Inherents' AI teammate just beat OpenAI and Anthropic at reproducing research

A startup founded by DeepMind alumni says its AI model can replicate scientific papers better than OpenAI's and Anthropic's — and it did it without any human prompting, just by reading the paper and running the code. The company is called Inherent, and it's betting that the next bottleneck in AI isn't writing but verification: can a model actually reproduce what a paper claims, or does it just sound convincing?

The claim is specific — the model reproduced results from published papers in ways that matched the original authors. That matters because the current generation of models can hallucinate with confidence, and the papers they write about look real even when the numbers don't add up. Inherent is trying to solve that by building a system that doesn't just generate text but checks whether the math and the code actually work.

This lands at the intersection of a couple trends: researchers are increasingly using AI to draft papers and run experiments, and the community is starting to notice when those papers contain errors or fabricated results. A model that can catch its own mistakes — or at least flag them — is the kind of thing that makes the difference between a useful research assistant and a dangerous one.

Why this matters for us: when AI starts writing papers and running experiments, we need tools that tell us what's real and what's not — otherwise the whole field gets flooded with stuff that sounds good but doesn't work.