What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
“What if there was an AI text detector that could recognize that you wrote something based on an idea that AI gave you—even if all the actual writing was your own original work?” That is the distinction behind IdeaLens, a research system that estimates where a document’s ideas came from rather than who wrote its sentences. Its output is an estimate of provenance, not proof of a person’s writing process or misconduct.
What does it mean to detect AI ideas rather than AI text?
Prose provenance asks who produced the wording. Idea provenance asks who supplied the underlying concepts or plan. Those origins can differ: a person may write entirely original prose from an AI-generated outline, while an AI model may turn a human’s detailed plan into polished text.
As an Amazon Associate I earn from qualifying purchases.
Conventional prose detectors inspect the words and their statistical patterns. IdeaLens instead tries to classify the conceptual structure represented in a document. The distinction matters for policies that care about the source of ideas, but a classifier’s estimate cannot establish exactly how a particular idea arose.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →How IdeaLens works
Rajendhran and seven coauthors describe IdeaLens in a preprint submitted to arXiv on October 5, 2026. The system converts a document into an outline: each entry pairs a discourse role with a short, paraphrased description of the content. The aim is to retain the document’s conceptual plan while reducing overlap with its original wording.
#1 Best Overall
In the authors’ formulation, this representation helps focus classification on ideas rather than prose. IdeaLens then classifies the outline. This is an indirect approach: it does not observe a writer brainstorming, consulting a model, or drafting; it infers provenance from the resulting document representation.
Training labels are estimates, not direct records
The authors trained IdeaLens on one million FineWeb documents using “silver” labels produced by Pangram, a prose-provenance detector. These labels are proxy signals, not a direct record of the origin of every document’s ideas. That distinction limits what the model’s training and evaluation can establish about real-world idea ownership.
Rank #2
What the reported tests found
The paper reports several evaluations. The results below are the authors’ findings on their study datasets and benchmark conditions, not independent replications.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minute- When AI wrote from increasingly detailed human plans: IdeaLens’s AI flag rate fell from 95% to 7%; Pangram 4 still flagged 92% in the reported comparison.
- When AI-generated plans informed the writing: IdeaLens’s AI flag rate remained above 96% in the study’s controlled comparison.
- Human-written stories from AI-generated plans: Among 50 such stories, IdeaLens flagged 68% as AI, compared with 8% for Pangram 4.
- Broader evaluations: The paper reports testing across 19 existing detection benchmarks and describes evaluations across domains, formats, and languages. It also says the authors released models and labeled datasets for future research.
The small 50-story test illustrates both the promise and the interpretive challenge: IdeaLens was more likely than Pangram 4 to flag human-written text when the plan came from AI, but a flag does not tell a reader what the writer did, intended, or understood.
A separate account by Martin Anderson in Unite.AI reports 95.3% accuracy for shared idea/prose provenance and 81.3% for mixed provenance across the benchmark suite it describes, and discusses a 24-language test. Those figures are attributed to that article’s account; the arXiv abstract confirms broad benchmark and cross-language evaluation but does not enumerate those specific figures.
How to interpret an IdeaLens result
An AI-idea flag should be read as a model classification under particular data and evaluation conditions, not as a forensic record. The system estimates whether the ideas appear AI-originated; it does not show who first conceived a thought, whether a writer used AI, or whether a use violated a policy.
- Keep the question precise: A prose detector and an idea-provenance model classify different things. A low prose-detector score does not rule out AI-suggested ideas, and an idea-provenance flag does not establish AI-written wording.
- Check the evaluation setup: Results can depend on whether idea and prose origins match or are mixed, which benchmark and threshold are used, and the domains, formats, and languages tested. Headline figures from different datasets are not directly comparable.
- Account for proxy labels: Since IdeaLens’s training labels came from Pangram rather than direct documentation of idea origin, its findings should not be treated as ground truth about every text.
- Do not infer intent from a score: The model estimates provenance from an outline; it does not observe the process by which a person developed or used an idea.
What the study establishes—and what it does not
The preprint presents a method designed to separate conceptual structure from surface wording, and reports stronger flagging than Pangram 4 on a small set of human-written stories based on AI-generated plans. It also reports broader benchmark and cross-language evaluations. The results are evidence about the tested datasets and conditions, not a guarantee of reliable attribution for arbitrary writing.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe paper’s linked demo, code, and model/data resources are not independently validated here; their privacy terms and commercial status are not established by the cited sources. Treat any result from a tool built on this approach with the same care: ask what it classifies, how it was evaluated, and whether the provenance labels were directly observed or inferred.
Quick Recap
Sources
- Rajendhran et al., “IdeaLens: Detecting AI Ideas in Long-form Writing,” arXiv, submitted October 5, 2026
- Martin Anderson, “Detecting AI Ideas, Not AI Text,” Unite.AI, October 6, 2026
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
