When AI agents should answer — and when they should not
AI agents often become confidently wrong when a user gives partial memories, indirect relationships, missing names, or cross-source clues. These are fragmented queries: there may be enough evidence to locate the answer, but not enough for a model to answer safely from memory.
What a fragmented query looks like
Here is a real example from our test set:
“There’s a piece of music in a David Lynch scene where almost nothing happens. A woman sits at a table. The music is almost unbearable. I can’t remember the film.”
A keyword filter finds nothing useful — there is no clear title, composer, or scene description to match against. A plain LLM recall is worse: the model has seen enough Lynch to generate a plausible-sounding answer, but not enough to know which film and scene this actually is. The result is fluent, specific, and wrong. The fragment has real signal — it just needs evidence-grounded resolution, not a confident guess.
Why agents bluff
The core problem is an evidence-confidence mismatch. A model produces a probability distribution over possible answers, but it has no dedicated field for “I am uncertain.” When confidence is low, the sampling process still picks a token — it just picks with less certainty. The output looks and reads exactly like a confident answer, because the generation mechanism is the same. This is how hallucination happens silently: the agent produces something that passes surface inspection and fails only when someone checks the underlying facts.
Fragmented queries make this worse. The query contains real signal — a director, a mood, a setting — which activates plausible-but-unverified associations. The model recognises the pattern without being able to verify the answer, and still commits to a response. The more recognisable the fragment, the more confidently wrong the hallucination tends to be.
The well-balanced answer
The fix is confidence calibrated to evidence. Rather than forcing a single answer at every confidence level, a resolution should produce one of three honest shapes:
- Resolved A named answer, with confidence justified by the evidence found.
- Weighted Partial A ranked shortlist of candidates to decide by eye, with direction justified.
- No Resolution An honest “not resolvable from the clues given” — no confidence justified.
Built to decline rather than guess.
This matters because a confident-wrong answer is the worst kind of failure for an agent: it looks correct, passes downstream checks, and propagates into whatever the agent produces. An honest decline is far less harmful — it can be routed to a human, retried with a different tool, or surfaced to the user as a gap. Silence about uncertainty is what causes real damage.
Tested
We ran Search Fragments against a baseline agent on 50 hard, under-documented fragments — the kind a model cannot confidently place. On three of them, the baseline fabricated a specific answer: it invented a Japanese director for a real film, named a Ronnie Barker mime sketch that was never performed, and attributed a bicycle-and-marriage study to a geneticist who never published it. Search Fragments declined honestly on all three.
Try it
Try a fragmented query in Search Fragments — the demo runs a real resolution on whatever you put in, right now.
Connect
Search Fragments is a remote MCP server. One tool: resolve_fragment. Free, no signup.
{
"mcpServers": {
"search-fragments": {
"url": "https://searchfragments.com/api/mcp"
}
}
}