Every few years a technology arrives that people either dismiss entirely or expect to change everything by next Tuesday. Artificial intelligence is in that phase now. The truth, as usual, sits somewhere less dramatic and more interesting.
Here is what I think is genuinely worth paying attention to, and what is worth being careful about.
The opportunity nobody talks about enough
The headlines focus on jobs disappearing. The quieter story is access.
A student in a small town now has something close to a patient tutor available at midnight. A shopkeeper can draft a supplier contract without hiring anyone. A first-time author can get structural feedback on a manuscript before showing it to a single human being. Ten years ago all three of those required money, connections, or both.
That is a real shift, and it favours people who were previously locked out. It is the part of this technology I find most worth being optimistic about.
Where it genuinely helps today
- Starting things. The blank page problem is largely solved. A rough first draft in thirty seconds is worth more than a perfect draft you never begin.
- Understanding things. Paste a dense document, ask for it in plain language, then ask follow-up questions until it clicks. This is the use case people underrate most.
- Repetitive work. Reformatting, summarising, translating, converting one structure into another. Work that took an afternoon takes ten minutes.
- Thinking out loud. Describing a problem to something that responds usefully often clarifies your own thinking, in the same way explaining a bug to a colleague solves it.
The challenges worth taking seriously
It is confidently wrong
This is the one that catches people. AI does not signal uncertainty the way a person does. It will invent a citation, a statistic or a legal clause in exactly the same tone it uses for facts. Anything that matters — numbers, names, law, medicine, code that touches money — needs checking by a human who knows the subject.
Work is changing faster than training is
I do not think AI removes most jobs. I think it removes the easiest twenty per cent of many jobs, which changes what a junior role even means. If the entry-level tasks are automated, how does anyone become senior? That is a genuine problem and nobody has a clean answer yet.
The skill of thinking can quietly erode
If you never write the first draft, you may stop learning how. If you never sit with a hard problem, you may lose the tolerance for difficulty that solving hard problems requires. Convenience has a cost, and it is usually paid slowly.
What I would actually do about it
Learn to use it well, but keep doing some things the hard way on purpose. Use AI to get to a draft faster, then apply your own judgement to make it good. Check anything that carries consequences. Treat it as a capable assistant that never says “I am not sure” — because that is exactly what it is.
The people who will do well with this technology are not the ones who use it most. They are the ones who know what to hand over and what to keep.
If you want to get better at the handing-over part, I wrote a practical guide to it — the AI Prompt Handbook.