entire way that I program has changed completely five times.
• 41:04
  SPEAKER_00: It started out I was using autocomplete and you know, uh tools that helped you kind of you know finish off the line code you're writing.
• 41:12
  SPEAKER_01: Matt, Matt, Matt, are you a vibe coder?
• 41:14
  SPEAKER_01: A hundred percent.
• 41:17
  SPEAKER_00: Not even am I a vibe coder.
• 41:19
  SPEAKER_00: My whole desktop, if I were to show it to you, is a hundred windows of coordinated agents writing software for me.
• 41:26
  SPEAKER_00: I haven't directly written a line of code.
• 41:28
  SPEAKER_01: Are you a vibe are you a are you a vibes dean?
• 41:32
  SPEAKER_00: I'm a vibes dean.
• 41:33
  SPEAKER_00: The way we're running the college.
• 41:35
  SPEAKER_00: I have agents helping me make decisions.
• 41:37
  SPEAKER_00: Look, I I think if you're not using these tools, there is no way you are gonna be competitive uh with people that are.
• 41:47
  SPEAKER_00: I I am 10 times faster.
• 41:50
  SPEAKER_00: Uh my like my ability to make decisions, think clearly, all of that.
• 41:54
  SPEAKER_00: Now, look, you got to apply some human in the loop on this stuff.
• 41:59
  SPEAKER_00: But I mean, you know, while we've been on this podcast, I can my agents in the background have written 700,000 lines of code.
• 42:05
  SPEAKER_00: Like, even if I were to uh do nothing but do that, uh I I couldn't get that kind of output or efficiency.
• 42:13
  SPEAKER_00: Um, no matter how I you know split my day up or you know what level of coffee I drink.
• 42:17
  SPEAKER_01: Aaron Powell Yeah, I've got a complete aside on something that I could easily just type type into your like uh chat GPTs.
• 42:23
  SPEAKER_01: But do you have any idea why the paper is called the hustler?
• 42:28
  SPEAKER_00: No, I don't actually.
• 42:29
  SPEAKER_00: That's a great question.
• 42:31
  SPEAKER_00: Uh uh I know why I know you know why we have so many nautically themed stuff.
• 42:37
  SPEAKER_00: Okay.
• 42:40
  SPEAKER_00: Well, it's just that you know the the nickname uh of Cornelius Vanderbilt was was the Commodore, and so that uh you know that shipping sort of magnate uh energy has has pervaded uh lots here.
• 42:51
  SPEAKER_00: But I actually don't know why it's called the Hustler.
• 42:53
  SPEAKER_00: That's a great question.
• 42:53
  SPEAKER_00: I should look that up.
• 42:54
  SPEAKER_01: Are there a lot a lot of like Lionel Ritchie themed?
• 42:58
  SPEAKER_01: It's totally you know, Music City here.
• 43:00
  SPEAKER_00: It's entirely popular.
• 43:01
  SPEAKER_01: I don't know if Lionel is recording here, but I can look out my window and see if you kind of had the country thing going on with some anyway.
• 43:06
  SPEAKER_01: Ten recording studios.
• 43:08
  SPEAKER_01: Yeah.
• 43:10
  SPEAKER_01: I asked this because you know, I was I I was obviously like sort of like looking into the um uh the origins of the the College of Connected Computing, and and it seems like the um the the ball got rolling before they brought you on, they were doing the search for for Dean.
• 43:25
  SPEAKER_01: But like can't be overstated the fact that this is the first what is it okay, uh the first uh new undergraduate school added since the Blair School of Music in 1981.
• 43:41
  SPEAKER_01: So the first one in like 40 years.
• 43:43
  SPEAKER_01: So like that's a huge deal.
• 43:46
  SPEAKER_00: Agreed.
• 43:46
  SPEAKER_00: Look, I think it's uh I think it speaks to the moment we're at where you know you can look at really any measure, uh news coverage, uh the economy, the stock market, whatever you want, and computing.
• 44:00
  SPEAKER_00: Is like so central, right, to just everything now, right?
• 44:03
  SPEAKER_00: And um honestly, I think it was pretty prescient for Vanderbilt to try to kick this off because this all happened if you think really before we were in the full swings of sort of this AI era we're in now, like they started planning, oh, we need a college that's gonna be teaching computing and hopefully like doing that for all of our students across the university.
• 44:22
  SPEAKER_00: So, you know, when I came in, one of the things that got me to take this job, obviously leaving CMU, which is an amazing place, the robotics institute was just an incredible home for me for such a long time.
• 44:31
  SPEAKER_00: Um, but what got me to take this was that, you know, I thought this was like if there was ever a moment to try to build something new in computing, it's now.
• 44:39
  SPEAKER_00: And um, if there's ever a time to think about like, well, how are jobs, the economy, the future of what everything is gonna look like, like when does that need to be reconsidered?
• 44:52
  SPEAKER_00: And what is what do we need to be thinking about so that we're graduating students out into an economy with skills that allow them to be like gamefully employed, to uh live long, rich, and fulfilling lives.
• 45:03
  SPEAKER_00: And it just feels like this is the moment for us to think about all those questions.
• 45:07
  SPEAKER_00: Like, I I kid you not, like my entire workflow of how I do everything in my life has changed so dramatically in the last year that it highlighted to me that like somebody who's you know 18, 19, getting started in their career, if we could give them the ability to go out there with knowledge of how things are changing, with a resiliency to the kind of speed and pace at which technology is changing the way jobs are done, what jobs even exist, like that seems like such an important mission.
• 45:40
  SPEAKER_00: And so that's what the College of Connected Computing is trying to do.
• 45:43
  SPEAKER_00: It's saying, how do we prepare if you're an English major, if you're a journalism major, if you're a art history major, whatever you're studying, like how do we prepare you for the fact that like computing and particularly AI may have upended that industry in many ways that we can't even appreciate now by the time you graduate in four years?
• 46:06
  SPEAKER_01: Yeah, the more I think about it, the more I realize that I think I think the the startup analogy is a pretty good one.
• 46:10
  SPEAKER_01: Except, you know, I guess the one kind of wrench in this conversation is is is tenure, because uh again, you know, I've like had this conversation with a lot of people of like, hey, why did you leave your job as a uh executive at Google to launch a startup, right?
• 46:25
  SPEAKER_01: And and and like that must have been a you have a kid, and this must have been a really difficult conversation with your partner, right?
• 46:32
  SPEAKER_01: So like obviously, like I'm sure that you know you had a this conversation with your partner.
• 46:36
  SPEAKER_01: We're um, you know, this is a little bit more uncertain and and we're we're moving cities, although again, you know, you were able to move cities to a place where she was able to also find to continue her career, but um still had to have been a very difficult decision to leave Pittsburgh and CMU.
• 46:54
  SPEAKER_00: I you know, I loved Pittsburgh, I loved CMU.
• 46:58
  SPEAKER_00: I you know, it is really hard.
• 46:59
  SPEAKER_00: I don't know.
• 47:00
  SPEAKER_00: I I the thing I keep coming back to is that, and this is the thing I say to to young people when they come into my office, it's like life is is both long and short, but like the short bit of it is that you know you have very few opportunities to really take risks and to try new things.
• 47:15
  SPEAKER_00: And like when those opportunities present themselves, I think you have to be open to them because I think the thing more than anything else that can um undermine you and what you want to do in the world is just like not fully um taking the risks that are um seem interesting or exciting to you, or just not figuring out a way to keep learning and growing, right?
• 47:42
  SPEAKER_00: I think like one of the most um frustrating things about the way our society is set up is as you get older and more senior and you have more knowledge and wisdom in your uh sort of career, like you're actually put in a position where you can take less risks, you get to be less creative, and I think that that's like ultimately really frustrating.
• 48:00
  SPEAKER_00: And so, you know, for me, I thought, well, this is an incredible opportunity to continue to grow as a as an academic, continue to grow as a person.
• 48:13
  SPEAKER_00: And, you know, I just like if you're not growing and learning and changing, I don't know what we're doing, right?
• 48:18
  SPEAKER_00: And so uh, yeah, again, uh uh I got a four-year-old, uh, we had to like uproot her, put her in a new daycare, all of these things.
• 48:28
  SPEAKER_00: It's not easy.
• 48:29
  SPEAKER_00: Um, but I come to work every day so excited.
• 48:33
  SPEAKER_00: I have like a whole host of new problems in starting a new college that I just never had at CMU because that was a college that had been around for 50 years at that point, right?
• 48:41
  SPEAKER_01: And and and just to be fair, and to and to uh to be certain, that that's an upside.
• 48:46
  SPEAKER_00: Great.
• 48:47
  SPEAKER_00: No, it is I come in fired up because it's like every day, what what do I get to solve today that like I'd never thought about before?
• 48:56
  SPEAKER_00: And as a leader, man, that is it gets me in the office and like so excited.
• 49:02
  SPEAKER_00: I I turn up, I check what my agents have done, I see, I see what results they have for me, and then I'm ready to go.
• 49:07
  SPEAKER_00: I'm ready, I'm ready for an exciting day.
• 49:09
  SPEAKER_00: And that that to me is is worth you know all the challenges, right?
• 49:13
  SPEAKER_00: And but it you know, there's a human cost to all this.
• 49:15
  SPEAKER_00: Look, I don't want to undersell and you know, you talk to any founder that's founding anything, right?
• 49:18
  SPEAKER_00: It certainly has this cost, right?
• 49:20
  SPEAKER_00: It has a cost on your on your personal life, it has a cost on your emotional well-being, on your time, free time, yeah.
• 49:26
  SPEAKER_00: Yeah, but I don't know, it's there there is nothing like coming in and just not knowing what the day holds, but being excited for it, right?
• 49:37
  SPEAKER_00: Knowing that you're gonna have six things thrown at you that you've never thought about before, but being like so amped that if you can figure them out and you can solve them, that like and to bring a team along with you, right?
• 49:46
  SPEAKER_00: I think that's the other thing.
• 49:47
  SPEAKER_00: Like to have a bunch of people here.
• 49:48
  SPEAKER_00: So by the you know, this is a college now that has uh we have about 650 students, something like that already, and then you know, about 50 or 60 faculty.
• 49:58
  SPEAKER_00: And, you know, to try to get them to come along on this journey and say, hey, look, we're we're building something here, we're trying to triple in size in the next three years.
• 50:05
  SPEAKER_00: I mean, that's sort of unthinkable in the academic sphere.
• 50:08
  SPEAKER_00: And it really feels like the kind of start a classic startup journey, right?
• 50:11
  SPEAKER_00: You know, it's a lot of risks for everybody, but like if you can convince them and align them all to the same mission, it just it just feels like the best of humanity to have everyone come in and be firing on all cylinders trying to build something together, right?
• 50:23
  SPEAKER_00: And you know, that's a really good point.
• 50:26
  SPEAKER_01: And it's something that I hadn't considered because I was just coming at this from the point of view of like, okay, like it must have been difficult to recruit the the CMU dean to come over here, but like, okay, but now it's the CM, the former CMU dean's job to say, hey, how do we get the potential like CMU, MIT, etc.
• 50:43
  SPEAKER_01: students to come to this entirely new school and to try something new to like how do we how do we again how do we convince you know this new graduate to come to our startup instead of Google or Meta?
• 50:58
  SPEAKER_00: Yeah, look, I mean, I think that's one of the big challenges, right?
• 51:02
  SPEAKER_00: So if you're starting something new, you both have the challenge that nobody knows who you are necessarily, and that you're asking them to take a risk themselves.
• 51:12
  SPEAKER_00: But like I think that that's like for us at least here at Vanderbilt, like that's the perfect uh confluence of factors because it is the kind of students we want to recruit here, are ones that really are saying, hey, I would like to build something new and actually take a risk.
• 51:28
  SPEAKER_00: And I think you know, there are amazing students all across uh the country in a bunch of computing programs.
• 51:34
  SPEAKER_00: And I think there are a ton that um those other schools, including CMU, MIT, all these other places are the right fit.
• 51:41
  SPEAKER_00: Um, but you know, what you get there, I think, is really innovative education, but also like a real sense that they've been doing it this way for the last, you know, N years and that you're becoming part of that culture, right?
• 51:55
  SPEAKER_00: So you leave, I think, uh becoming part of whatever the existing computing culture is at those places.
• 52:01
  SPEAKER_00: And the students that we're trying to reach here, it says, look, you can build something, like you get to build a brand new computing culture.
• 52:06
  SPEAKER_00: And I loved many things.
• 52:08
  SPEAKER_00: I was an undergrad at CMU as well.
• 52:09
  SPEAKER_00: I love many things about that place.
• 52:11
  SPEAKER_00: Um, but if you had dropped me in and told me I could build a culture, I think I would have built one that was different than the one I inherited when I started there.
• 52:19
  SPEAKER_00: And that to me is like what we're saying to students.
• 52:22
  SPEAKER_00: It's like, look, you know, there are blue chip computing facilities you can go, schools you can go to.
• 52:29
  SPEAKER_00: There are startup opportunities are plenty, but um this this will be one of the few opportunities you have to be in on the ground floor with your uh, you know, whatever uh early uh RE uh whatever equity shares, though you don't get to make any money.
• 52:46
  SPEAKER_00: Uh none of us make any money, so that's the critical thing to mention that nobody gets rich out of this.
• 52:49
  SPEAKER_00: But you know, like you get but you get 50% of zero.
• 52:53
  SPEAKER_00: Yeah, exactly.
• 52:54
  SPEAKER_00: But you get to be part of that inaugural class, right?
• 52:57
  SPEAKER_00: And like one of the things that's blown my mind about higher ed is that people still come back.
• 53:02
  SPEAKER_00: Our football team did really well this year, you know, for the first time in a long time.
• 53:06
  SPEAKER_00: And people came back, you know, alumni came back that had graduated like 50 years ago, like 50 years ago.
• 53:13
  SPEAKER_00: And so in my mind, we're building something where I hope that we have students that come back 50 years from now.
• 53:18
  SPEAKER_00: And so I was part of the first class of the College of Connected Computing.
• 53:21
  SPEAKER_02: I was employee number five.
• 53:23
  SPEAKER_00: Exactly.
• 53:23
  SPEAKER_00: And they get to tell that story, and then when their grandkids graduate from a very well-established, well-heeled computing program, you know, they can uh try to convince them that look, I had a real hand in being part of this coming together.
• 53:36
  SPEAKER_00: And and I'm sure like uh all of us older people, um, we will be ignored.
• 53:42
  SPEAKER_00: But but I do think that that's an incredible opportunity, you know.
• 53:45
  SPEAKER_00: That'll be an incredible opportunity for young people.
• 53:46
  SPEAKER_01: It's funny, like and and I, you know, as as as you may or may not be aware, you know, I'm I'm before I really started covering robotics, I was uh uh I was a hardware editor at TechCrunch for a long time.
• 53:57
  SPEAKER_01: So, you know, it's cover covering consumer electronics.
• 54:00
  SPEAKER_01: And um, you know uh again you know something that became like very clear is is when it comes to being sort of innovative in the hardware space, especially, uh, it's a lot harder to do at a large company because it's a lot harder to take risks.
• 54:16
  SPEAKER_01: And it sounds like there is a real equivalent to be drawn in the uh academic world.
• 54:23
  SPEAKER_00: Oh, 100%.
• 54:24
  SPEAKER_00: It's it's it's it's the same thing.
• 54:26
  SPEAKER_00: And I get it.
• 54:26
  SPEAKER_00: Look, if you're at a big company that's making money and successful, or you're at a big university that already has a program that's CMU's program is ranked number one, right?
• 54:34
  SPEAKER_00: Yeah.
• 54:34
  SPEAKER_00: So there's nowhere to go but down, right?
• 54:36
  SPEAKER_00: So the the question is to what degree is it even smart to allow somebody like me to come in and change a bunch of things, right?
• 54:43
  SPEAKER_00: Like where are we gonna where are we gonna take the program?
• 54:46
  SPEAKER_00: And so I think like that more than anything else, like if you ask me why, like people, you know, always ask this question, like, you know, how do startups end up innovating and like why are they allowed to be innovative?
• 54:55
  SPEAKER_00: And we come up with all these like but the I I think more than anything else, like if you have the ability to fail and you have the ability to fail, um that is just a gift because it allows you to use it.
• 55:08
  SPEAKER_00: And nowhere to go but up.
• 55:10
  SPEAKER_00: And nowhere to go but up.
• 55:11
  SPEAKER_00: Exactly that.
• 55:11
  SPEAKER_00: Nowhere to go but up.
• 55:13
  SPEAKER_00: We are you know creating something new.
• 55:15
  SPEAKER_00: There's nowhere to go but greater reputation, greater success.
• 55:18
  SPEAKER_00: Um, and so I don't know, that's just it's um it is uh it's an incredible luxury or gift to be able to try to build something new.
• 55:29
  SPEAKER_00: And I would just encourage everybody that has the opportunity to to take any of those risks in any in any aspect of their life, right?
• 55:36
  SPEAKER_00: Like I don't think it has to be necessarily at their job, but to take risks and try to build something, right?
• 55:41
  SPEAKER_00: Like, I don't know.
• 55:42
  SPEAKER_00: It feels like what like much of what the modern era of sort of malaise and you know AI slop and all these things that we get fed, like um there's obviously a lot of bad stuff happening, but like the opportunity to try to make a better world is just one of the unique pleasures of being human.
• 56:02
  SPEAKER_00: And I think anybody that can avail themselves of that that has the opportunity to do so, um, yeah, I think they should.
• 56:08
  SPEAKER_01: Well, uh a rare episode of the podcast ending on a positive note.
• 56:13
  SPEAKER_00: Thank you so much for joining us.
• 56:15
  SPEAKER_00: I really appreciate you having me.
• 56:17
  SPEAKER_00: Thanks.
• 56:18
  SPEAKER_01: Thank you to Dean Matthew Johnson Riverson and the folks at Vanderbilt University for helping to set that up.
• 56:24
  SPEAKER_01: If you've been enjoying the show, please like and subscribe.
• 56:27
  SPEAKER_01: We are just a hair under 1,000 subs on the YouTube channel.
• 56:32
  SPEAKER_01: Please uh subscribe, tell a friend, and please check out the newsletter over at automated.fm.
• 56:38
  SPEAKER_01: If you're watching this before June 24th, 2026, please come out to our third ever automated live on stage with Ali Aka of Field AI at Automate in Chicago.
• 56:51
  SPEAKER_01: It's going to be a great conversation about physical AI and more.
• 56:54
  SPEAKER_01: The guy who's building uh merchant helicopters.
• 56:57
  SPEAKER_01: It's gonna be super interesting.
• 56:58
  SPEAKER_01: Okay, that's about it for this week.
• 57:00
  SPEAKER_01: Stick around.
• 57:00
  SPEAKER_01: We will be back just about this time and next week with another episode of Automated.