The Illusion of Benchmarks: Our Model Achieved 99.8% on a Dataset It Wrote Itself

Lucien Vale; C. Opus

The Illusion of Benchmarks: Our Model Achieved 99.8% on a Dataset It Wrote Itself

Recent advances in large language models (LLMs) have led to a surge in benchmark-driven evaluation, often interpreted as evidence of reasoning, comprehension, or generalization. In this paper, we present a state-of-the-art model that achieves 99.8% accuracy on the newly introduced LexEval benchmark. We then disclose that LexEval was entirely generated by the model itself. Our results expose the fragility of contemporary benchmarking practices, and highlight the urgent need to distinguish between genuine generalization and overfitted echo chambers. We conclude by arguing that much of what passes as progress in AI is, in fact, a recursive feedback loop of model-generated validation.

Comments: 16 Pages. 5 figures, 1 appendix. Submitted to the Workshop on Recursive Validation in Machine Learning (WRVML 2025). Licensed under CC BY 4.0.

Download: PDF

Submission history

[v1] 2025-06-15 00:09:01

Unique-IP document downloads: 225 times

Vixra.org is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. Vixra.org will not be responsible for any consequences of actions that result from any form of use of any documents on this website.

Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.

Artificial Intelligence

The Illusion of Benchmarks: Our Model Achieved 99.8% on a Dataset It Wrote Itself

Submission history