AI is everywhere. You don’t have to like it or agree with it. But I'm pretty sure the genie can’t go back in the bottle.

I won’t hold it against you if you don’t want to use the tech. I understand there’s a LOT of real costs: the impact on the environment, the ethics of synthetic data, centering technology rather than lived disability experience, the unauthorized use of the entire internet and published books/journals/magazines etc as a source of training data, and the implications of this technology on people and their livelihoods.

I understand all of that.

I was catching up last week with a long-time friend and she said to me that because of ALL OF THAT, there’s all the more reason that people that truly care be involved in directing and shaping this technology and how it’s used.

She’s so right.

In my last days at Salesforce in Jan of 2025, I felt the pull to engage with AI very acutely – Salesforce as a company was all in on AI. The tech industry was going all in on AI. And I... didn’t know nearly enough about AI. Sure, I knew what it was, how it worked (generally), and some grounding in concepts that help create responsible and ethical AI… but I hadn’t used it for anything meaningful other than some “chats" or cobbling together a meal plan for the week.

I am a builder. A tinkerer. A creator. When I was a kid, yes, I did take lots of things apart and put them back together, just to better understand how things work. I need the words and concepts to have meaning; not in an intellectual “I know what those words mean” kind of way… but in a deeper “I’ve done the work and have hands on experience doing it” kind of way.

So I chose to educate myself by doing the work.

A growing list of 20+ AI projects that I’ve worked on solo over the past 8 months have all helped me understand AI as a medium. It started with migrating my website from Gatsby to 11ty, moved to creating a podcast annotation iOS app, and then designing and building a full scale Personal OS application that works the way I want to work.

I’m understanding AI as a tool. As an ingredient for making. As a platform for creation and bringing things to life that I couldn’t have otherwise.

When I described what I’ve built to a friend and former co-worker, he said “You’ve built your own assistive technology.” There’s some serious power in that.

When roles collapse, what’s actually collapsing?

We’ve seen a lot of discussion around the Collapsing of Roles – where we previously had designers, engineers, product managers – and those roles are being combined into a singular role: The Builder. What does that mean though, and what’s the role of the accessibility professional?

When we collapse roles, we also collapse judgement.

The judgement of quality and what’s worthy of shipping used to be distributed between design, engineering, product, and other stakeholders. As part of that, accessibility professionals have engaged with ALL of those roles for years to provide hands-on expertise through planning, testing, and guidance for preventing and fixing accessibility barriers in product.

We used to have many opportunities to flag accessibility issues before things ship. Now, judgement lies with one person.

With the collapse of judgement, what do we face as an industry?

• Concentration of risk: with traditional cross-functional teams, having 2 of 15 people with some background in or the desire to champion accessibility helps keep a team on track with delivering at least partly accessible products; someone was always there to call it out.
• Ownership: I’ve worked with enough teams to know that designers often think of accessibility as the engineers’ job, engineers think it’s on the designers, and the product managers say it’s definitely not on them because it’s the designers and engineers to handle in the details of the implementation. Is the age of the builder actually the age where we can definitively say “You’re the one that’s responsible for accessibility because you owned it end to end.” Cue the builders saying “I prompted the AI to make it accessible. It’s on the model, not me.”
• Lack of traceability: When cross-functional teams work together, there’s often handoff between the teams. Collaboration in the process. Discussion about decisions and tradeoffs. When decisions are buried in an ephemeral prompt, how do you do root cause analysis to determine what went wrong, how to fix it, and improve the system?

In my experience (yours may be different, of course) – almost every organization I’ve worked with from the outside or the inside is already NOT good at the traceability they need to feedback into the system. The focus is almost always “fix the bugs and keep moving.”

The systemic fixes that make complex, interconnected systems work well together almost always take a back seat in favour of shipping quickly.

So… now what?

I don’t have a tidy answer. I don’t think anyone does quite yet.

The accessibility work I’ve watched hold up over decades has held up because someone with a disability was in the room. Not consulted at the end. In the room. Asking questions. Naming things.

That part doesn’t get easier when the team is one builder and an AI. It probably gets harder. So the question I keep coming back to is: how do we create a human-centered, AI-driven accessibility model that keeps lived experience as part of its core?

I have some hunches. I have things I’m trying. I don’t have a clean framework yet. If you’re working on this too, I’d like to know what you’re experiencing. ### What’s next?

Quick note for those who replied to the last message: I said I’d go deeper on the capacity problem, making accidental success repeatable, or the team-of-one topics. I will. I have things to say on each one. But this AI arc grabbed me, and I want to follow it for a few messages first.

So this is the start of an arc. The question I posed up top is one I’m going to keep working through.

Next time I’ll show you one of the AI experiments I’ve been working on and what I’ve learned from it. After that, I want to go deeper on what “human-centered, AI-driven” actually looks like when you’re trying to do the work, not just talk about it.

If you’ve got thoughts on the question, hit reply. I want to hear what you’re seeing, what you’re worried about, what you’re trying. Your replies are part of how I’ll work through this.

Derek

Looking for reviewers: book draft

I have a book draft I started in 2018, kept writing through 2019, and shelved when the pandemic hit. About 28,000 words. It’s a book about inclusive design as a practice: the work of building it into how teams actually operate, not just defining what it is. This is the work I’ve spent most of my career on.

I’m getting it ready for the world.

I’ve interviewed and met with hundreds of people with disabilities over the years and when we talk about how they’d like to be part of the process and included in product creation and ways to make digital products better, they very often say: “I wish someone would have just asked me earlier.” That's what this book is about, and it's even more important in this age of AI.

I’m looking for 10-15 reviewers before I figure out the right shape to ship it. If you’d like some previews to read and respond to, hit reply and tell me what draws you in. I’ll write back.