Right now, peer review is built on a pretty simple assumption: a human came up with the idea, did the work, and wrote the paper. So when another human reviews i

Right now, peer review is built on a pretty simple assumption: a human came up with the idea, did the work, and wrote the paper. So when another human reviews it, they're not just checking facts, they're assessing originality, reasoning, clarity, and whether the author actually understands what they’re presenting. But when an AI system like Carl can generate the hypothesis, run the experiments, and write the full paper and it’s polished enough to pass peer review then it forces us to ask: what exactly are reviewers evaluating? Is it the quality of the output? The logic behind it? Or are we holding the human “author” accountable even if they didn’t do most of the intellectual heavy lifting? And that’s the real tension. If we can’t distinguish between human and AI contributions, the foundation of peer review starts to feel shaky. Not because the content is bad but because the system was never designed for this kind of authorship. This is something I’ve been thinking about a lot while building https://papercheck.ai/en https://papercheck.ai/en, which is focused on helping researchers improve their writing before submission. Most of them want to express their own ideas more clearly and not outsource the thinking. But lines are blurring fast. Maybe what we need are reviewers who look at the process, not just the paper. Or new standards around transparency: what parts were AI-assisted, and to what extent? It’s not about resisting AI it’s about keeping the review process trustworthy as the way we create knowledge starts to change.