- Welcome to "The State of Accessibility" podcast from TPGi on LinkedIn Live. I am Mark Miller, and this is my co-host, Dr. David Sloan, Chief Accessibility Officer for TPGi, co-author of "What Every Engineer Should Know About Digital Accessibility", and a user research and accessibility strategy specialist.

- And Mark is a sales director for TPGi and a member of W3C's Web Accessibility Initiative Accessibility Maturity Model Task Force.

- Hey, David, well, I'm happy to be here with you. We've got a lot of stuff to talk about. This is always an exciting time of the month for me. So we have a few things to cover this month, and then later, we're gonna talk about disability, the Disability Annual Conference, which you and I attended last week. And then most importantly, we're gonna talk about tater tots.

- Yes.

- All right, first, some legal action relating to the EAA already.

- Yeah, so we learned that last week, a group of disability advocacy organizations in France sent a complaint to four major French retail stores regarding the accessibility of their e-commerce websites. E-commerce websites are part of the services and products covered under the European Accessibility Act. So this group of organizations sent up demand or a request, depending on how you look at it, for those retail stores to meet EAA requirements for their websites by September 1st or face legal action. So these organizations are not waiting for France's market surveillance authority to act. They're already starting to apply the pressure early. To my knowledge, I think it's the first instance of where we've seen the EAA just to be used as a legal tool. So we'll definitely keep track of that case and anything else that arises over the summer. So it's an interesting to see already, you know, that's, what, two, three weeks after the deadline that things are already happening in terms of sort of enforcing the legislation.

- Well, that's a great update, and I'm glad that you brought it to us particularly since I think that was a big question we all had. I was like, "What happens after we cross over this deadline?" And there's probably a lot still to happen, of course. But this starts to give us an answer.

- Yep.

- Okay, so what I wanted to do from here, David, was go back to last month's conversation, which was on AI, and last month, we sort of dove deep into AI and accessibility. We talked a lot about AI's role in enhancing assistive technologies like Picture Smart and JAWS. And then we also talk about AI tools and how we can create digital content with those tools, which is starting to become a thing. And then the sort of final point I think we touched on was AI tools changing how we test for digital content for accessibility. And we're not gonna dive back into those things today, obviously. If anybody wants to see those, these "State of Accessibility" podcasts are also re-posted on the TPGi website and show up on YouTube. So you can go back and take a look at that. But there was a little bit more that we wanted to cover as well, right, David?

- Yeah, yeah, I mean, I think that any discussion of AI tools to help advance accessibility efforts also has to be accompanied by just being careful about the risks of pinning too much faith or hope in AI in improving accessibility when there's a question of whether AI reduces or perpetuates disability bias. And this is part of this broader ethical conversation over just how much can AI do for us positively versus potential negative consequences. You know, as we talked about last week, we know that AI can really help people with disabilities in terms of empowering assistive technology, tools that allow people to communicate and create content. So there's really positive things and it can help a lot of accessibility tasks, but the nature, especially of large language models and generative AI, is that they form and provide back to us knowledge and what already exists, what they've learned. And if what they've been trained on has inherent bias, and we know that out there in the world, there is still bias against disability or disabilities just kind of less prevalent or prominent, then if what these tools are trained on has bias, then it takes concerted effort to remove that bias. So we have this challenge of how do we get tools trained on content that potentially reflects the biases that exist in society to help us address those biases. You know, we'll have to be very, very careful about that.

- One other thing, the question marks that I've sort of had floating around in my head when it comes to AI and exactly what you're talking about here, the fact that it is trained on a particular data set, and it's only as good as that data set, right? Is that AI, you know, AI is an amplifier or an accelerator or whatever, like, that's why it's valuable is because it can parse through that big data set and do stuff faster and faster and faster. And so it makes me also wonder if there's a potential for AI to perpetuate things like biases. Like, if biases exist in that data set, that's what we're operating on, to your point, like, we need to be careful of those biases, but is there a danger that we then so rapidly reinforce that bias by putting it out in a world that it kind of runs outta control? So I think it's a, and I'm not saying that's happened, right? Like, I think I have a pretty good understanding of AI and large language models, but there's all these things that we just have to wait and see how they unfold. And that's one of the ones that I kinda think about and worry about. And I think we should all be cautious in paying attention in that way.

- Definitely. And I think the other angle is that, you know, we talked about how AI tools can help people with disabilities, you know, whether they're assistive technologies or other tools that make tasks easier or make them possible for some people who couldn't do them before. But there are other tools that are learning from how we use them. And those tools have inaccessible interfaces. So that means the tools are only learning from people, non-disabled people because of the accessibility issues with the tool interface. So the tool isn't learning from a diverse population and therefore, it's kind of embedding its learning from a subset of the global population.

- Ableist scenario, right, where more data is being fed in-

- Yeah, yeah.

- bias through something that's essentially almost, like, ableist.

- Yeah.

- Well, you know, I know there's a lot of discussions around AI and I really hope that this stays at the forefront because it's an opportunity. It really is an opportunity.

- Definitely. And I think it's worth calling out, you know, that there are definitely efforts to train models on quality information, and in our case, quality information about accessibility. You know, people are providing ways to interrogate high-quality knowledge bases, you know, we are doing this work, other accessibility provider are doing this. So we, in the accessibility field, know the limitations of general models and we know the value and the techniques that can be used to fine tune or, you know, other techniques to make sure that large language models are focused in a specific area, this kind of narrow AI concept. But there's a bit of a race between these advanced LLMs with really, you know, they're trained in quality data and then the ones that are more generic, the ones that general world knows about and is using, which may not be providing accessibility quality. So, you know, who's gonna win that race?

- Yeah, well, I think that's just as people get used to using AI, and that's one of the things that was a kind of a epiphany for me is, like, one, AI is not this sort of singular thing because the first experience a lot of people had was with AI models that had access to some version of the web, which is where that biases coming that you're saying. But that AI is sort of an engine that can be run in several different ways. And if it's run on a data set that's trained, it's got specific knowledge for specific, just like you have specific knowledge in accessibility user experience, right? Because that's what you've trained yourself on over the years, you'll have that kind of specificity. So I think as the world gets used to using AI, it will get used to identifying, picking out the AIs that makes sense in their circumstance and being critical, just like we're critical about the ingredients that are in our food, we may need to start becoming critical about the ingredients, AKA dataset, that's in our AI, right? And that might be something that people start to look at and evaluate and, you know, is posted and all that kinda stuff. It may already be and I just may not have seen it.

- Absolutely.

- I think that that's a great point and I think it's great to keep that conversation going so it stays forefront. But when we think about organizational accessibility strategy, which is essentially how an organization goes about including accessibility in its practice, whether that is employee-facing, customer-facing, public-facing, how does an organization become successful in baking that into everything, into the DNA of the organization? So how should AI help change our ability to do that?

- Yeah, and I think like any other adoption of AI tools, an organization needs to look at the risk and opportunity of AI tools to manage accessibility strategy. You know, there are definitely opportunities where AI can make accessibility work easier. You know, work that's thrown to human error or takes a human a long time to perform or is inconsistent, you know, within individual's ability to do that work or between individuals. So having that consistency and things like automated testing of code and content against standards, filing bugs in a standard way, helping humans ask questions, interrogating those quality knowledge bases that we were talking about earlier. These are all great opportunities for tools to either automate tasks or help a human do a task more efficiently or effectively. And that idea is that frees up humans to be more strategic and more creative. So rather than saying, "Well, now we can fire all of our accessibility team 'cause AI does the work for us", we know that that's just not a reliable strategy just now. And we need to caution against people who think that advances ain tools mean that there's no need for subject matter expertise. So the idea is that the tools free up humans to be able to do things like, you know, work with people with disabilities and research and design in validating progress and making sure that, you know, we're maximizing, improving what we build and minimizing the harm. One thing that I became aware of fairly recently, hat tip to Janis in Ascension for sharing this, was an AI tool accessibility leaderboard. So that URL is aimac.ai, that's AIMAC.AI. And this is a leaderboard that ranks the quality of AI code generation tools for the accessibility of their output. So this is a way, you know, if you're looking at the relative merits of tools that can be used to generate code and content, this website monitors and it's using kind of standard prompts rather than pumping a tool with, you know, provide me with an accessible panel, for example, provide me with a tab panel. So it's saying, can the tools generate accessible code without being asked for it? Which is what many people are gonna be doing rather than saying, "Build me a app panel and make sure it's accessible." Some of us will say that, but some of us don't think to do that. So this is a nice way to track the relative merits of tools.

- It goes back to the ingredient thing, right? Like, if we're looking at that tool and we're looking at an AI tool, and we as a human, or as a practice within our organization, understand that accessibility is important, we're gonna make sure that that is an ingredient that is part of that data set and included. And it's just like many other things that it would assume when it generates the code, it's gonna assume that accessibility is important.

- Yeah, absolutely. So ultimately from a strategy perspective, I think AI tools are a way to help organizations move toward accessibility standards conformance in a more efficient way, which gives them more capacity to go the next step on improving user experience for everyone, including people with disabilities. And that's the idea. But if we say, "Well, we move towards standards conformance and then just forget about the rest," you know, that's where the problem will really start to escalate. So let's try and focus efforts and improving user experience.

- Yeah, well, you mentioned user experience, David, and we sort of take it for granted that that's something that we should be doing, right? And there's a reason for focusing on user experience and it's so that customers have a good user experience. Next level of that is so that they return and you know, if somebody has a good time with something, they're gonna keep coming back to it. And that, I promised a discussion on tater tots, and we're gonna go back to tater tots in a second and we're gonna talk about it as it relates to user experience. But the reason why we promised a discussion around tater tots is because of a conversation that you and I had last week over lunch. And yes, if you're wondering, David and I hang out with each other quite a bit whenever we get a chance outside of this podcast, right? We don't just show up for this and start talking to each other. This is really an extension of conversations that David and I have all month long. So this tater tot conversation came outta that, but before we get into that, let's just talk a little bit about Disability:IN, 'cause I thought it was particularly good this year. And I really had a great time and learned and made some great connections and was very encouraged around how things are being accelerated within businesses and within the disability and focuses on people with disabilities as employees of organizations, right?

- Yeah, for me, so this was my second Disability:IN, and I always kind of contrast it with other accessibility conferences I've been to over my time working in this field. And it is a little bit different, you know, it's more corporate, it's more broadly, sort of large organizations focusing on disability across different aspects in a lot of case hiring people with disabilities and supporting them through career pathways, engaging with disability on businesses. And then from our perspective, the digital accessibility, you know, making digital resources accessible, providing assistive technology that helps people do their work and helps people be successful engaging with these organizations. So it's always interesting to hear about the, you know, what large corporations are doing. For me, there was a couple of big things around visible disability representation. You know, there's always a suspicion that disability representation can be kind of tokenistic. But it did seem that, you know, there was a lot of people with disabilities there, a lot of people who identified as maybe, you know, being neurodivergent. And people who are speakers, people who are managers, leaders rather than people who are not in positions of power or decision making. So that's very encouraging, that the disability representation is very present and often people, you know, in leadership positions. So I think that's encouraging. And I think, you know, a lot of talk about accessibility driving innovative product creation. You know, again, still there's this tension between do we have to do it to be compliant with laws and things, or we want to do it because it helps us make better products?

- Which we're gonna get into when we talk about-

- [David] Yeah, yeah. Yeah.

- And I think one of the things that, and this is sort of near and dear to my heart as a person with dyslexia, is that there seemed to be a really nice focus on neurodiversity in the workplace. In fact, one of the most interesting things that I ran across was it was this neurodiversity in the workplace study, a framework for human capital management, and to have a lot of the things that we suspected about having a diverse workforce and a workforce that doesn't all think the same, right? That has different strengths, different weaknesses, and thinks in different ways. You know, I think we've been talking a lot about what we assume or anecdotally see as positive outcomes of that kind of a workforce. But this neurodiversity in the workplace, a framework for human capital management came in and did a study that put data behind a lot of those suspicions. We don't have time to dive in deep today. I'll be posting a lot about it and integrating it into my personal database of things around neurodiversity. Anybody who follows me knows I post quite a bit about dyslexia and neurodiversity. So I was really encouraged to see that. And to your point about folks who have, you know, non-obvious, non-apparent, there's a few different ways to say it, disabilities, you know, hidden disabilities, when you're talking about that you are being serious because you're not, you know, it's not case. It's something you have to be intentional about. So from that aspect, it was encouraging as well.

- Yeah, and I think that one of the out, or when people were discussing that research, that report you're talking about, ways, practical ways, strategies for making a workplace more neurodiverse-friendly. Many of those practices seemed like very good common sense in terms of, you know, sending out an agenda for a meeting in advance, structuring meetings, allowing participation in different ways, clearly summarizing actions at the end so that people know what was agreed and what's gonna happen. A lot of very obvious things it seems after once you hear them.

- Very true, and I think the curb cutout effect from focusing on neurodiversity within an organization is gonna prove to be big in this case, based on everything that you just said. So anyway, stay tuned or follow me. There'll be more about that. But all people really care about right now is our discussion on tater tots, David. I'm sure people are like, "Yeah, that's all well and good, but you promised us tater tots." So I want to talk about this experience that Dave and I had in a restaurant in the hotel. We had snuck off the two of us to talk about some things and to have dinner on the first full day that we were at Disability:IN. And we found this nice sports bar, and we sat down and both ordered. I ordered fish and chips, and I can't even remember what you ordered, David. Was it like a burger or?

- I think it was a chicken-

- Chicken sandwich.

- Chicken sandwich, yeah.

- So when the waitress came over to David, so fish and chips, like, mine's like, you know, it's gonna be fish and french fries, you know, end of story. What do you want to drink, right? But with David's chicken sandwich, there were choices, right? And the choices were french fries, a salad or tater tots. And I know what everybody out there is thinking. "Only one of those is the right choice." And that choice is, and David knows the answer to this now, so we'll test him. What is that right choice now, David?

- Well now I know it's tater tots. But I just need to say, so I'm from Scotland. We don't have tater Scots... Tater Scots, wow, that's a slip. Tater tots.

- Maybe that might be even better.

- Hash browns arrived quite a while ago. We know what hash browns are. We like frying things, we like deep frying things and we like potatoes. So it's just surprising that tater tots, or tater tots have never quite made it over there. Maybe they have now. I don't know.

- Well, so this is how the discussion goes, right? Is that I go, the waitress walks away, and I said, "David, you answered that question wrong." He's like, "What do you mean?" I said, "She gave you these three choices and you picked fries." By the way, which I'm getting as well, right? So we don't even have a swapping opportunity here. And I said, "The only correct answer is tater tots." And this is where it got interesting. David looked at me and said, "Well, I see that every once in a while, but I don't know what a tater tot is." So I explained to him what a tater tot is and what a tater tot is, if you don't know, is that some brilliant person looked at the mash that was left after fries were made, fries are shot through a grit, a potato shot through a grit, shot through a grill like a, you know, think of like squares, and out the other end pops the french fries, and then this mash sort of falls down and some brilliant person said, "Geez, what could we do with that?" And they decided to sort of pack it into this little tot thing and drop it in a deep fryer. And you got a superior product there, in my opinion. So David asked me a bunch of questions, like, "Is it like a hash brown? Is it like this, is it like that?" Anyways, the result of this is we decided that we needed to change, and we figured it probably wasn't too late that we needed to change the order and change what his answer was. So when the waitress was in sight, again, I did that really rude thing, David, which I apologize for, but I didn't know what else to do where I sort of waved her over and she came over. I'm like, I'm so sorry to do that. It's like, I know the wait staff has like got a plan, and as soon as you wave 'em over they have to come over 'cause you're the customer, but you disrupt their plan. So that was maybe not a good part of the user experience for that person, but I went into detail about how David had made the wrong choice and didn't know what a tater tot was. Anyways, we had this back and forth. She started to play right along and she goes, "Oh my gosh, that's a terrible mistake. We've gotta correct you right away." And we had a ton of fun with it. The end result was David got his tater tots and now knows the correct answer. But you know how a different person will often bring the food out than the person that's sort of taking care of you? That person brought the food out and this wonderful waitress runs from the other side of the restaurant to stand there and watch David eat his first tater tot ever, which was, I wish you all could have been there. It was an incredible experience. He now, his life, I think your life has just started at this point now that you've had a tater tot.

- Also drooling, yeah.

- We were thinking about this experience, and this is just, in my opinion, this is just a great everyday human experience. We could have called that waitress over and we could have said, "Hey, can we change your mind? Can you throw some tater tots on there?" We could have done a bunch of different things, right? We could have, I don't know, pretended like we ordered tater tots and complained in the end. Like there's a lot of ways that this could have gone, but we had a bit of fun with it. And we did that because we knew that we were asking the waitress to do something that was a bit disruptive to her day, that was not ideal or whatever the case is, right? And it was our mistake, so we had fun with it, and we brought her with it. And the result was is that she enjoyed it, right? We enjoyed it. We had a great time. She had a great time. There was this experience between three people, one of them whom we didn't even know, that really went, I think, quite well. And David and I commented on that afterwards quite a bit too. She also got a decent tip out of it, because I really appreciate it when somebody's sort of plays along with you like that. And it dawned on us that in our practice where we're so focused on user experience, why should you go above and beyond to make a user happy with how things go? That's why you should go above and beyond. We're talking about this on a podcast. I remember that waitress, that waitress probably remembers us. If there's an opportunity to go back there, we will be back there and I'll be looking around and going, "I wonder if that waitress that gave, you know, switched David fries out for tater tots is still around." And if you take that analogy and you apply it to corporate, to user experience, to whatever your product is, to however you want people to think about you, think about your products, think about their experience, that's it right there. You want them to run across the restaurant to be a part of something. You want to inspire people to that level. And nothing that we did, David, nothing that we did was difficult.

- Right.

- It was fun for everybody. It was fun.

- Absolutely. Absolutely, yeah.

- I think that the UI discussions around AI, around what you've dedicated your profession to, you know, the maturity models and stuff that I focus on try to help organizations create, what TPGi does for customers every day. And this is really going beyond the guidelines, right? Using the guidelines as a basis, including beyond that. I really think that that was just a wonderful human moment interaction in which David and I kind of had this very basic aha, like, this is it, this is why we do it. So we wanted to share that because if you're head down working every day and you're thinking about these things, we hope that that gives you a bit of a pause and reinforcement around why these things are important and how they affect people.

- Absolutely. My one additional addition to that fantastic way of telling that story is just if you make a mistake, then intervene and fix it rather than live with it. And you know, because the correction might not be as hard as you might think. And when you're building digital products and you realize early that you make a mistake in direction, whether it's accessibility or anything else, try to fix it rather than sort of live with it because the mistake will just grow and it will be harder to fix. It may not be possible to fix and-

- Poor user experience.

- [David] You might have to eat french fries.

- With french fries, David? That've been horrible. Right? And by the way, just so people know, there was a lot of tater tot stealing that went on after that. Not a whole lot of reciprocal fry stealing. The waitress, I think, at one point even helped distract you so I could get one extra tater tot out of the deal. That's a great point. I mean, I think that that's... And when you do make a mistake, I think owning up to it. We all know that that's the right thing to do. But also, you know, taking what you do seriously, but not taking yourself seriously, and that's what we did in that. I mean, tater tots versus fries, I'll admit right now that it's... No, I won't, it's very serious but . But you know, my point is, is that we could enjoy it. We could have fun with it, we could have fun with a mistake because we didn't take ourselves too seriously in that circumstance. But we took the situation, 'cause it was a dire situation, you almost got fries, very seriously. Well, David, I think, we covered tater tots. There's not a whole lot left, right?

- Yeah, I need to go and get some lunch having heard that conversation. But yes, that's a very good analogy. and I've learned a lesson.

- Excellent. Well, this is Mark thanking David. Now you know the state of accessibility. I'm Mark Miller, thanking David and reminding you that the state of accessibility is always changing, sometimes from french fries to tater tots. So please help us affect change.