Spelling And AI

Your spelling checker never had to pass an ethics committee. Why does AI? Nobody has ever asked me to disclose that Word corrected my typos.

No patient has demanded a doctor to know whether a calculator was involved in their dosage. No board has requested a governance framework for the pocket calculator. No journal has asked authors to declare that they used a search engine.

Yet the moment a clinician lets an AI model draft a discharge letter, we convene a working group. I find that fascinating. And slightly hypocritical.

Here is my position, and I know it will annoy some of you: AI is a tool. Nothing more, nothing less. It is a very capable tool, sometimes unpredictable, but a tool. Just like the spelling checker you used to write your last "AI is dangerous" comment.

The double standard

We happily accepted that software could rewrite our sentences, reorder our search results, flag our fraud, route our ambulances and calculate our creatinine clearance. None of those got a dedicated conference. They just quietly became part of the work.

AI gets scrutinized not because it is fundamentally different, but because it is fundamentally visible. It talks back. It writes in full sentences. It looks like it is thinking. And that triggers something in us that a red squiggly line never did.

So we apply a standard to AI that we never applied to any other tool in the building. Ask yourself honestly: is that standard about safety, or about discomfort?

But it hallucinates

Yes. So does your spelling checker. It has confidently changed "hypokalemia" into something unrecognizable more than once, and it never apologized. Your autocomplete has sent emails you did not intend. Your reference manager has cited the wrong paper.

The difference is not that AI makes errors. The difference is that we stopped checking the older tools, because they became invisible. We should probably be more worried about that than about the new one we are all watching like hawks.

The real question

If AI is a tool, then the accountability question is simple and already answered: the professional who uses the tool is responsible for the output. Same as the surgeon holding the scalpel, same as the doctor who signed the letter Word helped spell correctly.

That is not a lower bar. It is the same bar. And it is a bar we already know how to hold people to.

What worries me is that all this scrutiny is becoming an excuse. "We are still assessing the governance implications" is a very comfortable sentence to hide behind while the nurse next door is drowning in documentation that a tool could have handled yesterday.

So, convince me otherwise

I know plenty of people believe AI is categorically different. Infrastructure. Hence this is my position, as I wrote in my latest essay "AI from tool to infrastructure". Down below, in making content- for pictures, texts, even videos I approach this more as tools for creativity. Some of you have heard me flirt with that framing myself. The use for creative is something different than how the underlying infrastructure of society is changing due to the capabilities of AI.

So here is the debate I actually want: If AI is more than a tool, what exactly is the property that makes it more? And why did that property not apply to the spelling checker, the search engine, or the clinical decision support system your hospital installed in 2014 without a single town hall?

Or is it just that this one talks?

Published earlier on https://www.linkedin.com/pulse/your-spelling-checker-never-had-pass-ethics-committee-lucien-engelen-mxabe/?trackingId=mlIljSQxQAGtp3CBMG9vDQ%3D%3D