AI Guides › Judgement & Guardrails

Why "The AI Said So" Is Never A Complete Answer To Anyone

By Nigel Guy · 2 min read

Someone asks why a decision was made, or where a figure came from, or why the document says what it says, and the answer that comes back is "the AI said so." It's said as though naming the source settles the question. It doesn't, because where an idea came from and why it's correct are two separate questions, and "the AI said so" only ever answers the first one.

The mechanism: the attribution test

1. Ask yourself whether you can restate the reasoning without mentioning the tool at all. If the explanation collapses the moment you remove "the AI said," you don't have a justification — you have a source citation standing in for one.
2. Ask whether you'd still defend the conclusion if the tool were unavailable to point back to. If the honest answer is no, that's a sign the reasoning was never actually yours to defend, and it needs more work before you hand it over as settled.
3. Separate disclosure from justification when you answer someone. "I used an AI tool to draft this, and here's what I checked and why I agree with it" is a complete answer. "The AI said so" on its own is not — it's the first half of that sentence with the actual reasoning left out.
4. Apply the same test to yourself before you accept an AI-sourced answer, not just before you pass it to someone else. If you can't restate why an answer is right in your own words, you haven't actually verified it — you've just relocated the question of trust from yourself to the tool.

What to skip

Skip the two easy extremes. Don't over-credit yourself by presenting AI-assisted reasoning as entirely your own original thinking when someone asks a direct question about its origin — that's a different kind of incompleteness. And don't over-defer by using "the AI said so" as a shield against being asked to explain something you're accountable for — attribution isn't a substitute for the explanation someone is actually asking for.

Guardrails

• This is about how you explain a conclusion to someone else, not a claim about whether AI assistance is appropriate for the task in the first place — that's a separate judgement call.
• The bar for "complete answer" should scale with the stakes of the decision. A quick internal note doesn't need the same depth of justification as something that will be relied on by a client or acted on irreversibly.
• If you genuinely can't reconstruct the reasoning well enough to explain it, that's useful information about whether you should be the one presenting the conclusion at all — not a problem to paper over with a more confident-sounding attribution.

All 751 AI guides · JulieMango plans from £17/mo