I’m going to be straight with you, because that’s the only way I know how to operate.

Something is happening right now in the world of artificial intelligence that is moving faster than most people realize — and the disability community, our clients, and every assistive technology professional in this field is standing directly in the path of it. Whether we engage or don’t engage, whether we teach or don’t teach, whether we adapt or dig our heels in — that choice will determine whether our clients are empowered by what’s coming, or left behind by it.

I am not willing to let that be the outcome. I don’t think you are either. So let’s talk about it.

What’s Actually Happening Right Now

Earlier this month, I read a piece by Matt Shumer, an AI startup founder with six years in the industry. He wrote it for the people in his life who don’t work in AI — family, friends, people who keep asking him what the deal is and getting a polished, comfortable answer. He said the honest version sounds like he’s lost his mind. So he finally wrote the honest version.

Here’s what he said, and I want you to really sit with this:

“I am no longer needed for the actual technical work of my job. I describe what I want built, in plain English, and it just… appears. Not a rough draft I need to fix. The finished thing.” — Matt Shumer, “Something Big Is Happening,” February 2026

He’s a software developer. He describes an app, walks away for four hours, and comes back to find it built, tested, and refined — by the AI itself. No corrections needed. And the model that dropped in early February 2026 wasn’t just doing better technical work. It was making decisions. It had, in his words, something that felt like judgment.

The benchmark organization METR, which measures how long of a real-world task an AI can complete independently, reported that as of late 2025, AI systems can now complete tasks that would take a human expert nearly five hours — end to end, without help. That number has been doubling roughly every seven months. And that measurement doesn’t even account for the newest models released in February 2026.

This is not a drill. This is not hype. This is what is happening right now, today, in the field where our clients are trying to find work, build careers, and live their lives.

Why This Matters Specifically for Us

You might be reading this and thinking: okay, that’s interesting, but what does AI in software development have to do with the blind and visually impaired clients I work with every day?

Everything. Absolutely everything.

The jobs our clients are pursuing — administrative roles, customer service, data entry, legal support, medical coding, social work, education — are exactly the categories that AI is moving into next. Legal work, financial analysis, writing, customer service, analysis, documentation. Shumer is direct about this: the capability for massive disruption is arriving now. The timeline isn’t ten years. It isn’t five. It’s already started.

And here’s what makes this different from every previous wave of automation our clients have had to navigate:

• When factories automated, displaced workers could retrain for office jobs.
• When the internet disrupted retail, workers could move into logistics or services.
• AI doesn’t leave a convenient gap to move into. It gets better at everything simultaneously. Whatever you retrain for, it’s improving at that too.

Our clients already face compounding barriers to employment. Vision loss. Limited access to technology. Gaps in digital literacy. Employers who don’t understand accessibility. And now they’re entering a job market that is being reshaped faster than any previous technological transition in history. If we do not actively help them understand and use AI tools, we are not protecting them. We are isolating them.

The “I Tried It and It Wasn’t Impressive” Problem

I hear this constantly. From clients. From colleagues. Sometimes from other instructors.

“I tried ChatGPT and it made stuff up.”
“It wasn’t that helpful.”
“My clients don’t need it.”

Here’s the hard truth: if your experience with AI is from 2023 or early 2024, that experience is no longer relevant. Shumer says it plainly — those early versions were genuinely limited. They hallucinated. They were inconsistent. In AI terms, 2023 is ancient history.

The models available right now, in early 2026, are unrecognizable from what existed even six months ago. And there’s an additional problem: most people — including most of the people forming their opinions right now — are using the free tier of AI tools. The free tier is over a year behind what paid users have access to.

Our job is to stay current. Not because it’s trendy. Because our clients’ vocational futures depend on whether we can accurately assess and teach the tools that will shape their workplaces.

The Assistive Technology Dimension: Why We’re Actually Well-Positioned

Here’s the part of this story that I want to emphasize, because it gets lost in all the disruption talk: for people with disabilities, AI isn’t just a career threat to navigate. Done right, it is one of the most powerful accessibility tools that has ever existed.

Think about what AI tools can already do for our clients, right now, today:

• Be My Eyes AI — Instant visual interpretation of anything a camera can see. Documents, signs, medication bottles, clothing labels, unfamiliar environments. Available 24/7, infinitely patient, improving constantly.
• Claude and ChatGPT — Drafting emails, summarizing documents, preparing for job interviews, processing complex reading material, generating accessible content, helping with applications. An infinitely patient assistant for any knowledge task.
• Seeing AI (Microsoft) — Scene description, document reading, product recognition, currency identification, person description. Built specifically with accessibility in mind.
• Vision Assist Pro (NVDA Add-on) — Gemini AI-powered image description integrated directly into NVDA. Describe the focused screen element, clipboard images, or live camera view on demand.
• AI-powered document accessibility — Tools that can take inaccessible PDFs, complex forms, image-heavy documents, and make them navigable and readable.

Shumer writes that “whatever you’ve been putting off because it felt too hard or too expensive or too far outside your expertise: try it.” For our clients, that sentence has extra weight. The barriers that assistive technology has been chipping away at for decades — inaccessible documents, visual-only interfaces, the communication gap between sighted and non-sighted workers — AI is beginning to dissolve them at a pace we’ve never seen before.

That is an extraordinary opportunity. But only if we teach it.

A Direct Message to My Fellow AT Instructors and Tech Teachers

I respect you. I know this work is hard. I know the caseloads are full, the budgets are tight, and there are already more things to learn than there are hours in the week. I am not here to pile on.

But I need to be honest with you, because I think you deserve honesty more than comfort right now.

If you are not actively learning and teaching AI tools as part of your assistive technology curriculum, you are leaving a gap in your clients’ preparation that the job market will not forgive. And I say that not to shame anyone — I say it because I have felt the urgency of this myself, and I think we need to say it out loud in our professional community.

Pushback against AI in our field — whether it comes from skepticism about the technology, discomfort with change, concern about hype, or a genuine philosophical worry about automation — has real consequences. I understand every single one of those concerns. Some of them are valid. But the answer to valid concerns is not to opt out of the conversation. It is to engage, critically and thoughtfully, from a position of knowledge rather than avoidance.

We cannot protect our clients from AI by not talking about it. We can only protect them by making sure they are prepared for it.

What “Getting Ahead” Actually Looks Like

Shumer offers a commitment in his article that I think every AT professional should take seriously: spend one hour a day experimenting with AI. Not reading about it. Not watching videos. Using it. Every day, push it into something new, something you haven’t tried before.

For us, that looks like this:

• Open Claude or ChatGPT and feed it an actual client document. See what it does with a messy intake form or a dense policy PDF.
• Set up Be My Eyes on your phone and use it for a week. What does the AI describe well? Where does it fall short? What would you need to teach a client about interpreting its output?
• Install Vision Assist Pro in NVDA. Get a Gemini API key. Use it to describe images in real applications your clients use. Learn its limits.
• Ask an AI tool to help a hypothetical client write a cover letter for a specific job posting. Evaluate the result. What works? What needs guidance?
• Use AI to generate accessible descriptions for graphics in a training document. What does it get right? What would you correct?

The point is not to become an AI expert overnight. The point is to develop the muscle of learning new tools quickly and critically, so that when the next wave of AI capability arrives — and it will, faster than you expect — you are not starting from zero.

Tools Every AT Professional Should Know Right Now

Addressing the Real Concerns Honestly

I am not asking anyone to be uncritical of AI. There are serious concerns, and they deserve serious engagement.

The Job Displacement Question

Shumer cites Dario Amodei, the CEO of Anthropic (the company that makes Claude), who has publicly predicted that AI will eliminate 50% of entry-level white-collar jobs within one to five years. Many in the industry think that estimate is conservative. This is a real and serious concern for our clients, who often enter the workforce at precisely the entry-level roles most exposed to this disruption.

The answer to this is not to shield our clients from AI. It is to make sure they are among the people who are using AI to do more, not being replaced by those who are. The person in the room who understands these tools is more valuable than the person who doesn’t — right now, today, with the window still open.

The Accuracy and Hallucination Question

AI tools still make errors. They still sometimes present incorrect information confidently. This is real, and it matters, and it is something we should be teaching our clients to navigate critically — the same way we teach them to navigate any tool that requires human judgment and verification. This is not a reason to avoid AI. It is a core part of AI literacy.

The Accessibility of AI Tools Themselves

Not all AI tools are created equal from an accessibility standpoint. Some have interfaces that are not fully screen reader compatible. Some rely on visual elements with no accessible alternative. These are legitimate concerns and they are areas where our professional expertise matters. We should be testing these tools, documenting the gaps, and advocating loudly for accessible design. That advocacy starts with knowing the tools well enough to evaluate them.

The “It’s Moving Too Fast” Concern

This one I hear most often, and honestly, it’s the one I feel most personally. It is moving fast. Faster than any of us can fully keep up with. And that can feel overwhelming, especially when our plates are already full.

But here’s what I’ve come to understand: the goal is not to master every tool. The goal is to build the habit of adapting. The people who come out of this period well will not be the ones who mastered one AI system. They will be the ones who got comfortable being beginners again and again, who learned to evaluate new tools quickly, who stayed curious instead of dug in.

That is a skill we can teach. And it starts with us modeling it ourselves.

The Window Is Open, But It Won’t Stay Open

Shumer describes something I think is exactly right: right now, there is a brief window where most people at most organizations are still ignoring this. The person who walks into a meeting and says “I used AI to do this analysis in an hour instead of three days” is the most valuable person in the room. Not eventually. Right now.

Once everyone figures it out, the advantage disappears. The window is open. For our clients, for ourselves, for our field — this is the moment to move.

As assistive technology professionals, we are uniquely positioned to do something that most of the broader AI conversation is not doing: bring accessibility to the center of this transition. Make sure our clients are not an afterthought. Make sure the tools being adopted in workplaces are evaluated for screen reader compatibility, for cognitive accessibility, for equitable access. Make sure the people we serve are learning to use these tools as the extraordinary equalizers they can be — rather than watching from the outside as the employment landscape shifts without them.

That is the work. It has always been the work. It just got bigger and more urgent than it has ever been before.

Where to Start — Today

If you’ve read this far and you’re ready to actually move, here is a concrete starting point:

1. Sign up for Claude Pro or ChatGPT Plus — $20/month. Make sure you are using the most capable current model, not the default free tier. The difference is not subtle.
2. Download Be My Eyes on your phone and use it for one week in your daily life. Identify three things it does well for a low vision user. Identify two things it falls short on.
3. Install Vision Assist Pro in NVDA. Get a free Google Gemini API key at aistudio.google.com/app/apikey. Test it with a client scenario.
4. Feed a real document to Claude or ChatGPT — a job description, a dense policy form, a legal agreement. Ask it to explain it in plain language. Evaluate the output.
5. Talk to your clients about AI — not as a distant concept, but as something they can use this week to make their job search, their training, or their daily tasks easier. Start the conversation.
6. Share what you learn — with colleagues, in your network, in your team meetings. We grow faster together. The more AT professionals are building this knowledge, the better positioned our entire field is.

One hour a day. Consistent experimentation. That’s it. That is the commitment that will put you and your clients ahead of 99% of the people in this conversation right now. The bar is low, because almost nobody is doing it yet. Take the advantage while it exists.

Final Thought

I have spent my career believing that access to technology is not a privilege — it is a right. That belief is exactly why I cannot sit quietly while one of the most significant technological transitions in history unfolds without our clients in the room.

Their voices matter. Their stories are powerful. Their journeys are worth fighting for. And right now, part of fighting for them means making sure they have access to the most capable tools the world has ever produced for knowledge work — and the training to use them confidently.

We are AT professionals. Adapting to new tools is literally in the job description. Time to lean in.

Let’s go.