The Toxic Simplicity of "Writing is Thinking"
How a seductive slogan is blocking progress on AI in education
If you ever feel the urge to say “writing is thinking”, you need to pause for a second and qualify it.
Instead, you could say:
Writing is thinking for me
Writing can be thinking
There you go. Easy.
When “writing is thinking” is unqualified, it is so simplistic that I believe it’s toxic and actively harmful to how we’re trying to address generative AI in education.
Those three simple words collapse cognition, language, authorship, and drafting into a single act and is the epicentre of the fear-based responses to generative AI in education. It leads to the belief that if generative AI touches the text, thinking must be absent, and it drives assessment policies that police writing rather than support learning.
In the writing process, there are legitimate ways to approach generative AI so that it can support cognition and expression without replacing ideation. This isn’t radical or new. We already accept this logic with translators, editors, speech-to-text, and collaborative writing. The ideas remain yours, even when the language is mediated.
Maybe for you, as for many people, writing genuinely is thinking.
And honestly: I love that for you.
But your experience is already centred, validated, and rewarded in education. Writing feels like thinking to you because the system was built around your cognitive strengths.
So when you tell me that, I really do believe you. All I’m asking is for the same grace in return.
Please believe me when I say that for a lot of people, many parts of the writing process block thinking.
As someone with a neurotype that features executive dysfunction and working memory deficits, writing was a barrier that very nearly ended my academic path 10 years ago. My ideas have always been there. I’ve always been capable of formulating and developing them. But the process of forcing them through text (starting, structuring, phrasing to meet prestige language norms) has always created unhelpful friction for me.
But now dictation and AI-assisted writing lets me think out loud and develop ideas through my preferred mode (speaking), and that barrier is lower. Not gone, the cognitive friction is still there, but it is lower and I can redirect my cognitive resources to where they are most useful – ideation and creativity. It means my thinking can finally move forward instead of continually getting stuck at the point of transcription. The same is true for many multilingual people, neurodivergent thinkers, and anyone excluded by narrow academic language norms.
Our brains are legitimate.
Our ideas are legitimate.
Our thinking is legitimate.
People might believe they are preserving standards when they insist that writing is thinking, full stop. But a lot of the time they’re gatekeeping participation. And because of that, we’re losing contributions we will never even get to hear.
Qualifiers matter.
Without them, “writing is thinking” is too blunt to survive contact with real cognitive diversity. It hardens into a toxic assumption that then shapes our response to generative AI, leading us to police writing practices rather than interrogate what we actually mean by thinking, learning, and evidence of understanding.