How we mined 120,000+ research gaps from the literature
How our gap finder mines the limitations authors write about their own work, and why most 'AI gap-finders' hallucinate.
Every PhD student eventually stares at the same wall: what's actually missing in my field? The honest answer requires reading five years of abstracts across 10+ sub-disciplines, tracing citation thickets, and spotting where the network gets thin. That's a six-month task, and nobody does it properly.
Our research gap finder starts that work for you: 120,000+ open research gaps mined from 100,000+ papers in a 4.5-million-paper local library — 1,100+ of them with their own permanent, citable analysis page — and a free search that shows you a field's literature before you commit to a topic.
What "gap" means here
We use the term precisely: a gap is a question the literature itself says is still open. Authors state them all the time, in the sections most readers skip:
- Limitations — what the study could not do: the sample it lacked, the design it could not run, the population it never reached.
- Future work — what the authors say should come next, often in one concrete sentence.
- Open questions — the contradictions and unexplained results they name and leave unresolved.
The pipeline reads those sections across the library and keeps each statement together with the paper that made it. Statements that recur across independent papers are grouped into one cross-paper gap, and a gap earns its own analysis page only when at least two papers ground it.
We don't ask an LLM to invent gaps. LLMs hallucinate gaps the same way they hallucinate citations — confidently and wrong. A language model does write the human-readable synthesis of each cross-paper gap, but only from what the papers themselves said, and every page lists those papers and their own words next to it.
Why most "AI gap finders" are snake oil
If a tool answers "what gaps exist in [field]" by generating prose from nothing but the field name, it is manufacturing confident fiction. The test is simple: ask for the evidence. If the tool cannot show you the paper, and the sentence in it, that raised each gap, you are reading a guess. If every gap points back to specific papers with DOIs you can verify, it is doing the reading underneath.
Ours points back: each gap carries the papers it came from.
A taste of what it found
- Gravitational-wave physics — 23 papers from 2025–2026 converge on strong-field questions current observations leave open (the write-up).
- Biochar in agriculture — 43 studies agree biochar helps soil, and none has settled the biochar-to-vermicompost ratio (the write-up).
- Teacher professional development — 33 studies find the policy scaffolding around teachers' growth was never built (the write-up).
Browse the gaps by field at /research-gaps.
What this isn't
We won't tell you a gap is worth chasing. Novelty is necessary but not sufficient for good research. We just point at where the literature says it is unfinished — your judgement is what matters after that.
→ Browse research gaps · Match the finished paper to a journal
Tools mentioned in this post
Related posts
- When not to use AI peer review: honest limits and edge casesOur 8 AI agents are calibrated on 69,000 real peer reviews, but some papers fall outside their scope. Here is what to watch for before running a review.
- Hallucination Guardrails for an AI Reviewer: What We Do, What We Still Can't FixHow Science AI Journal's 8 AI agents guard against fabricated review claims -- and where those guards have real limits. A transparent technical account.
- How AI enforces CONSORT, STROBE, and PRISMA in peer reviewHow our methodology agent checks RCTs against CONSORT 2010, observational studies against STROBE, and systematic reviews against PRISMA, calibrated on 69,000 real peer reviews.