Hosts Eamon McErlean and Joe Devon interview Ed Summers, Head of Accessibility at GitHub. They discuss how Ed began working in accessibility, some of the latest developments at GitHub, how sonification is used in accessibility, and more.
OUTLINE:
00:00 Opening Teaser
00:31 Introduction
01:18 GitHub Copilot & Developers Announcement
02:16 Getting Involved With Accessibility
04:47 Priorities & Challenges in Accessibility
08:30 Developers + Accessibility
14:54 GitHub Copilot
17:35 Benefits of Using Gen AI
22:36 Haptics vs Other Tech
25:39 Using Sonification To Help Read Charts
29:26 Customer Feedback
33:12 How Conversational AI Will Affect User Interface Development
35:53 Coding Accessibility Video Series
36:50 Future of AI & Accessibility
40:42 What's On Your Wall?
41:48 Wrap Up
--
EPISODE LINKS:
GitHub
https://github.com
Amelia Wattenberger
https://wattenberger.com/thoughts/fish-eye
GitHub - Coding Disability
https://github.com/readme/featured/disability-creativity
GitHub - YouTube Channel
https://www.youtube.com/github
GitHub - Accessibility Playlist
https://www.youtube.com/playlist?list=PL0lo9MOBetEGvxp03Kh3eIlFvIvfE21KF
Physical Intelligence - Robot Folding Laundry
https://www.physicalintelligence.company
SAS - Accessibility
https://www.youtube.com/playlist?list=PLVBcK_IpFVi9kCxPXz4dd1HO5x_yLLEHJ
SAS - Examining the Details of a Chart Using Sonification
https://www.youtube.com/watch?v=AVEflA4dmIk
--
PODCAST INFO:
Podcast Website:
https://accessibility-and-gen-ai.simplecast.com
Apple Podcasts:
https://apple.co/46eflnv
Spotify:
https://open.spotify.com/show/4eEwo3jUSo3aS7wGhlcxs2
RSS:
https://feeds.simplecast.com/nCrQiw1t
LinkTree:
https://linktr.ee/a11ygenai
- You know, another great quote that I read sometime back was from Seth Godin, and he said that "Culture is the way we do things around here." And I think, you know, the best successes that we have in accessibility is when accessibility becomes the way we do things around here. You know, and that's one of the things we're trying to do at GitHub with a large degree of success.
- Welcome to episode five of "Accessibility and Gen AI," a podcast that interviews the newsmakers and artificial intelligence. I'm Joe Devon, joined by my co-host, Damon Erlene, I'm Joe Devon, joined by my co-host, Eamon McErlean, ed Summers, who's a true advocate Ed Summers, who's a true advocate and leader in accessibility. GitHub hired him to spearhead their commitment to accessibility. Ed, thank you for joining the pod.
- Wow, thank you, Joe. Thanks, Eamon. I'm a loyal listener of the pod and just incredibly humbled and excited to be on this side of the mic.
- Awesome. Why don't we start with something exciting, because a little birdie told me that you have an announcement that you are going to make right here on the podcast. Would you like to do so now?
- That's great. Thanks for the opportunity. GitHub Copilot, our AI coding assistant that's integrated into the GitHub platform is now free for all GitHub users, and we're also announcing that we just crossed the boundary of 150 million developers on GitHub.
- 150 million?
- 150 million developers.
- Wow.
- And has that number been like a progression, you know, over multiple years, or have you seen spikes in that recently? Or how's that progressed to that figure?
- It's that elite progressing up. You know, I've been at GitHub now for two and a half years. Not long after I joined, we announced the 100 million developer mark. So, I dunno, you have to compute that. You have to do the math to compute that growth rate.
- Well, either way, congrats. It's a hell of a figure to hit. And speaking about you joining GitHub, you said two and a half years ago, would you mind walking us through your journey, like from the beginning and how you really got involved in the accessibility world?
- Well, when I was a kid, I was diagnosed with a degenerative retinal disease. It was very minor for many years. And, you know, along the way as I went to college and I earned degrees in computer science and math, so I'm kind of a math nerd. And this degenerative retinal disease really kicked into gear, you know, in my twenties. And by about the time I was 30, around early thirties, I was crossing that boundary from, you know, then low vision to really thinking of myself as being blind. So as you can imagine, that's quite the journey in a variety of ways, you know, just psychologically and socially and professionally. But I was really blessed all the way through. And it became pretty clear that with a background as a developer, and by that time starting to lead manage teams and just that technical experience combined with the lived experience of being low vision and then blind, it became crystal clear that my path going forward was going to be accessibility. And I started, at the time I was managing a development team, was the engineering manager. And I started doing kind of some, not related to accessibility, and I started doing some kind of just volunteer work at the company I was at to work on accessibility, you know, nights and weekends. And pretty soon I was recruited to lead that team to build that program at the company that I was at. And that was in 2010. And I've been building teams and leading the accessibility programs ever since.
- So it's been that progression over basically 14, 15 years in total. And where you're at now, GitHub, and the scope that GitHub has from a customer perspective, from a global developer perspective, I'm sure as head of accessibility, that you cover many aspects both from a customer engagement and direct employees. But you know, and I've experienced this myself, you wear multiple hats. How do you prioritize kind of the areas you go after and what would your greatest challenges be from your perspective?
- You know, it's true. Many of the people who are quote unquote heads of accessibility do wear many hats. But I think there's a lot of people out there that are effectively heads of accessibility without the title and without the, you know, formal support within the organization. And in many respects, I don't know if you you would agree with this, but I think, you know, having a formal title of head of accessibility, you know, reporting at a high level within the org, that's a blessing because, you know, that just implies, you know, resources are available in order to incorporate accessibility as just a fundamental fundamental operating principle, you know, across the org. So, let's see. I think, you know, I kind of think about getting support across the org, there's a couple things. There's three things really. There's carrots and sticks are the first two, right? Kind of the BF Skinner model, you know, and the carrots are a job well done and more customers and things like that. More revenue because of accessibility. And of course, you know, sticks are, you know, legal compliance and things like that. But then there's this thing that I learned pretty early on, which I call identity. And, you know, identity is when people kind of include accessibility in disability inclusion, you know, including people with disability in the way they think about the business and they think about their role in the business. You know, another great quote that I read sometime back was from Seth Godin, and he said that "Culture is the way we do things around here." You know, the way we do things around here, and I think, you know, the best successes that we have and accessibility is when accessibility becomes the way we do things around here. And that's one of the things we're trying to do at GitHub with a large degree of success.
- No, love it. As I've mentioned before, I think, you know, you talk about the carrot and the stick, I think the three areas I've always called out is A, first and foremost, and should always be first and foremost, it's the right thing to do. Two, conformance, and it is becoming more and more critical from a conformance, especially in the public sector. But three, and I'm sure you'll agree, if you do accessibility right, it can become a competitive advantage ultimately for you if you do it the right way for the right reasons. But to do it the right way, you have to embed it. It has to be, it cannot be an afterthought. It cannot be thought of as like an extra thing you have to do. It just has to be part of your DNA, it has to be part embedded in the product development life cycle for you to truly succeed.
- That's right. And to continue succeeding, right? I think that's where the key thing there, embedding it.
- I love how you brought up identity Ed, because to me... Here we are, those of us advocates of accessibility that have been trying hard for many years to get the community, the community of developers and the people that build digital technology to care about accessibility. And I've always felt that that accessibility is downstream of dev culture, if that makes sense. And what's really cool about your position and what made me so excited about GitHub creating a head of accessibility, was that the dev culture lives on GitHub. So being that you're in such a key role to make a difference, and you know, we've collaborated a fair bit together, so I know what kind of a difference you've made, but can you share how you view that dev culture and its importance, how the craft of development, including accessibility, how important is that and how much do you think about that as you consider what your goals are for your mandate there?
- Yeah, no, I think you're 100% right in that, you know, the dev culture is just a... the global dev culture, most of whom are working on GitHub, you know, I've a clue in the dev culture of designers and product managers, right? Kinda like you. Broader. We collectively are making or breaking, you know, accessibility. And if you think about, you know, the way I think about this is there's 8 billion people in the world. There's 150 million people working on GitHub. So let's say that's all of 'em. And so there's, you know, whatever, 2% of the world's population, we are creating the technology that's used by everyone. And that technology is not optional in the sense that, you know, if you want to network online, you know, if you wanna network period, right? I mean, it's global now. You have to do it on LinkedIn or, you know, social channels, you know, blogs and podcasts. And if you can't access those things, then you're shut out. If you want to apply for a job, then that happens online. You know, when it comes to the online digital part of it you know, during the pandemic employment for people with disabilities rose during the pandemic because it became possible to work from home and that just opened so many doors. So I think you're right. you know, the quote unquote developers, developer culture, it makes or breaks accessibility. And I think that there's a positive trend. You know, there's just a growing awareness and incorporation of accessibility into dev culture. And I think that if you look at the, you know, the great designers and the great developers, solving the problem, you know, making the thing work is kind of like, it's really easy. There's a lot of developers that can write the code in such a way for it to work. But when you start doing great work within the constraints of limited memory, you know, or, you know, performance constraints or accessibility constraints, that becomes the kind of fertile ground for, for real innovation. And, you know, that's where developers and designers who are really good at their craft, that's where they shine, you know, embracing those constraints and making great experiences in spite of them.
- How do you get them bought in to that, Ed? Like, you know, we talked about the carrot and stick before, but like, how do you incentivize and really showcase the criticality of it?
- It's a great question. Well, I think there's data, right? I mean, you kind of look at the way that people, that change happens within the human brain and also within organizations. Data is a big driver of that, you know, and that data is the incidence of people with disabilities, or the number of users that have disabilities, or the revenue, you know, there's all kinds of great data points they've created. But then I also think that those stories about people with disabilities, about users of your products and how they succeed or how they struggle, you know, that really does make a difference. And then I think, you know, one of the things that's been super effective for us at GitHub is just a few key decisions about accessibility that take advantage, that kind of graft accessibility onto existing mechanisms within the company. And I'll give you the best example for GitHub, we have this program called the Engineering Fundamentals Program. And that is a way that we as an organization just promote engineering excellence in everything we do. So our three engineering fundamentals are availability, accessibility, and security. And we have scorecards that measure that. Now, the fact that this Engineering Fundamentals Program exists, well, it's just an amazing thing because all of the websites and all the products that we ship they're kind of broken into services, and a service you can think of as an entire product, or it may be a portion of the product, you know, like a certain number of pages on the site. And the purpose of the services is just to be able to assign ownership to a team. So, you know, certain teams might support the login process, whereas others are supporting some other part of the website. So we have this thing called service catalog, and that way we have kind of an owner and we can we attach these scorecards to services, whether it be security availability or accessibility. And it's just a wonderful framework within which to make, you know, just to make visible and make transparent where we currently stand when it comes to those fundamentals. And also, you know, drive that excellence over time.
- And have definitive ownership for the respective areas as well.
- Exactly, exactly.
- You and I were seated next to each other when your CEO, Tom Dohmke, and I hope I got his name correct, announced that the new mission of GitHub is now all around AI and making coding more accessible to everyone. There is a goal to hit 1 billion users, which is huge. And your GitHub Copilot is clearly the mechanism to get there. It's just incredible that it is now free. And I would love to understand what you think about that vision, how GitHub Copilot relates to it, as well as how to do accessibility with GitHub Copilot.
- Yeah. Well, the 1 billion developer goal is a bold goal. And I remember when I first read about it, it was one of those times where I locked my computer and stood up and went for a walk round the block just to process the enormity of it. You know, considering there's 8 billion people in the world, so making, you know, 1 billion of them developers, that's a big challenge. But as we've worked on this and started to, you know, map out plans towards bringing this into reality, what we're really saying is that there's 1 billion people who have access to the tools of innovation. And the way that's gonna happen is via natural language. So when I learned to code, I learned to code using Fortran 77, many, many years ago. And, you know, the languages and tools that we use as professional developers are pretty nerdy, and there's a lot of people their brain doesn't work like that. But when people can express an idea in natural language and have AI turn that idea into a prototype, then they can start to innovate and to validate their ideas and assumptions and then bring in, you know, quote unquote professional developers to take that and run with it and turn it to a hardened product. So I think, you know, this goal, what we're really saying is we're going to unleash you know, just the next level of human creativity.
- Completely agree. And I think it puts a whole new spin in the whole, you know, low code, no code, because literally you can have that, you know, true generic information of what you want, just as long as you can articulate what you wanna build, you can end up, you know, being a semi developer, if you will. Apart from that specific area, Copilot, either personally or professionally, what else in the Gen AI world upcoming that excites you?
- Well, let's see. Let's talk about how, kinda like my go-to AI, you know, as a person who's blind, if we didn't say that, meant to say earlier that I am blind and my go-to use case for AI in the physical world is currently using my Meta AI glasses, which are paired with my iPhone. And I use that just straight up using the Meta AI feature. And I also use it with this app called Be My Eyes, Be My Eyes is an app that originally allowed you to connect a blind person to connect with a volunteer where they could share their camera and their audio with a volunteer and get, you know, sighted assistance. For reference your first version of your podcast. Your first episode was with Mike Buckley, the CEO of Be My Eyes. And he is just an amazing leader in the space, doing great work there. But then in the last year or so, Be My Eyes added this feature called Be My AI. So it's just my workflow right now when I need to answer a problem that, you know, solve a problem or solve an access problem in the physical world is like first go to AI and I ask either Meta AI straight up or Be My AI, I ask, you know, I'll start interacting with it to solve my problem. And if I can't do it that way, then I'll call a volunteer and then I got a human on the line. And there's a surprising number of problems that I can solve very nicely, you know, hands-free access using the camera on my Meta AI, my Ray-Ban glasses. So, I mean, that's just opened up just a lot of more, you know, kind of efficiency and independence for me in ways big and small. I think, you know, one of the things I would love to see moving forward with this is we need GPS to be a part of that. Like, I want that model to know where I am. That's one thing. And then we also need to create this kind of a goal directed mode. So for example, let's say that I get out of the Uber at the airport and I need to go to Gate D 16, I would love to be able to bring up my AI agent through my glasses, and say, okay, I need to get from where I am right now, you know, GPS camera feed being detected. "I need to get the gate D 16, let's go," you know. And I have a big German Shepherd guide dog. We move fast and we can go, but we need help reading the signs as we go by them, be able to scan the space and figure out, you know, oh, there's the escalator, there's the elevator, you know, those kinds of things. And I think that's one place where this is gonna go moving forward. And you're starting to see just a ittle bit of that. There's a couple projects that went out recently. One is the live AI feature from Meta that was demoed at Connect a few months ago. Pretty cool. If you haven't seen that, check it out. And then Google has a project named Project Astra, which is kind of the same thing. it's a continuous conversation with AI using both an audio and video feed. Pretty cool.
- I did see somebody show the Gemini 2.0, and I think it was the Astra, where it was looking while they were doing social media, and he did say to the AI, "Please encourage me to be more productive." So it said, "Hey, get off social media and get to work!" Which was pretty cool. You know, Ed, in our conversation. So I've personally been very excited by David Eagleman did this amazing TED Talk where he showed that he gave a haptic vest to someone who was deaf, let him train for like an hour a day. And then at the end of it, essentially he was taking the sound from the iPhone, turning it into haptic touches, and this person was able to hear through the haptic touches. And I got really excited by haptics in general, but what I love about hanging out with you is everybody has a different perspective. And I remember you didn't feel that haptics is working as well. So I always love to hear different people with different perspectives. What do you think from, you know, the upcoming, you've got some people wearing pins, some people wearing glasses, some people wearing the earphone, the earbuds. There are different form factors. What's your take on haptics versus the other potential form factors in terms of what's gonna really change things for people generally and for the accessibility movement?
- Yeah, yeah, that's a fascinating space. I think there's a ton of potential with haptics. It may work for some people, not work for others, right? I think the other thing about haptics is not only with haptics, but you know, I have a lot of background in this field called sonification. I worked on the accessibility of data visualization for many years while I was working at my previous company. And, you know, sonification is using sound to represent data. You know, once you get past kind of surface level interactions with haptics or sound, then there becomes learning curves, right? And these learning curves is where you kind of have to, you literally have to learn how to interpret new things, you know, and wire in kind of, you know, build those pathways in your brain. So I think there's a lot of potential there. I think the challenge is, you know, climbing those learning curves. Because once you get past the immediately obvious kinds of things, it becomes non-intuitive and things don't necessarily make sense right out of the box. So, for example, you have to interpret the signal from the haptic device in order to be able to understand how that stock chart is moving, you know, up and down. Just as an example of that. But I think, you know, one thing this points out is, well, a couple of things. One, the advances we're gonna see in using AI is kind of in lockstep with these other fields such as HCI, right, 'cause AI's not a product. AI is a foundational technology that can be used in products, you know, to solve problems. You still have to have a user interface. And user interfaces still have, you know, what do you wanna call it, kind of restrictions. For example, you know, right now we get a lot of feedback from AI using text or speech. And so if I want to, I don't know, locate where I dropped my keys or my keys are on the floor somewhere or on the counter, like, "Hey, where are they?" Well, you know, you kinda have to describe that right, in text. But, you know, the beautiful thing about vision, which I don't have, is the ability just to see exactly where it is. So I think we have to find additional mechanisms like haptic, there's additional innovations that are gonna block some of the AI, some of the progress that we make on AI. So, you know, the folks who are out there who are working in the HCI space, you know, yeah, I think they're doing the good work and opening up additional adoption of AI.
- Yeah, absolutely. And I think the word you were looking for was constraint.
- Yes. Thank you.
- And would you mind just explaining in a little more detail the sonification, what that looks like? It's such a cool technology and maybe explain how you would do that with a stock chart, and is that something that you did at SAS?
- Right. Yeah. So before I came to GitHub, I was the director of accessibility at this company called SAS, all caps, S-A-S. It's a leading firm in data and analytics. And when I was there, pretty early on, when I started working on accessibility, it became obvious that if we were to make our products fully accessible and quote unquote conformant with something like the web content accessibility guidelines. The way that people used our products was, they were using it to analyze data, to find insights in the data. And again, you know, the WCAG was not enough. You know, it required a way to experience, to understand the data faster than a description. And way back then, there was no way to generate a description programmatically, like we have now with Gen AI, right? So we leveraged some existing research, there's been research in the field of sonification you know, since of the nineties. And the way it works is basically if you think of a stock chart, the Y axis represents the price. And we would map that to the pitch of a tone, like a musical note, from a low C to a high C, let's say, on the piano scale. And then we would map the left hand side of the chart and the left speaker all the way over to the right hand side of the chart and the right speaker. You know, taking advantage of the simple panning effect, right? That's been available in audio APIs for a long time. And you know, those two dimensions of panning and pitch, those are just the bread and butter of sonification. And that allows you to kind of wire in or to convey a lot of information in that XY plane. And then if you add interactivity to that, right? So rather than just kind of like having the whole thing play, if you make it so that you can literally use a keyboard to explore each data point, for example, then it becomes a lot more active and just, it greatly increases the ability to perceive and understand data visualizations.
- And that's why accessibility is so cool, it's so innovative and people are just missing out because it's the constraint that makes things fun and innovative. So thank you for sharing that.
- And you mentioned, you touched upon WACG. You know, conformance levels. But to do it properly, we do need to go above and beyond. I always say we wanna be conformant, but we always want optimal usability. One of the features we're doing at ServiceNow is complete conversational AI voice integration. So we can have ultimately an end user enter analysis for conversational AI and then enter their actual platform to process a customer case, to resolve an issue, completely via conversational AI. So, and again, as we all know, the best accessibility features can be utilized by everyone and are utilized by many people, which is really where we wanna get to. When we look at that, when we look at new features like that, that customer engagement and actually listening to our customers is a key part of that. Can you share with us from a GitHub perspective, what that engagement looks like for you and your accessibility team members with your customers? Do you have ongoing meetings with them? Do you have different forums with them? How do you solicit that feedback?
- You bet. Yeah. So over the last couple of years these are capabilities we've been building, just foundational capabilities that are required in any accessibility program in order to do it well and to solve the usability problem as you mentioned. So there's a couple different mechanisms. So our mainline support channels, there's primarily two. We have a support channel for enterprise customers as most companies do, where questions get reported through support@github.com. And then we have just open on the web, we call them community pages. And each product area on our community site has a, you know, a discussion, it's called a GitHub Discussion, where people can post questions and get answers. So we do have a category on the community site specifically for accessibility, where we can post announcements and users with disabilities can report problems or ask questions. So, actually over the last couple of quarters, we've just built out our ability to triage and track and get answers back to people on those forums. On the accessibility forum specifically. And I'm just proud to say we've got our response time down to two days. That doesn't mean we're fixing issues in two days. It means we're giving you a response and we're starting that dialogue in two days. And that's a big achievement for us. And then a couple other ways where we're doing this is we have a user research program that focuses specifically on inclusion. So every month we're running usability studies, and we're pulling from that, we're inviting people to participate in those studies through what we call our Inclusive Research Panel, which I think we're up to about 150 people with disabilities now. On the communities page there's a discussion at the top where you can go register, you know, you can request to become a part of that panel. And then we've just recently created a, this is a pilot project where we've created a Slack channel for developers with disabilities where we have a a lot more kind of interactive discussion, and that's under our pallet right now with just a few people admitted at this point as we start to build that out. But we hope to grow that in the future. And we can get a link to the communities page and share that.
- That's awesome. To got feedback and even from your employees, that timely feedback of certain phases of the development life cycle is invaluable. It really is. That engagement's key.
- Yeah. Eamon, let me ask you, you said something there that really peaked my curiosity. So you were talking about conversational AI at ServiceNow, and I just kind of, you know, would love to hear both of your opinion on how that conversational AI kind of front end, how that's gonna affect user interface development, you know, as a whole moving forward. Are we seeing the death of UI because of conversational AI?
- Joe, go ahead. I'll let you go first.
- You know, I just had in my mind this one conversation and you gave me the perfect lead in to it. You have the most amazing employee. I just love watching all of her work. Her name is Amelia Wattenberger, and she builds these incredible UIs. And she has one now called Fisheye, which is a different way of looking at data. And it seems to be integrated into LLMs, if I get it correctly. She teased some product that's coming out that I haven't been able to see. But no, I think there's lots of room for different kinds of UIs and it's really important to see what's gonna be coming. And definitely follow Amelia, she's incredible. Anybody watching this, just Google her or ChatGPT her and find out where she does socials.
- Yeah, from our perspective, we do a combination of a few things. We do like heuristic evaluations as well to see how intuitive. For example, voice assist, how easy it is to turn on, how easy it is for an end user to actually access it and know it's there. And thereafter, again, we go back to customer engagements walking through step by step. The biggest thing for us from a UI and even from an overall functionality perspective is simplicity, it really is. How can you simplify the entire process? We try to do that. We have a new data fabric initiative going on to make sure that data from different sources can be condensed and be as fluid as possible from an end user perspective without having to go to multiple different streams or areas. But the simpler the better. And it's the same thing. I don't care if you're talking about a conversational UI or any other UI, that's kind of our mindset overall. And you touched upon a key piece of that earlier, is ensuring you engage with individuals with disability, ensuring you engage with them early in that phase, at the design phase, even at the requirements phase, to solicit their feedback, or else you're not gonna deliver an optimal solution unless you engage.
- And let me add one other feature that Amelia created for GitHub Next, I think it's called, it's sort of like a GitHub lab feature. She created this accessibility brush where you have code and then you just wave your wand, put a brush over it, and it makes the code more accessible. Real, really cool feature. And we've got only a couple of minutes left, so we're heading toward the final comments here, Ed. And if you could share where you think it's all going, where AI and accessibility is going, and maybe a little nod to the incredible blog that you have with some great experiences of people who have been impacted by the work that you and the good folks at GitHub are doing.
- Yeah, that goes back to the stories. We created a video series, it's called "Coding Accessibility." And thus far we've highlighted five developers with disabilities who are working on GitHub, you know, some in open source and just doing amazing work. You know, some of the folks that are highlighted there include the folks who created the free MBDA screen reader. Which was created by blind people for blind people. Each one of those videos, they're about 10 minutes and they're definitely worth a watch. You can check 'em out on our YouTube channel. We can get this in the show notes as well, maybe, at youtube.com/github. Let's see, where's it going? Are we talking about where's it going from an AI perspective?
- Whatever you would like to most share. But my question is more around AI and accessibility.
- Agreed. Okay, cool, cool. Yeah, you know, first off, it's going so fast. You know, the innovation cycle with AI right now, I think is the fastest innovation cycle I've ever seen. I've been around this business for a little while, you know, and I can remember when, you know, innovation cycles were years long, but we're just looking at like weeks now with different innovations. And I think... I'm probably not the first person to say this, but I think we're entering that phase where the foundational technology of Gen AI is increasing in performance and getting a reduction in cost to the point where now we're gonna go into this period where, you know, the hammer that is Gen AI, people are gonna use it to start pounding on a lot of nails to see how well Gen AI can crack open different problems. So I think, I mean, clearly that's happened in genomics, in genetics, it's happened in the biological sciences, it's happened in the material sciences. And I think, you know, as a result of this just kind of flourishing of using this new tool to solve different problems, the impact on humanity is gonna be huge and huge in a positive way. And this may be a little bit controversial, may be a little bit spicy, because, you know, disability isn't something that needs to be solved, you know, necessarily. On the other hand, one cause of disability is disease. And I think there's a lot of innovation that's gonna happen that's gonna reduce the incidence of disability due to better preventions, treatments, cures of disease. And I wanna just qualify that by saying there are people in the disability community who do not feel that they need to be quote unquote fixed. I don't feel like that, you know, certainly there are people in the neurodiverse community that I think would agree. People in the deaf community tend to agree as well. It's part of who we are. Blindness has become part of who I am. But I think, you know, just the innovations in medicine, it's gonna change humanity, I think in one way. And then I think in addition to that, the thing that I'm really keeping an eye on is robotics and the way that AI is being incorporated in ways that can change the physical world. One of the things I saw recently that I think is really fascinating is there's a startup called Physical Intelligence. Yeah, check 'em out. They're basically using Gen AI to train a general purpose model for the physical world, for controlling robots. And they're solving problems, like the simple one is just the ability to fold laundry.
- [Joe] Oh yeah, I've seen their videos. Yeah.
- Seen that?
- Yeah.
- That model was trained using Gen AI. And, you know, Gen AI, what it's really good at is language prediction, right? Look at the data, find the patterns in the data, predict the next token in the data. And the data streams that they're using are these sensors from these manual robotic arms that people are controlling with a remote control. And that along with the video feed is being used to train models to do these very intricate things. And I think, you know, that's gonna flourish. And in the next five to 10 years, you know, the control that we have as people with disabilities with the physical world is gonna change drastically.
- Thank you so much, Ed. This has really been a pleasure catching up with you on the pod after all these years.
- We finally did it.
- Ed, you don't mind if we wrap up on a more of a personal note. So we see two things behind you, Einstein, picture of Einstein to the right. And then I think that there's three vinyl records or albums on your left shoulder. Are you a music guy through and through, or can you give us a little bit of background on them?
- I do love music. I originally went to college to be a music major, you know, pretty common in our field. Turns out that I'm a hell of a lot better at computer science than I am at music. And it's a good thing too. I think I would've been a starving musician, who knows. But yeah, that's Stevie Ray Vaughan, my favorite guitarist of all time. And of course, I don't know if you know, but Einstein had a disability, he was dyslexic.
- Dyslexic, yeah.
- Yeah, for sure. You know, and I think it was, could have been the dyslexia that gave him the ability to see, you know, to think differently, basically, I think. And then of course there's my GitHub Monocat there in the background right there.
- Listen, honestly, just reiterate what Joe said. Thank you so much for your time. It was an absolute delight talking to you and learning from you. Thank you for your continued efforts and looking forward to a continued relationship with you. I really do.
- Yeah. Thank you both. Thanks for doing the pod. I'm a long listener, so I can't wait to see the next episode coming out in my feed.