Hosts Eamon McErlean and Joe Devon interview Ted Drake, Global Accessibility and Inclusive Design Leader of Intuit.
OUTLINE:
00:00 Opening Teaser
00:36 Introduction
01:35 Working with Victor Tsaran on Accessibility
05:38 Getting Involved with Accessibility
10:47 Soliciting and Maintaining Leadership's Support For Accessibility
13:39 DeafBlind Potter
15:17 European Accessibility Act
18:44 Scaling Accessibility at Intuit
23:00 Incorporating AI at Intuit?
29:39 Personal AI
33:57 Open Source vs Closed Source
36:36 Will AI Replace Testing?
41:15 CSUN Assistive Technology Conference
44:55 Detecting Autism Through Web Browsing
47:12 Raising The Floor
48:46 Advice For Newcomers of Accessibility
52:48 Wrap Up
--
EPISODE LINKS:
Intuit
https://www.intuit.com/accessibility
DeafBlind Potter
https://www.deafblindpotter.com
See Me Cane
https://www.seemecane.com
European Accessibility Act
https://commission.europa.eu/strategy-and-policy/policies/justice-and-fundamental-rights/disability/union-equality-strategy-rights-persons-disabilities-2021-2030/european-accessibility-act_en
Personal AI
https://www.personal.ai
CSUN Assistive Technology Conference
https://w2.csun.edu/cod/conference
Axe-con Digital Accessibility Conference
https://www.deque.com/axe-con
Inclusive Design 24
https://inclusivedesign24.org/2024
Raising The Floor
https://raisingthefloor.org
Ted Drake's Website
https://www.last-child.com
Ted Drake on LinkedIn
https://www.linkedin.com/in/draket
- I think AI with agents should be able to see an image, understand the context of a page, and does the alt text relate to the context of the page? It should be able to do keyboard testing, and it should be able to say, these are the things on the page that look like they should be keyboard accessible. But I think there's still going to be the need for a manual review just to make sure that it's usable, that things, even though they hit all the check marks, that you actually have an accessible experience.
- Welcome to episode seven of Accessibility and Gen AI, a podcast that interviews the newsmakers and thought leaders in the world of accessibility and artificial intelligence. My name is Joe Devon, and I am joined by my co-host, Eamon McErlean. And today we are interviewing Ted Drake, a leader, a true OG who is an encyclopedia of technology. He's the Global Accessibility and Inclusive Design Leader at Intuit, and he is a good friend. Ted, welcome to the pod.
- Ah, thank you so much. I've been looking forward to this one.
- Now, Ted, I know, from having known you for so many years, you've worked with some incredible leaders in the accessibility space. Can you share a story or a memory of how they've had an impact on you in the past?
- First thing that comes to my mind is working with Victor Tsaran. Victor Tsaran, who's now at Google, he was my manager at Yahoo. And together with a few others, we co-founded the Yahoo Accessibility Lab. The thing about Victor though was the lab was set up so that developers and designers could come into the lab and actually explore how their products were working with assistive technology. And Victor had this way of knowing enough about web that he could talk to them in the same language and the same dialogue and the same ideas. So a couple instances that I saw regularly, he would be working with a team. It was the first time they ever met him, probably the first time they ever worked with someone blind. And Victor would reach out his finger, and he would touch the screen and he would say, you have something right here that's not working for me. And that was 'cause he understood layout, he understood where the dom was, and where an advertisement or something would sit on the screen. But that would always amaze people that he would be able to find something on the screen that was problematic. And then another thing was Yahoo User Interface Library, which was one of the original JavaScript libraries. They wanted to create graphs that were accessible. Now, this was a long time ago. So at that time, graphs, bar charts, circle graphs, things like that, they were SVG files so they could expand and they could grow and stuff, but they were kind of flat and they could work with a mouse, but that was about it. So sitting with Victor, the team figured out how they can make 'em keyboard accessible. So you could actually sit there and tab and use your arrow keys to move up and down the graphs. And as you put your mouse on the graphs, they would have these little tool tips that will pop up and say, you know, your stock on this date was $100. So they came up with this concept of having a hidden element on the page, and that hidden element on the page would represent whatever they're focused on. So you're not using the mouse, the mouse is not tipping up and things like that. But they would have this little hidden object. So this little hidden object had a new ARIA role called alert. And what that meant was that whenever anything in that container changed, it would be announced by a screen reader. And this was the first time that they had actually, as far as I know, the first time anybody had ever used this to create a dynamic announcement of whatever was changing on the page. And so with that, just to be able to sit in the room and work with developers and designers, and focus on it for a few hours and then come back the next week, and tweak a little bit more, they were able to create that concept of how do you take something that's truly visual and make it interactive, and make it so that people with a screen reader can now navigate a stock chart? This concept has been used for maps and charts, and all sorts of things since then. It's become part of our programming methodology is to use that in content.
- Yeah, and in fact, Victor did a video where he showed a screen reader, and that was the first time I've ever seen assistive technology. And Yahoo in fact were very supportive of Global Accessibility Awareness Day. And I was so excited because they flew Victor out as well as, I'm trying to remember his name. But there was another developer who's amazing, and he gave an amazing speech. The entire audience was completely blown away by it. So yeah, Victor's a good egg.
- I wouldn't be surprised if that other person was Todd Klutz.
- Todd, that's it. Thank you.
- Todd was the best keyboard person around. He could make things work with a keyboard that nobody else could. I learned a lot from both the two of them.
- Ted, thanks for joining us today. Really, really do appreciate your time. We know you're a busy man. You've been involved in the accessibility arena for quite some time now. But for the listeners that are not aware of your journey, would you mind just giving us an overview of, you know, really how you got involved in accessibility, and how that transitioned into where you are now?
- Sure. This, it started for me around 2000, year 2000. I had a degree in fine art, and so I was working for the Museum of Art in San Diego and they needed a website manager. And I had been building websites on the side. So I raised my hand and said, I can do this. I know how to do it. Keep in mind, I'd never coded before. I was using WYSIWYG editors where you would sit there and drag things around, but I understood the concept of the websites, the architecture, I just didn't know how to code. But when I became the website manager for that site, that's when I found out about Section 504, Section 508, ADA, all of these, WCAG 1.0 came out around that time. So that's where we started realizing, hey, I've gotta make this accessible. And what does that mean? At that time, all I knew was that images had to have alt text. That was about as far as I knew at that point, but it was a matter of building bad stuff and then fixing bad stuff. And then we got into this, at the time, you have to think back then, everybody was trying to get your eyeballs. And so they were creating these really super gaudy, animated, brightly colored, horrible websites. And it came to a point where people are just like, stop, we've gotta stop this. We've just gotta go back to basics and build websites the way they were meant to be. And that was called standards-based web development. And I got in on that early, and so I was building stuff on my own, I was participating in chat rooms, I was going to conferences and meetups, and trying to learn as much as I could about that. I left the museum, then worked for a couple companies helping them on their transition. And then one day I got a call from Yahoo and applied for a position at Yahoo, and joined Yahoo Tech. And the thing about Yahoo is everybody there was self-trained. Nobody went and got a computer science degree. And if they did, it was something they did, but it wasn't front end engineering. Because we were all self-taught, we had this broad spectrum of, you know, we'd be a philosopher working with, I was a fine art working with the musician, and we were all in these rooms together, and we had all kind of learned from each other. And accessibility was just the way things were done. But also at Yahoo is where I met the first people with disabilities, and those that were using the products, and realizing while I had been coding to a standard, and checking the boxes to make sure my stuff was successful, I didn't really know what that meant until I actually saw someone using it. And that's why at Intuit, one of the things that we do so much is customer interviews and customer research. And I encourage everyone to volunteer. And that not only gets you to understand why we're doing or how to do something, but the why. And we always try, I always try to push the why, 'cause the why is when you get someone interested in accessibility, and they're doing it for the rest of their lives. As opposed to I've gotta fix this bug, and the next week I forget about accessibility and build the next thing.
- 100%. And we try to mimic that approach at ServiceNow, both from a customer engagement perspective, but just as importantly from an employee perspective. And we've been fortunate enough to build up a panel of like 300 employees, individuals with different disabilities, and we engage with them as often as, you know, as frequently as possible at the right time to solicit that feedback, because it is invaluable, it truly is. And it opens up people's awareness and their eyes, and understanding more so than anything else. It really does.
- You know, one of the things about, you mentioned 300 people, so I'm assuming that it might be like your employee resource group or something like that. One of the things we found is that those employee resource groups and our accessibility champions are great for finding people to help us test things that are not the standards. So testing reduced motion for people that have motion sickness, testing readability for someone that might be dyslexic, or colorblindness, or anxiety, things like that. That's what really also helps is you put a note on the, we will put a note on the Slack channel saying, we're working on a project, and we'd like to get some feedback from someone that has this. If you're interested, you know, direct message us so we can ask you those questions. That's made a tremendous impact for us to be able to have a more well-rounded design process.
- Just tying into that one second, like that engagement's invaluable. But at the end of the day, you still need that leadership support. And one of the most frequent questions I get asked is how do you solicit and maintain that leadership support, you know, of that continued growth and that continued engagement?
- I've been pretty lucky, and I shouldn't say lucky at Intuit, that's why I'm at Intuit. That's why I've been at Intuit for 14 years. Intuit is a customer-focused company. So for me, I don't have to do return on investment. I don't have to say that if we fix this, we're gonna get X amount of sales. That's not part of my job. What I do is I show here's the impact that's having on a customer. Here's how we can improve it for the customer, create videos, create user feedback, things like that. We, if we find that we're building something that is not usable and it's impacting the customer, that's what I have to show. And that's how we prioritize. And our CEO, Sasan, when I first started, he was with TurboTax, and he understood the importance of accessibility at TurboTax. He actually founded, started the accessibility efforts at Intuit, and that was before I joined. So I always knew that I had an ally with him, and so it's been like a combination of working directly with developers and designers and project managers. It's also getting the buy-in from the CEO, and the previous CEO, Brad, was also, and then in between. Before Sasan was the CEO of Intuit, he was the Head of QuickBooks. And I went to Sasan, I said, "There's this project I want to do, "there's this company called Aira, "and they provide visual assistance for people "that are blind or low vision, "and I wanna be able to include them with QuickBooks "as like QuickBooks customers." And he was like, "No, that's not really big enough." He was the one that said, "We need to go bigger." And so I went back to Aira, and what we did was we created the promotion so that it was available to anybody that was blind or low vision and a small business owner. And they could use Aira for up to 30 minutes. I think now it's 20 minutes to run their business. Doesn't even have to involve our products. But he was the one that really pushed for us to think about how we can really power prosperity. That's kind of our motto. So having that as your support from the very top really does help. But it's all the way through our company. It's about making sure that the customers can succeed.
- It's interesting, I think that any organization that is not actually speaking to users with disabilities, they just don't even understand what they're missing, and how it sort of changes how your product turns out, because just looking at it from different perspectives, it changes everything. Tell me, I saw you mention in another interview about The DeafBlind Potter. What's the story behind that?
- Yeah, his name is Kelvin, and he's a really interesting story. This goes back to our Aira promotion. Kelvin is deafblind. He still has some sight, some hearing. He was at a stage where he needed, for mental health, he needed to get back into something that was creative. So he went back to his pottery wheel, and he started throwing pots and cups. And then someone suggested he do a TikTok video. So he and some friends, they arranged the cameras and they created some Tiktoks, and he started this TikTok channel, showing what it's like to be a deafblind potter. But he also mentioned regularly that he was using the Aira promotion, and that his wife was a QuickBooks user. So that's how we found out about him from Aira. So together, we have been working to help support him. We've done customer research with him. We purchased a bunch of mugs, 'cause at Intuit, if we're gonna buy something, we try to buy it from small businesses. And he joined our Adopt a Small Business Program, to launch the See Me Cane, which is a white cane, but it's lit up, because he's been hit by a car three times, and something like 30% of people that are blind have been hit by a car, because you don't see the white cane very well. And this one you can see from 100 yards away. So we helped him launch his website, and his fundraising as part of our Adopt a Small Business.
- We have the European Accessibility, or EAA coming up. How do you think that that is going to impact organizations? Are they ready for it?
- I don't think American companies are ready for it. I'm not sure if European companies are ready for it, to be honest. It has been around for a while. We knew it was coming. Some of the details are, we're still waiting for some of the details, like how each country is gonna implement it. We've been planning on it. We put together a proposal last year to expand our team, to hire someone that would specifically focus on global compliance. So here's what I'm seeing now is that the European Accessibility Act incorporates WCAG 2.1. It will soon be 2.2, so you should really be on 2.2. But it's more than that. It's also going back to the focus of the customer user experience. And there's a new conformance documentation process. It's really gonna be important for anybody that has the consumer electronics seal on their hardware. That's gonna be, in order to get that seal going forward, you're gonna have to show that you're accessible after June. For websites, it's a little bit more vague, but essentially we're gonna have to have a conformance. If you're a tech company and you're planning to sell to consumers and small businesses in Europe, you should be already a WCAG 2.2 AA as your standard, and you should already be investing in compliance, whether it's a VPAT or something else. But that's what you're gonna have to really be focusing on. And starting in June, you're gonna have to start really making that a part of your development process. I think that we're gonna see the European Accessibility Act become the new standard. It's already being adopted by Canada, and India, and Australia, Kenya. As a platform, as a tool, this is going to become the new, and we're not gonna see the American laws be as relevant. So we're going to stop focusing on, as a company, we're still gonna meet Section 508, but we're not going to be treating that as the gold standard. We're gonna have to start focusing on EAA as our gold standard.
- Yeah, and the fines are pretty severe. They're, the way that it's set up is that every single member can create their own fines. So it adds complexity to the situation, and potentially there could be very, very large fines.
- Yeah, we're waiting for the countries to release their fine structure. We still don't know what it is, but a lot of people have been reading the tea leaves from GDPR, and thinking, well if GDPR is an example, will the accessibility fines be similar to GDPR for? For those that don't know, GDPR was the European Global Privacy Act.
- Yeah, the cookies that pop up, and the fines are huge there.
- Yes, it's gonna be interesting to see how it plays out. It really is, especially country specific. Talking about being international and trying to support international initiatives and team members across the globe. I'm not sure, Ted, what size your own team is, but I don't think any accessibility product team is ever large enough. It always comes down to scaling across the organization. So what's your approach and what's your strategy at Intuit to scale accessibility across the org?
- I've known a lot of people that have scaled quickly. You know, they go from one person to like 10 people in one year. And I've always been kind of hesitant about that. We started with one person. That grew to an intern, and then another intern. And we've slowly built up, right now we have five people. But we're five people that have very specific roles. But then we also have our Accessibility Champion Program, which we launched about, I wanna say about six years ago. And that has three levels. The first level is we want everybody to learn about accessibility. They learn what is accessibility, what is inclusive design, disability etiquette. They set up their computers for success, you know, like set up your Mac for keyboard access, install some plugins, and we ask them to make a commitment statement. That's all they have to do. So it takes, depending on their role, up to 30 minutes. After they do that, they get added to a Slack channel for accessibility, and they get a badge on their profile at work. And I just sent out a bunch of stickers to people. The goal of that is to have that common knowledge. There was a quote at the recent conference in London last year, Tech Acts, no, I can't, I'm sorry I'm forgetting the name of that one. But the comment was--
- Tech Share Pro.
- Tech Share Pro, that's it. The comment was, we don't have a lack of experts, we have a lack of common knowledge. And that's one of the things we've been trying. Because of this Champion Program, we've had like 3000 people complete it, out of a company of about 15,000. You know, we've had a significant portion of our company complete it. I don't have to talk about the basics anymore. We're not having conversations about what is alt text, or what is a form label, or you know, color contrast. That's all everybody has a good understanding that we're talking about the intricacies of keyboard navigation, and ARIA usage, and content design. Then we have a second level for those that wanna be our product leaders, and our regional leaders. And they have to learn much more about customer empathy. They have to do volunteering, customer reviews. They have to do things like create a closed caption video, create audio descriptions, take online courses, 'cause we want people to understand who our customers are. And that builds into what Intuit is about as customer empathy. They then go into our level two champions where we have a community of people that can then work with each other. And then our third level is for our global experts, and they have an impact across the whole company. So that would be everybody in our team, plus we have people across the company that have reached level three. And the process for that is about leadership. So we're not, it's not all about being an accessibility expert, it's about understanding customer empathy and leadership. And in the along the way, you're gonna learn about accessibility.
- Yeah, and have you seen like a continued success on the return on that as it relates to, you know, finding issues quicker, or resolutions with customers faster, and just, I'm assuming there's multiple benefits for that type of hierarchy of engagement?
- Yeah, absolutely. It's a matter of, instead of spending a day trying to figure out who to talk to, it's now go over to the channel, type in a question, who's responsible for this? Or we just had a customer ask for this help, and then a champion will say, I'll take on that, and they'll work directly with the customer to solve a problem.
- ChatGPT, when it arrived, that was a real pivotal moment. What are your thoughts in terms of the impact that it has had, and has it changed anything on the enterprise level? If you can speak about Intuit specifically, or just your thoughts in general on the enterprise, and then after, let's talk about how it might've affected your friends or just people in the community.
- It was about six years ago that Intuit decided to invest heavily in AI. And what that meant was we spent five years just cleaning up the data, 'cause that's what people don't realize is if you want to create effective AI, you have to have clean data. And so we created our own language model that's focused on finance, whether it's business transactions or tax. And so after investing heavily in AI for those, we are then able to build the AI functionality on top of that. I would say ChatGPT kind of accelerated the course for companies like Intuit that have been on a progression. When ChatGPT was released, it was like, okay, now people wanna see these in action. So we kind of sped up the release of those. An early example of that that I really like to say is we had QuickBooks Capital. Now the problem with small business owners is you need capital in order to survive. In order to expand, you need to have cash flow. A lot of people are either afraid to go to a bank, or they've been to a bank, and they've been rejected multiple times. So they don't get loans. And when I'm talking about those people, I'm talking about underrepresented communities, people with disabilities where you're not only asking for a loan, but first you have to prove that as a blind person, I can run a business. So when we ran QuickBooks Capital, we didn't base it on, you know, judgements. We looked at the content of their QuickBooks, how much, what is their cash flow, what is their average invoice level? How quickly do they pay? Things like that. From there, we were able to approve people for loans within, you know, 15 minutes. And I think it was something like 60% of the people that got loans in the first year had never gotten a loan before. And that is the AI that I wanna see. I wanna see AI that can take away bias and provide real good answers. On the other hand, I also wanna see, which is something called explainable AI. If you did not get approved for a loan and that was determined by AI, I wanna know why I didn't get approved for that loan. So there should be something in that AI response that I could say, where in that thought process did it decide that I wasn't able to get a loan? And that might mean that you have, you know, your cash flow balances out of whack, and you've gotta work on that first. But that's the kind of thing we need to do with AI. Let's get rid of bias and let's get real good strong functional results, but also let people know, how did that happen?
- That's really great because I think small businesses, it's really hard to get a loan at all unless you don't need the money. And then if you don't need the money, then they'll throw the money your way, but then they're gonna expect you to use it even though you don't need it so that they can get their margin. It's really kind of messed up. So it's good to see that you all have been creating something that can work for small businesses.
- One of our most recent ones, and this also goes back to accessibility, making sure the interface is accessible is important, so that when we're delivering a page that has functionality, we wanna make sure everything is well coded. But I also like this one, because it's knowing that people need cash flow as a small business, it automatically looks at your records every day as a small business owner, and it finds where you have outstanding invoices, or where you have a client that doesn't pay their invoices regularly on time. And so it presents you with the list, like here are five invoices that need to be paid, and it suggests how you would write the letter to 'em. Is it a friendly note? Is it a demanding note? It gives you the suggested text, and then you can read the text and make any changes. If it all looks good, then you hit Send, and it sends the invoice back to the customer and reminds them to pay for it. But it also has that explainable AI. So there's a section that says, how did I choose this company and this invoice? And it explains the factors that went into choosing that invoice, and why it decided you should use a friendly versus a, you know, aggressive tone.
- Tone, you can dictate the tone of it.
- Yeah.
- But the first point you touched upon about the data, and ensuring that the data is as inclusive, as comprehensive and inclusive as possible. Joe and I have been talking about this for some time now, and you know, we agree that a key part of that, again, is engaging with individuals with disabilities as you build out their models is such a key piece, or we'll be playing catch up again, which is the last thing we want to do. What other areas either you see short term, long term, or tools that you use from an AI perspective do you see helping with accessibility as we move forward?
- Going back to what you just said about incorporating users, I think I'm hoping that we'll see a good model with Be My AI, where Be My Eyes has worked with OpenAI to work on image descriptions, and they had a really good cycle from what I understand, as to customers saying this was a good, or this was not a good reading description, this was helpful or not, specifically with customers that are blind or low vision, because that really, that's like the extreme edge of image descriptions. The most, people that need it the most to be clear and accurate. If they did that with the general population, the feedback mechanism wouldn't be anywhere near as good, and they wouldn't have image descriptions that had developed so quickly and so well over a short period of time. I would like to see more of that. So as companies are developing their AI models to incorporate customers, a broad spectrum of customers, but specifically those that are typically not included in research studies so that you can get that the edge cases and the better data.
- When we were preparing for this podcast, you mentioned a company called personal.ai. Can you describe a bit about what personal.ai is and then what does the future of accessibility leaders look like in a world where you have personas like that?
- Yeah, Personal AI is something that a lot of people may know Aira, I've already mentioned him a few times. Aira was created by Suman, and I don't want to mess up his last name, so I'll just say Suman. Most people know him as Suman. So Suman started Aira, and then after Aira got established, he left Aira to create Personal AI. And originally it was gonna be a small language model. So it was basically training an AI on yourself, your friends, your text messages and such so you can create something that helps you manage your day. But they switched to a new idea where it's like you're creating personas. And those personas, if you think about it, can be really good. So a Chief, I don't wanna say Chief Accessibility Officer easily, because that could be scary to a lot of people. But let's just assume that they create a persona for a Chief Accessibility Officer, like myself. I'm not a Chief, I'm not a C level, right? But just assume that you wanted to see staff accessibility person. Most of that work is going to be redundant. Not redundant, but like repetitive and tedious. That seems like expense reports, gathering reports, data metrics, and creating those. Those are things that an AI engine should be doing really well at. I think that there is a really good potential for companies to adopt a persona that could take over a single position. And Personal AI is doing that. Their CIO, their COO, their brand marketing, and their mascot, the dog, are all AI-powered. I'm curious as to how that's gonna happen in the future. My hope and expectation is because of Suman's background in accessibility, that everything that Personal AI has done has been done with the best practices. And I have no reason to think that it wouldn't be. But I think that we're really in a kind of an interesting shade right now is like what is the future for, A, an accessibility leader, if you can be replaced by an AI persona? And I think that a lot of companies will, especially as we're moving towards a global compliance space, instead of a user space that a company is low on resources may end up using a persona that can send out the weekly notices, the reports, understand the law, what's required, give code examples, review code, review possibly design. That could be what we see in the future.
- I'm proud to say at ServiceNow, we are trying to get ahead of the game with accessibility and gen AI on a couple of new features. One coming out next month, which is that conversational AI interface, which we are trying, will be embedding across our platform. Another one's a more intuitive page summarization that everyone can use. From an intuitive perspective, is there anything going on in your world that ties the gen AI and accessibility words together?
- So one of the things that I think every company our size is doing is using design systems. And design systems make it a lot easier. 'cause not everybody's building the same button, the same components. But one of the things we're trying, and I'm sure everybody else is doing, is taking your design system components and documenting them and standardizing them in a way that AI can then understand how those components are working. You don't necessarily want someone to be using Figma to build a page, and then all they have is like images of an accordion. It would be really nice if they could build a page, and it's got the actual accordion component in there, so that when it's exported, it's got all of the richness and the accessibility and theme ability. I think that's gonna be a real game changer when designers can put together a design, export it to a page, and it's built with all of the best practices, and not something like Angular or something like that, which, you know, a lot of us are still licking our wounds from Angular.
- Yeah, so it's interesting conversation about the point at which AI would take over jobs, and we're getting closer and closer to agentic AI. That also brings up a closed source versus open source debate. I don't know if you're familiar with LangChain, but you have like these frameworks, and you have people creating open source agentic AI, which so far, I haven't heard anybody super excited by any of it. You have the closed source deep research that you have to pay OpenAI $200 a month to try. And it looks like people are saying it's a pretty big game changer. I almost bought it today, but I don't wanna spend that 200 yet unless I'm convinced. And then on the other hand, you had China coming out with a DeepSeek, and so it seems to be a battle between the open source and the closed source. Which one do you think is going to win this battle, and just how do you see the landscape from the news you've been looking at lately?
- For me, I think open source is going to empower true innovation. And what I think about that is, I think going back to Africa and the M-PESA mobile payments, where you had people that needed to make payments digitally, and all they had was a SMS-based phone. They had limited resources, limited technology, but they had a big problem. And they solved that problem in a way that scaled tremendously. And in the way that they built it, they established whole new businesses and infrastructure, and it is spread across the world. I hope that the open source AI is going to help us get some true innovation from people that don't have large amounts of resources, where they're constrained, they have problems, they wanna solve those problems, and they can use open source AI. But I don't think it's going to be used for enterprise. I think enterprise, where you have a lot of resources, you need that security, that privacy, that strength of a closed system. I also think, I know when Llama first came out, there was a lot of problems with people that were able to take Llama as an open source and bypass any kind of safety regulations. And so if you have a lot of open source AI, there's really nothing keeping people from using it maliciously. So that's also something we have to keep an eye out for.
- Do you ever see, like, and I've had this question posed to myself about AI replacing testing, manual testing.
- I think that's gonna happen. I don't think it's gonna happen entirely. I just spent a week doing nothing but manual testing on one of our products, because we've been using automated testing, and generally the automated testing provided a good product. But it took me a week of going through manually, and actually testing the alt text, and finding those issues that were created with pseudo headers and pseudo buttons, and things that can't be done with manual. I think AI with agents should be able to see an image, understand the context of a page, and does the alt text relate to the context of the page? I think it can look and it can start doing things like, if your alt text has hyphens in it, it's probably not good alt text. So let's highlight that as something that needs to be redone. It should be able to do keyboard testing, so it should be able to mock up a keyboard, and it should be able to say, these are the things on the page that look like they should be keyboard accessible. And now let's run a keyboard through it, and let's see if everything gets, you know, treated as if it's keyboard accessible. I would hope that it could also look at a page and say, this text is really big, what is the code underneath it? Is it a heading or is it just a div? So I think AI is going to be able to replace a lot of those. So we might see that automated testing is doing, you know, the 40% right now. It might go up to more like 60 or 70%, but I think there's still going to be the need for a manual review just to make sure that it's usable. That things, even though they hit all the check marks, that you actually have an accessible experience. 'Cause there's still, AI can still only mimic a user's experience. They can't replace a user's experience.
- My thought on the alt text is that it's gonna move to the client side, to the assistive technology, because they think that different users have different preferences. So they might want you to really go into a lot of detail on the image. Others might say, I want it to be really short. And then maybe what it's gonna look like from the business side is that you have branding where you may wanna make sure that the alt text is worded in a particular way that's good for your brand as opposed to allowing that to be personalized on the other end of it. So, just a thought there.
- Yeah. I'm actually gonna be doing a presentation at CSUN with Marco Salsiccia about that concept, because it's like, alt text, we had really bad alt text for the longest time, and then AI started to appear, and then we're like, hey, we could actually, as you're uploading the image, can we do this into it? You're uploading an image to our content management system, we're running that image through object identification. It's then creating a description of that. We're then running through our content design guidelines to make sure that meets our voice and tone. And then it's putting that description into the images within our content management system. At that point, the person that uploaded that image should review it, and they should say, yeah, this is a good description or it's not a good description. But it's still, so what does that mean when it goes onto a page? And that's where we're bringing in screen reader users. So if it's a picture of two people pulling, you know, creative tomatoes off of the back of a truck, that might be a really good description. But what does that have to do with the page? Is it appropriate description? And so this is where we're saying, well maybe we should have a smaller description, and if someone's intrigued by it, they can always take the image and then ask their AI to describe the image further.
- Yeah, query it.
- We were looking, I did a review with Lucy Greco from UC Berkeley, and we were looking at a recipe website, where they had alt equals quote quote, on all of the images of the recipes. I was like, which of these recipes would you like to make? But because they were alt equals quote quote, they were removed from the image list. So she couldn't actually highlight them to have her assistive, her AI describe the images. She's like, there are no images on this page.
- Oh, that's so frustrating.
- So it's, you know, it's a rollercoaster.
- Since you mentioned CSUN, perhaps you could explain to the audience what is CSUN, what you do, you know, how you're the Program Chair essentially, and what it means to you, because I know it's so important to so many people.
- Yeah, a lot of times people in the accessibility industry, we just say CSUN, and assume that everybody knows what we're talking about. CSUN is the California State University of Northridge, and every year they put on the Assistive Technology Conference. And this year is the 40th anniversary of that conference. So if you think back in 1985, these were people that were really working on the initial screen readers, and DOS, and initial Microsoft products, moving on towards Windows and then the web and everything. It's that one conference around it that people from around the world come to, and you have this amazing gathering of people exploring accessibility, inclusive design. My favorite part of it is, it's not just engineers, but there are actually three sections. There's engineering, design, web, and then there's also teaching, and then there is also rehabilitation and independent living. And they're all in the same building together. So you have a broad spectrum, and I always spend most of my time in the teaching and rehabilitation spaces, 'cause I know enough about web development, I don't need to see yet another ARIA presentation. But I'd really like to see that, you know, $300 electric wheelchair that was created in Brazil that can go into someone's trunk. I mean that's the kind of thing I wanna see. I wanna see like there were, the sessions that stand out to me are the ones that are things I don't know anything about, when you're, there was a fantastic presentation from Japan about the visibility of contrasting colors on steps. Should it be like really high contrast yellow and black? Or should it be a medium contrast brown and grit tan, you know, things. So that's what CSUN is. A lot of people go and they never attend a single session 'cause they just sit in the hallways and talk to everybody.
- So on that note, for those individuals that have never been but would like to go, what would your suggestions be to them?
- If you're too late to see CSUN this year, then I would say definitely put it on your calendar for next year. If you work for a company as their accessibility team, you should really lobby heavily that you need to go, because it's, you're going to learn things that will change your company's accessibility experience. If you can't meet CSUN, I would highly recommend you go to Axe-con, which is a free online conference. Very, very good. And Inclusive Design '24, which is another online conference that has speakers from around the world.
- I'm proud to say, I brought my entire team to CSUN last year, and we also used that, we added on an extra day and used that as a team offsite, and it all, everybody found it in valuable. It really was. It was an incredible experience in many ways.
- This year, we're taking our entire accessibility team to Los Angeles the week before, and we're doing an offsite so that we can all gather and plan for the next year. And then we have the CSUN, so we're actually making it into like a two week event.
- You once gave an award to a researcher, who through, I think it was web browsing, that would detect autism, am I saying this right? Autism through web browsing, something like that. Can you explain what that was about? And then is that related to AI as well?
- Intuit's been a sponsor in the Web4All Conference. And the Web4All Conference is for accessibility research around the world. And then once a year they come together, and it's co-sponsored with the Web Conference, which is all web technology, and then the first two days they have sub conferences. So this was a project, a team in multiple countries, and they were trying to figure out what's keeping people that are autistic from effectively navigating search results. They used, you know, specific page that they could, you know, fine tune what they're looking at. In that research, they realized that people that were autistic, their eyes tended to go in different patterns, like bounced around a lot compared to the non-autistic control group, which was more like a linear, what we kind of expect, which is kind of a Z pattern or just a linear. And that they showed one year they said, this is why we think people are autistic, have a, take longer to review pages. Then they thought maybe we can flip this on its head. If we know that people that are autistic have a different eye gaze pattern, can we then look at people's eye gaze patterns to determine if they're autistic? So they then reran those results against the original group, and then they expanded that group, and what they got was about a 75% accuracy that you can detect that someone's autistic from looking at their eye gaze. And this was never meant to replace medical diagnosis of autism, but it's a great way of early detection in regions where they don't have that kind of medical support. And if you can detect that a child is autistic at three years old or four years old, and start giving them that kind of education, I don't wanna use behavior therapy, because that's a kind of a loaded term, but to be aware early on can have a great impact.
- And can changing the way a page is displayed be helpful as well, because that's something that does seem like a good thing, good use for AI to personalize. Oh, why don't I just change the order that the content is displayed for the user.
- Yeah, you know, this year, Gregg Vanderheiden just retired. For those that don't know who he is, he started the, he's one of the original accessibility leaders, and he had this project called Raising the Floor. And the concept of Raising the Floor is that every person should have a personal experience. If you go to an ATM, and you put in your card, the ATM should know that you need bigger text, a higher contrast. And so the ATM should adapt to your needs. If you're on an airplane and you're watching a movie on an airplane, it knows who you are, so it should be able to automatically turn on closed captions. That's always been the goal is this concept where technology adapts to the person. And I think that that's where we're gonna start seeing is that personalization. Why does, you know, why do I need to have that standard three column layout on a page? Maybe I don't, maybe I want you to take that information and blast it into a radial diagram, you know, where like the key concept is in the center of the page and I can actually navigate through, you know? I'm just throwing that out from the top of my head.
- Yeah, I mean.
- Why not? These are the things that we should be seeing in the future.
- Yes, it's something that we're actually looking into and other companies looking into. Like even when you register as a new employee, maybe you have dyslexia or low vision. When you, right away, all your applications are set up within preferences to suit you, the personalized use. Just a simple example, but we're really starting to go down that path. For those listeners, Ted, that maybe don't know a lot about accessibility, want to get into it, or just starting their journey, what would your advice be?
- I think what's always been important for me is networking. It used to be that LinkedIn was really great, because we were, it was new, and you know, you may have like five or 10 connections, but nowadays people have 5,000 connections. So I wouldn't say that LinkedIn is gonna be your best solution, but what I would suggest is reaching out to people that you admire that has done something, contact them. It usually they have some kind of contact mechanism with real questions. Join community boards, work on your stuff, explore things, create examples, have a blog, join Blue Sky, join other social media networks so that you can actually start building that network. The thing that made me effective over the years is I know what I don't know. I always knew that. At Intuit, I knew I was the worst JavaScript programmer in the company. So I knew that that was my weak spot. So I always worked with people that were great at JavaScript. And so if I had a question, I always knew who I could go to to find those answers. At the same point, I was like really, really good at web semantics. I knew every single item in the HTML and how they're used, and people would come to me and say, "Well, how do you use this tag?" So I mean, we all have our strengths and weaknesses. Get to know what your weaknesses are, and then move forward. Now going back to AI, that's one of the problems I have is that a lot of people are gonna shift to ChatGPT as their guru, as the person they ask for all this information. And it's not always good. We've been talking about alt text. You go to ChatGPT and say, "How do I make good alt text?" And you're only gonna get like that one response, and you don't get the broad spectrum of how alt text is useful for different people. So I would say definitely networking. If you're in a university, see if you have Teach Access, which is a collaboration between technology companies and universities to increase education of accessibility, I would say volunteer. So in your local communities, find organizations you can volunteer with. Go to conferences, even if it's a virtual conference. There's chat rooms, so that you can start reaching out to people. On LinkedIn, there's groups. So for instance, the Champions of Accessibility Network is a great group for any kind of company that has an Accessibility Champion Program. And they host monthly meetings that are wonderful. So I think that that's where it is, is it's about expanding your scope and your network. And the other thing about accessibility that's great is that we tend to have very few people with big egos. So the barrier to entry is much lower. If you reach out to someone and you have, you're sincere, and you're saying, I have a question, or I'd like to know this, you're probably gonna get a really good response.
- It's also one of the few areas, I don't know any other area, vertical, department, whatever you wanna call it, that like even with our competitors, Salesforce, Microsoft, like we'll share everything. We'll share our best practices, we'll share how we're embedding, how we're scaling our KPIs. There's very few other areas that do that. Very, very few others. It's very collaborative area. It really, really is. It's awesome that way.
- You'll hear a lot of people say, we don't compete about accessibility. There might be now, your third party contractors might compete. But when you're talking to Salesforce or Intuit, or any of the other companies, you're not going to see us competing and hiding information.
- Ted, this has been really great having you share your wisdom with us. Before you go, can you tell our audience how they can read anything you might be writing or how they can contact you?
- Yeah, I had a blog that was alive for many years, and it got hacked, and hacked, and hacked, and hacked. And so I lost a lot of the content, but it's back. So the address is last-child.com, which is a reference to a CSS role, but also I'm the youngest of six kids. So last-child.com
- That's cute.
- And what I've been doing is going on internet archive and finding some of the better articles that I had used and people referenced, and I've been restoring those articles. So slowly, it's building its way back up. But that's a good opportunity. And you can also catch me on Spill, and Blue Sky, and Mastodon, and LinkedIn.
- Yeah, thank you so much. Awesome discussion. As always, we learn so much with humble individuals like yourself, but we do, we thoroughly appreciate your time. Thank you.
- If there's one thing I can say, I was telling Joe this earlier. When it comes to AI and accessibility leadership, we're all imposter syndrome victims. None of us know what's happening with AI. So if you're worried about yourself as an accessibility leader in AI, the best thing you can do is just learn, and try, and keep your mind open. And just know that, I mean, Joe's probably the one in the accessibility world that I know that has like the most broad expanse of AI, outside of like, you know, like Suman and some of the Envision AI. But even then, he feels like he has imposter syndrome. So just be, you know, take it easy on yourself if you feel like you're falling behind in AI, 'cause everybody is.
- I actually think there's a lot of positivity from imposter syndrome, because it keeps you, people are humble, and they're always willing to learn, because you never think you're good enough. So I think, you know, it's not all negative. I think there's a lot of positivity from the syndrome, I do, you know?
- I'm reading Fei-Fei Li's book, "The Worlds I See". And she's one of the pioneers of AI, and she, you know, referenced imposter syndrome. So if Fei-Fei has even a tinge of that, I think we can all just take a deep breath.
- Exactly, and enjoy the humility.
- This is a great point. Thank you, Ted.