AI assistance leaves timing patterns that reveal overreliance in writing and coding
Does AI Assistance Leave a Temporal Fingerprint? Detecting Overreliance in AI-Assisted Writing and Programming
Human-Computer InteractionArtificial Intelligence
Summary
It can be hard to tell if people used AI tools too much just by looking at their finished writing or code. This paper finds that the timing of how people type or accept AI help leaves a kind of fingerprint that can show if AI did most of the work. The researchers studied writing and programming and showed that bursts of AI-generated content happen in clear time patterns. Their method can detect when someone mostly lets AI do their work, which could help keep things fair in schools and other places.
What this means in practice
- •For academic integrity teams: Check student work process data for timing patterns indicating if AI did most of the writing or coding, helping to find unauthorized AI overuse.
- •For software development tool makers: Design programming editors that track timing of code acceptances to detect when developers rely heavily on AI-generated code suggestions versus their own input.
Authors
Eduardo Davalos, Yike Zhang
Abstract
The rapid adoption of generative AI has made final artifacts unreliable evidence of student learning, and AI detectors that examine only the finished product are inaccurate and ethically contentious. Process data offers an alternative, but prior work covers only English essay writing. We ask whether AI assistance carries a temporal signature, whether it generalizes from writing to programming, and whether it distinguishes ordinary collaboration from wholesale delegation. We analyze three public corpora: CoAuthor (1,447 keystroke-level co-writing sessions), RealHumanEval (editor telemetry from 243 programmer records), and a pre-LLM CS1 corpus (5.1 million keystrokes) as a human-only baseline, comparing minimal-AI work, collaborative AI use, and simulated wholesale delegation. Three findings emerge. First, the signature generalizes: AI contributions arrive in bursts far outside the author's own baseline in both mediums (paired d_z = 1.13 and 3.54). Second, engagement diverges by medium: 93% of AI-inserted characters survived to writers' final documents, while only 14% of accepted code suggestions survived intact. Third, classifiers using only observable temporal features separate simulated delegation from authentic work nearly perfectly (F1 $\geq$ 0.997; at most 0.5% of real work misclassified), while ordinary collaboration remains hard to distinguish from unassisted work. Temporal evidence flags wholesale delegation rather than assistance, positioning process visibility as a candidate evidentiary basis for academic integrity, pending validation in authentic coursework.