エクステンデッドリアリティで働く：AR・VR と AI が変えるインターフェース

In 1968, Ivan Sutherland built the first working head-mounted display. It was called the Sword of Damocles. You can see it today at the Computer History Museum in Mountain View, California. Virtual reality, thus, is not a new technology, or at least not a new idea. It's 58 years old, and for most of that time it was an expensive thing. My guest today has research-grade headsets in his lab that cost more than a brand new car. And then the Oculus Rift development kit arrived at about $300 instead. Roughly the same field of view, same resolution, and in some respects, better tracking, maybe. So if the hardware problem is largely solved, what's next for VR? That turns out to be an interesting question, and it's what we talk about today when we get into interface design. My guest Doug Bowman is a professor of computer science at Virginia Tech and a member of its Center for Human-Computer Interaction.

He's run the 3D Interaction Group there since 1999. He co-wrote the book on this, 3D User Interfaces. In this episode, we get into why obvious design instincts, like reaching out and grabbing something with your hand, isn't necessarily the right choice for this environment. It seems like VR is a whole new design paradigm. and I was grateful to get to talk to Doug about his expertise in this area. My name is Doug Bowman. I'm a professor in computer science at Virginia Tech. I'm also a member of the Center for Human-Computer Interaction, which is an interdisciplinary group of faculty and students here. Can you share a few details on what the Center for Human-Computer Interaction is all about? The Center for HCI is this interdisciplinary group across more than 20 departments here at Virginia Tech. And it involves faculty and students who are interested in all aspects of user experience design, understanding technology's effects on people.

And in particular, there's a lot of emphasis on extended reality, augmented reality, virtual reality. We have a lot of expertise in that area. But then we also involve many faculty from various application domains. So many of our HCI faculty work with people in the Department of History, for example, which is really interesting. We have a large number of people from construction who are interested in the use of interactive technologies. So it's pretty diverse. Well, I could guess that construction is interested in the way it could assist in that safety and efficiency and sort of things on a construction site. How do historians play a role in the picture? Because we're at a technical university, even our humanities departments are often thinking about how to use digital technologies and things like that. And so it just turns out that we have several historians who want to think about telling historical accounts using technologies like augmented reality. For example, I worked on a project where we wanted to give an augmented reality tour at the Petersburg battlefield from the Civil War.

So understanding how the battle took place, what life was like for people on the surrounding plantations at that time, you can use augmented reality to kind of transport people and explain things in a way that you can't with traditional technologies. So that's been a lot of fun. Well, there are some off-the-shelf and commercial technologies for both VR and augmented reality. To what degree do you think about working on those versus exploring research questions that involve hardware? Our group really does no hardware research whatsoever. So we rely almost exclusively on off-the-shelf technologies. So we're using the same MetaQuest or Valve Index or HTC Vive that you might be using at home to play games or what have you. We're, of course, writing our own software to do application prototypes or new forms of interfaces or interaction. Sometimes we do experiment with different forms of input devices or sensors that are not natively part of the heads up.

So sometimes we're modifying the hardware, but typically we're using the off-the-shelf hardware to create these prototypes and then run experiments with human users to understand how their user experience changes when we make some modification to the interface or to the experience or how they understand being in an immersive space versus using a desktop application, for example. I think there's always been virtual reality since I've been alive. But, of course, if we rewind to the 80s, it's primitive by today's standards. It would have to be we're not in the dark ages. But where do you see us as on, like, the development curve? Do we have really amazing state-of-the-art things that just need applications? or will we continue to see more and more impressive hardware devices opening new opportunities in the future? I've been in this field for more than 30 years at this point. So I did my graduate school work in the mid to late 90s at Georgia Tech, and I was doing virtual reality at the time, and the hardware was definitely primitive when I started.

And even during my grad school years, there were some nice advancements. It's interesting that, to go back a little bit, virtual reality as a real technology has been around for more than 50 years at this point. So the first working system was 1968, Ivan Sutherland's sort of Damocles system, which you can see in the Computer History Museum, actually. I saw it this summer. So it's been around for a long time. But really, we didn't start to see kind of rapid development of the hardware capabilities until the Oculus Rift really was kind of the turning point or the inflection point. That device took what I had in my lab at the time, which was a $30,000 research grade head-mounted display, and it gave the same capabilities in a $300 developer's kit. more or less the same field of view, same resolution, and in some ways better tracking than was on the research grade. So that was obviously a big cost reduction, but once there was a consumer market for these devices, even if it was small, that really drove lots of hardware

optimization and lots of investment is what really made the optimization go much faster. In terms of where we are today, I mean, there's some very nice specs that you can get on VR headsets in particular. And if you want to be fully immersed and you don't want to walk around with these things on and use them, you know, in your everyday life, but use them for special purposes in your home or at work or whatever, I'd say we're almost where we want to be on VR devices. But where the really interesting hardware development is still got a long way to go is in what I call everyday AR glasses. So think of, you know, the capabilities of an Apple Vision Pro in the form factor of the eyeglasses that you're wearing or of today's kind of AI smart glasses. So there's still a lot of room to improve in terms of optics, in terms of heat dissipation, in terms of battery life, in terms of compute that you can fit.

on a device like that. So still a lot of hardware development to go, I would say. What kind of research questions can you ask with the hardware you have today? That's a really interesting question. So we tend to ask questions about future user experiences. So I mentioned everyday augmented reality. I want to know what sort of user interface should we use or what sort of user experience should we design in the future when we do have those eyeglass form factor devices that we can wear all day, every day. And so we tend to try to simulate those experiences on hardware that we can get off the shelf today, even though they're future experiences. So we can't simulate the full experience because these headsets are going to be bulkier and heavier and so on. But we have figured out ways to get those headsets, those existing glasses, into some more everyday sorts of situations and have people use them for shorter amounts of time. and we can still ask and answer some pretty interesting questions

about what those future user experiences should be like. I don't have any kids myself, but I have some nieces and nephews, and I've been surprised at how adept they are with tablets. They seem to just understand gestures like evolution gave it to them, not that it was learned, even though I'm sure it was. Do you find people either get or will achieve the same sort of natural experience with augmented and virtual devices? Is there a learning and training curve? So part of the job of an HCI researcher is to think about how to avoid there being a learning curve in many cases. There are some cases where you're designing a system for experts, and it's okay for there to be a training period because they're doing something really complicated that may not be natural, and they're going to be doing it for work all day, every day, and so a little bit of training is not necessarily a bad thing there. But if you're designing for kind of everyday user experiences for consumers, you really want them to be able to get up and running immediately. And so a lot of people think that, you know, the best way to do that is to try to make interaction in extended reality the same as interaction in the real world to the extent that that's possible.

So when people think about well how should I manipulate this virtual object the naive design is okay well let just reach out with our real hand and grab that virtual object and move it the same way that we would move a real object. And that works to some extent, but it's really limited. Our arms only have a certain amount of reach. Well, we can't take advantage of all the dexterity of our hands with the current hand tracking systems that are available with these devices. So it's not actually always the best idea to try to design a so-called natural interface that completely emulates the real world. Instead, we try to come up with more magical interfaces, interfaces that extend your capabilities beyond what you can do in the real world, but that still use some of the same natural affordances that you might use for real-world interaction. You know, for example, in the real world, you might refer to something by pointing at it with your finger from a distance.

Well, in virtual reality, that same sort of gesture can be used to select distant objects, for example. So you're giving the user a new capability to interact with things at a distance, but you're doing it in a way that makes sense and feels natural. Well, when I think of the web, maybe specifically like HTML and its form options, I've got input fields, check boxes, radio buttons, and a collection of other kinds of things, toggles, what have you. There's sort of a language of two-dimensional form design. Do we, I guess, as a culture or a society or maybe just the research community, is there such a language of 3D interactions, like the way users will interface in the same comfortable way? That's a really interesting question, and I don't think we have time in this whole podcast to fully explore it. But I'll say a couple things. One is that that language that we already have from 2D user interfaces doesn't have to completely go away. And so you will see if you go to the home menu of a meta quest or something similar,

you will see some of those familiar user interface elements like buttons and checkboxes and sliders and so on. The way you interact with them might be slightly different because the input devices you're using are different, But the same visual language might apply to some of those. The same metaphor for how they control things is going to apply to some of those. So we can take advantage of people's familiarity with 2D user interfaces to get them up and running in 3D user interfaces as well. But then, of course, at the same time, we want to invent better ways to do things in 3D. And the kind of naive idea often is let's just take advantage of this third dimension. So let's lay things out in depth as well as width and height. Let's use three-dimensional objects instead of two-dimensional windows and so on. And sometimes that can be appropriate, but often it makes things more complicated than they need to be. So we often find that if you're doing some of these standard sorts of computer interaction tasks like selection, like a multiple selection like you would do with checkboxes, like setting a parameter value like you would do with a slider, let's just use those existing methods that are proven.

But then when it comes to interacting directly in the 3D environment, what we find is that there are, again, kind of some standard actions that you have to do. So there is a concept of 3D selection. How do we select objects in the scene that are scattered in three dimension? There's 3D manipulation. How do we move those objects around and position them and orient them? There's 3D navigation. How do we move ourselves through the scene? And then there's system control, which is kind of everything else. all the commands and buttons that you would normally think of. So for at least those first three, there's been a lot of thought about how to design unique interfaces, unique interaction techniques for VR and AR systems, and to do those in a way where people can understand the metaphor, even though they're not used to doing those things in standard computer interfaces. So I was a member of a team that wrote a whole book on this topic, and we'd go through in painstaking detail the categories of interaction techniques you can use for things like selection, manipulation, and navigation.

I think you're referring to 3D user interfaces theory and practice. I notice now in its second edition, and first edition lasted for quite a few years, what brought on the need for a second edition? Well, as you say, the first edition had been around for many years and surprisingly held up as well as it did given how fast the field moves, the pace of innovation in the field. So we really just wanted to, we didn't change the structure of the book to a great extent in the second edition. Those kind of universal interaction tasks and devices are still relevant, but we have newer examples and references to newer literature and things of that nature. And in that time, how have you seen the field evolve? Are there any maybe highlights that stand out for you as major leaps forward? Well, in the current moment, the thing that is obviously changing a lot is how AI is going to affect interaction in these 3D environments. So for the early part of my career, the first 20 years of my career at least,

we were primarily thinking about all manual interactions. So given an environment and a set of objects and some tasks that the user wants to perform, what are all the manual actions that they need to do in order to accomplish that goal? And we were able to, you know, develop some interactions and interfaces that allowed people to be quite efficient and make few errors and things like that. But it was all manual interaction. And now people are starting to think, what can I do with AI in this domain? And a lot of it is assisted interaction or, in some cases, automatic interaction. If a user is pointing to select an object, maybe they don't have to point as precisely as they would otherwise because the system can infer which object they're interested in based on other signals from their gaze or from their head pose or something like that. I saw an interesting paper just last week at a conference where they were predicting whether or not the user intended to click on something in 3D.

And so they would look with their gaze and fixate on an object, and then based on a large data set that they collected, they're able to predict with a high level of accuracy whether the user intended to click on it or not, and so the user doesn't have to click. The click can happen automatically and with very low latency. So there's all sorts of examples of this, and my lab is very interested in this AI plus XR field. So going back to your concept of all day, every day for a device, I have quite a console set up for my home working area. you can't see it all because you're seeing from my laptop camera, but vertical monitor here, big one over here, you know, everything the way I like it. But when I get up and walk away or I go to the coffee shop, it doesn't exactly come with me. Only this, you know, one small piece in front of me can come with me. I know there's some opportunities out there I could be exploring. I haven't. What kind of success do people get, like, when they look into augmented reality as extending their device screens? Well, the first thing I want to say about that, which is not directly answering your question,

is that some people have a conception of virtual reality and augmented reality as only being for gaming or only being for truly immersive experiences where you're transported to another world or you're adding three-dimensional objects or background to the real world. But we do find, as you're implying, that one of the most promising applications and use cases for these technologies is just the everyday work that we do. And a lot of us are knowledge workers. We're using computer screens all day, and we are really limited by the size, location, shape, configuration of these physical screens that we use. And so, yes, if you're in your home office where you have the perfect setup, then those physical screens are ideal. But as you said, you can't easily pick those up and take them with you to the coffee shop or on the airplane or what have you. So a great kind of feature of augmented reality and virtual reality is that you can have more or less unlimited display space.

So we can create virtual displays that exist anywhere in space. And those displays can be any size. They can be any shape. They can be dynamic. You can change their position at will. You can add more display space at will. And so the whole world potentially becomes a display. And so now instead of having these bulky physical displays that are fixed in size, fixed in location, we instead can have whatever setup we want wherever we want. And so we've done several studies along these lines to try to understand whether that sort of work in different physical locations actually makes sense and what do people feel about it. How do you go about studying something like that? Well, it goes back first to a concept that I mentioned earlier, which is simulating future experiences with current devices. So we certainly are doing some things that look a little weird from the outside. So, for example, in one study, we took the Microsoft HoloLens 2, which is now discontinued, but it's an optical see-through augmented reality display,

and attached that to a pretty high laptop and took that out into places on our campus We took it to a cafe We took it to the library We took it to a meeting space a classroom and we asked people to perform knowledge work tasks that they would be doing as part of their work or study. So we made sure that they had all of the documents and apps that they need available through their Google account so they could log into their Google account on our laptop and then they'd have access to their real work. And so that allowed our study to be more what we call ecologically valid, that we're studying people doing tasks that they care about instead of just predefined kind of lab-based tasks that we give them. That gives us a much better idea of how people will respond to these technologies when they're really doing their own work with their own devices in the future. And what did you hear back from your participants in terms of how eager they were to have their own HoloLens or maybe not eager? I don't know. I wouldn't say that they were necessarily eager

to have the device that we used. I mean, the HoloLens 2 is an amazing piece of technology, but it's not what consumers would want to wear all day, every day. Still too big, still too strange looking, still way too low of battery life, too small of a field of view, etc. But it does illustrate the potential of using these sorts of virtual displays to do work in different locations. And people were very excited about the future possibilities of this sort of setup. They did see that they could do their work with it. It is feasible to do work with it, the same as with your laptop or what have you. But it's much richer, potentially more expressive, maybe more productive to have these kind of extended virtual displays beyond just your small physical laptop display. And so I think people were quite excited about the potential for the future. In comparison to the HoloLens, which you have real data on, since then the

Apple Vision Pro has come out, and just by right of it being newer, I would guess it's somehow better. Do you have any thoughts on how incremental versus revolutionary a device like that is? I think comparing those two devices, comparing the HoloLens 2 to the Apple Vision Pro, There's a lot of differences. So the HoloLens is an optical see-through device, meaning that you're looking directly through lenses at the real world. The Vision Pro is a pass-through or video see-through device where there are cameras looking at the real world, and you're seeing video from those cameras piped into your display. So that's one major difference. The way that you interact is quite different in the two as well. The Vision Pro uses mostly gaze-based interaction. The HoloLens uses primarily gesture-based interaction and so on. So it's hard to compare them directly. I would say that in this domain of doing knowledge work or just using virtual displays, there's probably not a huge step change between two devices like that.

Different people will prefer one or the other device for different reasons. I think where we'll see the big leap is when we have the eyeglass form factor glasses that I was talking about. earlier, where it really is possible to wear and use all day and to use in many different situations. So in our study, we did test a variety of situations, but only one at a time and one per participant. I think another interesting question that we haven't really been able to study yet is what happens when you're doing a piece of work in one location? Then you need to get up and walk to class and you continue that work after class in another location. or maybe you need to finish up that email you were writing while you're walking to class. How does that user experience work? So it's these transitions between contexts that are the next kind of interesting question I'd love to study. Is there much scholarship in this area, or do you find it's very much an open field to figure out how context and AR fit together?

It's an exploding area of scholarship for sure. As I mentioned, I just got back from a conference. It's the ISMAR Conference International Symposium on Mixed and Augmented Reality, which is one of the conferences I like to go to every year. And this topic of everyday AR, of adaptive AR, of AR in different contexts in the real world is very much at the forefront of scholars' minds. And so there's a lot of people trying to study this, a lot of people trying to design user experiences that fit this new way of working in XR. And so it's pretty exciting, but there's still plenty of open questions to answer. And where do you see the field going? Well, as I said before, I think this combination of XR and AI is the most exciting thing right now. So thinking about how that impacts user experience and interface design, one thing I like to say is that the interface that I need for working in my office in XR

is probably very different than the interface that I need for working on an airplane in XR and much, much different than the interface I need for doing some work while I'm walking from one building to another, which is much different, again, from the interface that I need for viewing entertainment in my home, right? And so if we rely only on manual interaction with these systems and we're wearing them all day through all these different contexts of use, We're always going to be fiddling with the settings, and nobody wants to do that, right? So you don't want to have to stop what you're doing, open the settings, you know, change all of these different parameters, or even just go into a mode in order to fit the context that you want. You really want the interface to adapt to your context automatically. And so that, I think, is the next frontier. And I would say that in HCI, people have tried adaptive interfaces for a long time, and many of them have been quite unsuccessful.

So think about Microsoft Clippy from the Office suite of 15-ish years ago. There was a lot of jokes and memes and sketches about that and how poor the experience was because it might be able to do a decent job sometimes of predicting what you were trying to do and offering you help, but most of the time its recommendations were just not helpful and, in fact, just annoying. So how do we design an adaptive experience that really can recognize the context of use, really can understand what I'm trying to do, really can give me suggestions or help that is going to give me just the information that I need, just when I need it, just in the way that I need it? That's a tall order. If we don't get it right, then these adaptive interfaces are going to fail. And so we don't want to fail like that again. Well, our culture is currently exploring questions about privacy and just sort of what's appropriate. Should I wear these sorts of devices into my coffee shop or down the street?

And I don't know that we're going to, you know, use a theorem to figure out what's the right answer. But do you have any thoughts on it being so close to the topic? Absolutely. I knew you were going to ask this question. Yeah, it's been in the news a lot lately. And if we don't solve this problem of privacy, both privacy for the user wearing the device and privacy for bystanders who might be captured using the device, then this, you know, as far as a market, it's going to be dead in the water before it ever gets off the ground to mix the metaphors. So, yeah, we really have to address this problem. And I think it's got to be a multi-pronged sort of approach. So are there some technological things that we can do to improve privacy? Absolutely. So, for example, we had a piece of work in my lab recently where, again, we were kind of simulating these future everyday AR glasses, and we were looking at when they might capture your own personal information just as you were wearing them around your house while you were doing some other task.

And we compared kind of a simulated privacy assistant that would try to identify those objects from the video stream and then obscure them or blur them before that video stream ever got sent off the device into the cloud. And we compared an assistant that did that automatically versus one that gave you recommendations. And then the user had to kind of confirm, yes, I do want to obscure this information. And in terms of preference, it was actually kind of a 50-50 split. So some people prefer kind of just the convenience of the device obscuring things for you as long as it got it right most of the time. And other people really wanted to have kind of the final say. They wanted to monitor what the privacy assistant was doing before they agreed to that. Anyway, that's an example of a type of technological solution that might help your sensitive data not go off your device into hands that you don't want it to go into. But technology alone really can't solve this problem. We also need policy solutions, which I'm not an expert in, and we need social conventions.

So if you look at the history of technologies over many years, we've had to evolve kind of social conventions for their use and how we perceive their use as well. An example that I like to give is with smartwatches. You know, when I first got my Apple Watch and I would get a notification during a meeting and I would lift my watch to see what the notification was, the other person invariably thought that I was checking the time and they thought that I was bored with the meeting. But now everybody understands, oh, it's just a notification. They're going to check it. It's only going to take a second and then we'll be back to the conversation. So that an example of kind of evolving social convention that to some degree can help alleviate some of these problems Well my first computer was very heavy and very bulky and I never moved it off my desk So I was glad as that form factor evolved into the laptop I have today, and I would say even the extension of my phone. So I really have ubiquitous computing, but I don't have the same desire for my stove that can stay in the kitchen. Maybe I only need VR in a room or one

location. I know that doesn't necessarily fit your all-day-everyday kind of vision, but could that be the world we evolve into? It's possible, but I don't think so. And the reason is, if you look at kind of the general-purpose computing platforms that we've used over the years, there's like a very clear trend. So, you know, before PCs, we started with, you know, mainframes and other kind of centralized computing resources. And then PCs came along, and what do they do? Well, they made the device smaller, slightly more portable, more convenient to use, more personal. And then, you know, after PCs, we have laptops. They increase all of those same characteristics. It's smaller. It's more portable. It's more convenient. It's even more personal than a PC. And then smartphones, right, same thing. And so we keep moving in that direction. But at the same time, with all of those devices, our capability for kind of rich and expressive display has gotten less, right?

So with PCs, you can attach a large number of large monitors, like in your home office that's up there. But with your smartphone, you're limited to this tiny few-inch by few-inch screen. You're getting bigger, but not that big, right? So the amount of information you can fit and the amount of interaction you can do is very limited. So when we think about AR glasses, are they going to be smaller? Yes. Are they going to be more convenient? Yes. Are they going to be even more personal? Yes. You wear them on your face. But they're going to buck the trend of having a smaller display. So like I mentioned before, with an AR glasses setup, you can have more or less infinite display. And so I think now with this sort of device, you're getting kind of the best of both worlds. and that's why I think this is really the next paradigm of general purpose computing. So if all those hardware problems that I mentioned before do get solved, and I'm pretty sure that they will, then I don't see a future in which most people don't have a set of all-day wearable glasses.

And do you have any thoughts on good opportunities for students interested in HCI? Maybe it's a CS undergrad thinking about their master's degree. what are some of the best avenues they might follow? Come and work for me, first of all. There are several entry points into HCI because it's an interdisciplinary field by nature. So a lot of people do get into it through computer science, for sure. And for computer science students, most major CS departments now have one or probably multiple HCI courses as part of the undergraduate curriculum. And then after that, you know, doing some undergraduate research or independent study sort of project can really help you understand what it's like to kind of push the boundary of the state of the art in this field and prepare yourself for grad school doing more special purpose work in HCI. But it doesn't all come through computer science. Psychology is a huge part of HCI. So perceptual and cognitive psychology in particular are critical parts of the HCI world. And there's many psychologists these days working on the perception and cognition and neuroscience related questions about XR systems and interactive HCI type systems in general.

And then there's the more creative side as well. So there are many people in the HCI world who were trained in graphic design or other sorts of visual arts, industrial design, if it's more on the hardware side of things. And those are critical sorts of areas to go towards a career in like user experience design and that sort of thing. When you send people out as part of an experience like the students who wore the HoloLens and tasks. My background, having read a lot of machine learning papers, I'm thinking like, oh, you're going to measure accuracy and recall and AOC and time to task completion and these very precise measurement kinds of approaches. But of course, there's also just the interview process, asking people what experience. Can you talk about how you view both this qualitative and quantitative ways of collecting data on your work? It's definitely a yes and. We do both and we really have to do both. So in a study like the one we were talking about earlier, where we sent people

out to do their work in a HoloLens, those low-level kind of quantitative objective measures like task completion time and so on are not really relevant. I mean, first of all, everybody's doing a different task, and those tasks are not really comparable. But even more than that, we're not really interested in our research questions for such a study. We're not really interested in, you know, small increases in productivity or in, you know, level of accuracy with virtual displays versus physical displays. That's not the question we're trying to answer. In a study like that, we're trying to answer a much more high-level experience type question, like what's it like to use virtual displays in a public place? Questions of that nature. And so the measures are necessarily much more qualitative. We observe people and we write notes about what they do. We do log some of their actions in the system, so we can have basic quantitative measures like how many windows did they have open at one time or how did they lay

out their windows, things like that. But then really it's the interview that gives you the richest data. How did it feel with people watching you using this device? Did you believe that this expanded display was actually helpful compared to using your regular laptop screen? Those are things you can't measure with those kind of traditional low-level measures. But then we also do lots of laboratory-based experiments where the research questions are about smaller differences in performance or attitude or experience. And to answer those questions, we have to use quantitative measures. We have to do statistical analysis. So it's a wide range of things, and it really just depends on what's the research question. What sort of claim are you trying to make from the study? So there's an argument to be made that we'll see the first major deployments in industry because they can afford it. If it's purpose-driven, helps the company, they're going to push it out. It can be a controlled environment. Do you find that to be a compelling argument, and do you think things will go that way?

It's hard for me to predict the future on that because I'm just a lowly academic. I'm not a business person. But there certainly are several companies who have made bets in that direction. So the HoloLens part of Microsoft was almost exclusively focused on enterprise applications for a time before the HoloLens went away. They wanted to sell them to big corporations and to the military and so on. Magic Leap, which also recently decided to stop making their own hardware and just provide optics to other companies. They also made this bet on kind of enterprise applications. and I guess so far that hasn't really worked out. And it's unclear if that's because the premise is wrong or simply because the hardware wasn't ready for prime time yet. I think eventually, as I've been saying through this whole interview, I think eventually kind of the consumer application is the biggest market that will and will certainly happen.

It's unclear to me whether there will be like this transition time where it's used more in industry or not. Well, Doug, what's next for you? Well, I'm running this 3D interaction group at Virginia Tech. It's been my lab since 1999. And what's next for me is to keep on mentoring more and more students. I work mostly with PhD students, and they're amazing. It's always very gratifying to see their growth over the five or so years that they're part of this program and then how they go off and make an impact on their own. So I'm going to keep doing that because that's the thing that really gets me up in the morning. But, you know, related to what we've been talking about in terms of research topics, we really are going all in on this so-called intelligent XR approach, where we really want to crack this case of how do we provide adaptive, intelligent interfaces in XR that minimize the amount of input that people have to do, that automatically change based on the context or the user task or the need. We want to figure out that user experience and then marry it with these modern AI technologies and some AI technologies that probably haven't been invented yet to really enable huge gains for people who are using these future devices.

Well, definitely an exciting area. I look forward to keeping up with your work. Where can listeners follow you online? So I'm primarily on LinkedIn. So I've got my personal LinkedIn profile, which you can find Doug Bowman. And then the Center for Human-Computer Interaction at Virginia Tech has its own organizational profile as well. And there are a lot of really interesting talks related to this area that the Center for HCI puts on. There's a weekly online seminar. And so people can find information about that on their LinkedIn page. Very cool. We'll have some links in our show notes for people to follow up. Well, thank you so much for taking the time to come on and share your work. Thank you, Kyle. This has been great. Thank you.