Transcript: AI

Introduction

Hi everyone. Thanks for coming. Today we’re going to be talking about the topic of AI. As always, a quick note before we get going: this call is open to everyone. You don’t need to be neurodivergent to be here. If you’re an ally, if you’re curious, or if you just wanted to listen in, you’re welcome.

The format today is the same as usual. I’ll kick us off with a short overview on the topic, around 10 minutes. After that, Naz is going to give us a quick demo of his new AI productivity game, which I’m really looking forward to. We’ll give him about 10 minutes for that, including a few questions. Then we’ll open things up for wider discussion.

One more thing before we start: you don’t have to have a “position” on AI to contribute. If you’ve never used it, that’s a valid experience. If you use it every day, great. If you have mixed feelings, even better.

Why talk about AI

AI is one of those topics that’s hard to avoid right now. It’s showing up in workplaces, in apps, in search results, in creative tools, in customer support. And the conversation around it tends to happen at a very general level: is it good or bad, will it replace jobs, is it trustworthy.

What that conversation often misses is that AI doesn’t land the same way for everyone. And for neurodivergent people specifically, there are some genuinely interesting dimensions to explore: ways it’s been more useful than expected, ways it’s fallen short, and some questions worth sitting with as it becomes more embedded in daily life.

What I mean by AI

For the purposes of today, I’m using “AI” broadly. That includes large language model chatbots like Claude or ChatGPT, AI writing assistants built into tools like email clients or document editors, transcription and summarisation tools, image generators, and AI features built into apps people already use day to day. The landscape is enormous and growing fast, so rather than cataloguing all of it, I want to focus on the kinds of experiences people are having.

Where AI helps

One of the areas where AI comes up most frequently as genuinely helpful for neurodivergent people is written communication. For people with dyslexia, ADHD, or autism, drafting emails, reports, or messages can involve a disproportionate amount of effort: not because of a lack of ideas or intelligence, but because translating thoughts into the expected format, tone, and length is its own separate task. AI writing tools can reduce that overhead significantly, acting as a kind of translation layer between how someone thinks and what a professional context expects.

Executive function is another area that comes up a lot. For people with ADHD in particular, tasks like breaking down a project into steps, deciding where to start, or maintaining momentum through a task that doesn’t provide immediate feedback can be genuinely difficult. Using an AI tool to externalise that structure, to ask “help me plan this” or “what should I do first” and receive a structured response, is something a number of people find useful as a coping strategy.

There’s also the area of summarisation and information processing. For people who find long documents, dense emails, or lengthy meetings cognitively demanding, tools that can summarise, extract key points, or answer specific questions about a body of text can reduce the load considerably.

And some people describe the interaction style of AI as unexpectedly comfortable. It’s patient, it doesn’t show frustration, it doesn’t carry social subtext, and it doesn’t require the same kind of social monitoring that human interaction does. For people who find certain kinds of social interaction effortful, there’s something in that which some people find genuinely useful, whether it’s for working through a problem out loud, drafting something difficult, or practising how to phrase something.

Where AI falls short

At the same time, AI tools come with real limitations and frustrations, and these aren’t evenly distributed.

Inconsistency is one of the most commonly cited frustrations. AI tools don’t always give the same answer twice. They can contradict themselves, produce confident-sounding responses that are wrong, and change their output based on subtle variations in how a question is phrased. For people who rely on predictability and clear rules, that inconsistency can be more than just annoying: it can make the tool feel fundamentally untrustworthy.

The interfaces themselves aren’t always accessible. Many AI tools assume a level of comfort with open-ended, free-text interaction. If you’re not sure how to phrase a request, or if you don’t know what the tool can and can’t do, getting useful results can take trial and error that not everyone has the bandwidth for.

There’s also the question of cognitive load. AI tools can generate a lot of output very quickly: long responses, multiple options, extensive suggestions. For people who already find processing large amounts of information difficult, more output isn’t always better. Knowing how to get a concise, focused response is itself a skill that takes time to develop.

And there are more structural concerns: data privacy, the tendency of AI to reflect existing biases (including about neurodivergence itself), and the risk that organisations start to expect neurodivergent employees to use AI to compensate for things that should instead be addressed through proper workplace adjustments.

AI as an accessibility tool

AI is being positioned by a lot of organisations as an accessibility tool, and there’s something real in that. But it’s worth holding that framing alongside a few questions. Who designed these tools, and who were they designed for? What happens to people who can’t or don’t want to use them? And does framing AI as an accommodation let organisations off the hook for providing the human support and structural adjustments that are actually required?

These aren’t reasons to dismiss AI as a tool. They’re reasons to think carefully about how we talk about it and what we expect from it.

Naz’s demo

Before we open up for discussion, we’ve got something a bit different today. Naz has been working on an AI productivity game and has offered to show us what it does. Naz, over to you.

Naz then gave a demo of his AI productivity game. That part isn’t in my script.

Thank you Naz, that’s so cool! Does anyone have a quick question before we move into the wider discussion?

Thank you

Thank you so much for this conversation. As always, the most valuable part of these calls is hearing directly from real people, and today was no exception.