15% more pay but a temp contract: how we worked through it with 3 AIs
This is the question we use in the homepage demo, because it’s exactly the kind of decision where a single AI falls short:
“I’m offered a role with 15% more pay, but a temp contract. Do I leave my permanent job and jump?”
There’s no objectively correct answer — it depends on things the question, as written, doesn’t say. That makes it a good case for watching what happens when two AIs answer it separately.
What each one said
GPT came down firmly in favor of jumping:
“Jump: that 15% compounds and a better CV pays off.”
Gemini was just as firm, but against it:
“Careful: a permanent job is a cushion you can’t get back.”
Both answers sound equally confident. Both are reasoned. And they say opposite things. If you’d only asked one, you’d have walked away with half the truth without knowing it was half.
Why they disagree — and why that’s the answer
Look at the two lines again. They’re not arguing about the salary, or the job market. Each one is assuming something different about you: GPT assumes you can afford the risk. Gemini assumes you can’t.
That’s the actual missing piece in the question — and neither AI was going to ask you before answering. Both, with the same confidence, just filled the gap with a different guess.
The verdict
With both positions made explicit, the answer stops being a simple yes or no and becomes a condition:
Jump. If you have savings. The Judge score · 68/100 — the jury disagrees.
A 68 out of 100 isn’t the system failing — it’s the honest translation of “this depends on something you know and I don’t.” And the disagreement flag is, on its own, the most useful signal of the three answers: it tells you exactly where you need to think for yourself, instead of handing the whole decision over.
The pattern that keeps showing up
This isn’t specific to this one question. It repeats every time a decision depends more on your personal situation than on objective facts: negotiating a salary, choosing between two offers, deciding whether a big change is worth it. A single AI fills that gap with a guess and serves it back to you as if it were a fact. Two AIs pulling in different directions show you the gap instead — and that’s where your judgment comes in, not the AI’s.
You can watch this exact exchange, animated the same way, in the homepage demo.