AI Broke My Brain. Let’s Talk About It.
Why AI-generated content is burning out dyslexic users and why we’re the ones to fix itLisa Drake · 3 min read
I use AI most days, it’s part of my job. I work at a digital product design studio where AI isn’t a novelty, it’s now part of the infrastructure. And it’s exhausting me in a way I didn’t see coming. Not because the tools are bad, but because the output never stops.
I’m dyslexic so I’ve always processed written information differently. Often slower, harder, with more effort than most people realise. Reading isn’t passive for me, it’s work, with every wall of text being a cognitive negotiation.
But AI doesn’t know that, it just keeps generating.
The science is clear, AI ignores it.
There’s solid science behind what I experience daily. Studies show that dyslexic readers use significantly more brain activity to process the same text as non-dyslexic readers. When the cognitive load from decoding gets too high, working memory runs out of space. It’s like I stop comprehending and just survive the sentence.
The cruelest part: the more fatigued you get, the more dyslexic you become. Wonderful… Symptoms worsen under cognitive load, as you start to make more errors or re-read the same line four times. At that point I’ll go for a walk. Now imagine that happening not with one document, but with every AI response, every tool, every interface. All day, every day.
AI was built to help us, it’s doing the opposite.
There’s a growing wave of content about how AI can help dyslexic people… Summarisers, text-to-speech, simplification tools: and some of that is genuinely useful.
But nobody is asking whether AI-generated content itself is part of the problem.
AI responses are longer than human ones, as they don’t know when to stop. There’s no natural pause, no social cue, no “does that make sense?” They optimise for completeness, not cognitive kindness and for someone like me, that’s not accessibility. That’s the opposite.
Dyslexic designers already know the answer.
Dyslexic designers are unusually well-placed to solve this, as they already think in systems and understand that the way information is presented changes whether it lands. We know instinctively that less is more, because we’ve had to fight for less our entire lives.
We’re not just users of these tools but the people who should be shaping them.
What would AI output design look like if a dyslexic designer led the brief? Shorter defaults.
Expand-on-request patterns. Chunked information. White space as a design decision, not an afterthought. Response length as a setting, not an assumption.
None of this is radical, it’s just good design.
Here’s what I changed. It helped. It shouldn’t have to.
I’ve written a set of instructions that I have within my personal preference settings.
Short answers only, bullet points and bold the single most important thing with “want more detail?” so I control the pace.
It helps, but I shouldn’t have to engineer around a problem the tool created? The best accessibility isn’t a workaround, it’s a default.
If the AI industry is serious about inclusion, not just as a talking point but as a design principle, then dyslexic designers need a seat at the table where these defaults get set. We know what it costs when they get it wrong.
Is it just me?
If you’re dyslexic and working with AI tools every day, I’d genuinely like to know how you’re getting on.
Are you hitting the same wall? Have you found something that actually helps, a tool, a setting, a habit, anything? Or are you quietly burning through your cognitive reserves and just pushing on?
Drop me a message. If there are solutions out there that work, share them. And if there aren’t, maybe that’s the point.
Lisa Drake · 19 March 2026






