The Dragon with Golden Toes: Why AI Can’t Be a Shopping List for Neurodiversity

Image, Dragon Toes and Nose, by Mariann Hardey, 2025

A recent comment on my LinkedIn feed stopped me in my tracks. A superstar researcher in dyslexia, asked a question that was both practical and profound. She asked: What is one adjustment that has made an AI tool actually work for your neurodivergent students?

It is a fabulous question. It is the kind of question that comes from a place of care and a desire for solutions. Yet as I sat down to answer it, I realised I could not provide a bucket answer. There is no single app, no specific prompt, and no digital overlay that solves the equation of the neurodivergent mind. To offer a list of tools would be dishonest. It would imply that neurodiversity is a static problem waiting for a software patch. Tempting though, right?

The reality of my lived experience, and the experience of so many others, is that a one-size-fits-all approach does not just fail. It suffocates.

The Shifting Sands of Survival
Every morning I wake up and face a different internal landscape. The executive function strategies that made me a productivity machine yesterday might be the very things that paralyse me today.
Yesterday, I was a master of logistics. I managed administrative tasks with ease. The tools that helped were structural and visual. Trello was my best friend. It organised my chaos into neat, satisfying cards. It felt like a scaffold holding up a building.

Today is different. Today, I am skirting the edges of burnout. That same Trello board is no longer a scaffold. It is a place of overwhelm. The sheer volume of information on the screen is a sensory assault. Padlet is not my friend. The notifications are not helpful nudges. They are demands I cannot meet.

So I turn to pen and paper. I retreat to the tactile, slow friction of ink on a page. Later, I might take a photo of these notes and ask an AI to transcribe my handwriting into a Google Doc. But note the distinction here. The AI is not the solution. The AI is merely the janitor cleaning up after the real work was done. The solution was the permission to abandon the digital tool entirely.

The Carnival of Online Diagnosis
This brings me to my deepest fear regarding the intersection of AI and neurodiversity. I worry that the nuanced (aghhh, I am now allergic to this word as it is over-used by AI’s but here we are…), human-led pathway to diagnosis will soon be paved over by an algorithm.

If you spend any time online, you have seen them. The ads are relentless. They are predatory and harmful. They scorch the earth of genuine clinical inquiry with thirty-second clips designed to pathologise normal human behaviour. A frantic millennial points at text bubbles floating above their head. Do you doom scroll? You have ADHD. Do you find small talk exhausting? You are Autistic. Do you have a drawer full of cables you might need one day? Here is a subscription to our app.

These online tests are a farce. They are digital carnival games rigged to funnel you toward a monthly payment plan. They rely on the Barnum Effect offering statements so vague that they could apply to anyone with a pulse and a smartphone. Do you sometimes lose focus? Do you ever feel tired? (uh-oh). Of course you do. You are a human being alive in late-stage capitalism. And everything is unsettling. (Ok, I am about to nerd out about this aspect, a classic autistic trait, stay with me). 

The Barnum Effect as the Digital Clinic: To understand why the "Are You ADHD?" ads on TikTok/Insta and so on feel so uncannily accurate, and why they are so dangerous, we have to go back to the 1940s.

The Barnum Effect (also known as the Forer Effect, thank you Wikipedia) is a psychological phenomenon where individuals believe that personality descriptions apply specifically to them, even though the description is actually filled with information that applies to everyone. It is named after the showman P.T. Barnum, who famously declared that a good circus has "something for everyone."

In the classic 1948 experiment, psychologist Bertram Forer gave his students a personality test. A week later, he handed each student a unique psychological profile based on their answers. The students were amazed. They rated the accuracy of these profiles as 4.26 out of 5.
​Ooooo, science, right?

Every single student had received the exact same text, which Forer had copied from a newsstand astrology book. It contained statements like:
“You have a tendency to be critical of yourself.”
“At times you are extroverted, affable, sociable, while at other times you are introverted, wary, reserved.”
“You have a great deal of unused capacity which you have not turned to your advantage.”

These are Barnum Statements. They work because they are high-frequency, low-stakes generalisations. They rely on subjective validation: our brain's desire to find connections between generic information and our own lives.

The Weaponisation of the Barnum Effect
In the 20th century, the Barnum Effect was mostly used for harmless vanity. Horoscopes and Myers-Briggs tests used "flattery" to keep us hooked. They told us we were "critical thinkers" or "misunderstood geniuses." They played on the Pollyanna Principle, where we are more likely to accept positive feedback than negative feedback.

But the algorithm has mutated this effect into something far more sinister.

We are now witnessing a Medicalised Barnum Effect. The modern algorithmic ad does not try to flatter you. It tries to pathologise you. Instead of telling you that you are "disciplined but insecure" (a classic Forer statement), the modern Instagram ad asks:

“Do you have a drawer full of cables you might need one day?”
“Do you hate small talk?”
“Do you doom scroll at night because you didn’t feel productive during the day?”
“Are you a woman who is spacey? Forgetful? Or chatty?” (basically a person with a personality)

These are the new Barnum Statements. They take universal human experiences, boredom, clutter, procrastination, social fatigue, and reframe them as symptoms.

These diagnostic algorithms are worse than a polygraph test on The Secret Lives of Mormon Wives. At least reality TV admits it is spectacle. At least when the wires are hooked up on screen, we know it is for the drama. These online tools masquerade as medicine. They wear the lab coat of authority but underneath is nothing but a data-harvesting engine. They reduce the complex, lifelong architecture of a neurodivergent brain into a binary output. Pass or Fail. Subscribe or Leave.

Real diagnosis is an archaeology of the self. It requires digging through layers of masking, trauma, and learned behaviours. It requires a human witness who can see the difference between anxiety and autism, or between trauma and ADHD. An algorithm cannot see the history in your eyes. It can only calculate your click-through rate. My fear is that future generations, including students I teach, will be handed a QR code instead of a conversation.

The Threat of Flat Stanley
We are promised a future of seamless voice interactions with AI. I assume this will function much like the speech-to-text apps I currently use, which have been lifesavers at times. However, there is a cost to this convenience that we rarely discuss.

When I speak to an AI, I am feeding the machine. My data, my cadence, and my real voice are harvested to train a model that prioritises averages and norms. As a neurodivergent woman, I fear what happens when my distinct creativity is processed through these algorithms. Will (yep, goes my brain) my thoughts be flattened out like Flat Stanley? Will the jagged, interesting edges of my thinking be sanded down to fit a generic model of "professional communication"?

Side note, ask an AI to compose an ‘out of office for a university professor’, and the default pronoun will be ‘He’. Isn’t that something. 

No, thank you. I want to keep my womanly dimensions.

Diagnoses and Dragons
This year has been a watershed moment for my daughter and I. We both received new diagnoses. I am dyslexic, and I have now been diagnosed as autistic. My daughter has started her own journey during her school years.

I look at her and I see the difference in our paths. I am an adult who survived my entire education without support. I built my coping strategies out of necessity and instinct. My daughter has only been in school since 2019. She has not yet modelled these extensive defences. What she does have is a strong, innate sense of self. She knows what she likes. She knows what causes her to feel "blurgh." She knows what is fun. She is fun. 

It would be horrific if a diagnosis report simply stated: Have D use an AI.

Why? For what purpose?

If we simply shovel AI tools at her, we are bypassing the human work of understanding how she learns. We are replacing a helping hand with a predictive text generator. My fear is that we are heading toward a future in Education where human support is considered out of reach. By the time my daughter reaches university, I imagine professional support services will be rarer than a dragon with golden toes. I fear students will be handed a generic "AI Toolkit" and told to get on with it. This is already happening, btw. 

The Philosophy of the Ape and the Wolf
Mark Rowlands (I'm reading a lot of his work lately) reminds us that there is a difference between the instrumental value of the "ape”, who schemes and plans for a future result, and the intrinsic value of the "wolf”, who lives entirely in the moment of being.

AI is the ultimate tool of the ape. It is obsessed with efficiency, output, and results. It tries to civilise the wildness of our thoughts. But the neurodivergent mind often has more of the wolf in it. It does not always want to be efficient. It wants to wander. It wants to make connections that an algorithm would label as errors.

I can answer the original question without a single piece of software. What makes the work possible for me? What would make it possible for my daughter?

Flexibility and time.

That is it.

We need the flexibility to use Trello on Tuesday and burnout on Wednesday. We need the time to process the world without a predictive engine rushing us to the end of the sentence. If you apply flexibility and time to neurodiversity, you will be surprised by what happens. We have sophisticated, instinctive strategies that bloom when we are not being forced into a standardised shape. This is not a superpower btw. 

The answer is not in the code. It is in the space we leave for the human.