'Just ask ChatGPT': how AI is quietly shifting the expert's role
There’s one phrase my manager repeats constantly.
“Just ask ChatGPT.”
And it hasn’t been about IT for a long time. Not just IT. It’s about everything. Documents, copy, ads, ideas, letters, campaigns. If something’s unclear: GPT. If something isn’t selling: GPT. If you need to think something through: also GPT.
You’re sitting right there. Working. Every five minutes you hear: “put that in GPT” — “no, I’ll ask myself” — “GPT will phrase it better” — “let’s see what it says first.”
At some point you catch yourself with an uncomfortable thought. If everything goes through GPT, why do they still need someone who was hired as a specialist?
Short version: ChatGPT lowers the barrier for writing, searching and phrasing. But a good-sounding answer isn’t the same as the right decision. GPT doesn’t know your processes. GPT takes no responsibility. The specialist’s role doesn’t disappear — it shifts: you’re no longer the person with the answers. You’re the person who judges whether the answers are actually right.
GPT always gives an answer. Even when it doesn’t know.
That sounds like criticism. It’s just a fact.
GPT writes fast, searches fast, phrases things fast. For managers that looks like magic: why spend time consulting an expert when you can get a clear answer in twenty seconds in a chat?
But clear isn’t the same as right.
GPT doesn’t know your company’s weak spots. It doesn’t know what went wrong three months ago, doesn’t understand the constraints you’re working with, has no idea what’s already been tried and failed. It gives an answer regardless — that’s exactly what makes it useful, and exactly what makes it dangerous when nobody’s checking anymore.
Sometimes you’re not the expert. You’re the interface.
A lot of IT people and specialists recognise this already.
You were hired as someone with experience. But then you hear: “do it this way, GPT said so.” Or: “I preferred the GPT version.” Or simply: “can you adjust what it wrote?”
You’re still technically the expert. But in practice you’re editing a result someone else already chose.
Before, a specialist would analyse and propose a solution. Now what’s increasingly expected isn’t expertise but confirmation. Not “what’s better for the business?” but “GPT phrased it like this, can you finish it off?”
I’m exaggerating slightly. But not much.
Two weeks playing with AI doesn’t make you a specialist
Right. And yet people are starting to think it does.
GPT lowers the barrier for generating answers. That’s not the same as lowering the barrier for good decisions. Rewriting a text or developing an idea now takes five seconds. Judging whether that idea actually makes sense in your specific situation: that still takes experience.
I use GPT every day. For summarising, for searching, for quickly writing up something I already knew. But the question after is always: does this hold up? Does it fit here? Will this cause problems I already know about but GPT doesn’t?
That part GPT doesn’t do. That part I do.
And that’s the shift that’s happening. You’re no longer the person with the answers. You’re the person who filters, checks, and takes responsibility for the outcome.
Less spectacular than before. But it’s what actually matters now.
So yes: just ask ChatGPT. I do it too. But make sure there’s someone afterwards who figures out whether the answer holds up, or whether you’re just efficiently executing a nice-looking mistake.
Three questions I get asked
Should a small business use ChatGPT for IT decisions? For formulating questions, searching and summarising: yes. For technical decisions about your specific systems: not without someone who knows the context. GPT doesn’t know what went wrong at your place last week.
Is my IT specialist still relevant when AI can do so much? More than before, precisely because AI produces so much. The question is no longer who has the answer — GPT has that too. The question is who judges whether the answer applies here. That’s experience, not information.
How do I use ChatGPT sensibly without blindly trusting it? Use it for first drafts and quick summaries. Never let it have the final say on anything that affects your data, your security or your clients. And if you’re unsure: ask someone who understands the consequences, not just the answer.
Want to know how AI tools like ChatGPT can be used sensibly in a small office, and where the real risks are? I help businesses in Limburg with practical IT and judging what tools like this can and can’t do.
GPT also gives security advice. But who checks that advice? Read: ransomware at your office — what do you do when it goes wrong.