The problem isn't that AI wrote that sentence. The problem is that the sentence is bad, and we need to say so clearly regardless of who or what generated it. Bobrodsky's anonymized example, "outperforms the best baseline costing no more at none of six operating points and a permuted placebo at none of three, with gains up to about 1.5 not excluded", is nearly unreadable. It obscures meaning behind a tangle of conditionals and qualifiers, making the reader work harder to extract a simple result. This matters because if we cannot critique the substance of writing without defaulting to "this sounds like AI," we risk losing the ability to hold any technical communication to a standard. The real issue isn't provenance; it's clarity.
For our readers building and reviewing with AI-native tools, this is a practical concern. When you're Finding Your Balance Between AI Effort and Real Results, the last thing you need is output that buries its conclusions under jargon. The sentence above fails because it uses passive constructions ("not excluded," "costing no more") to hedge results, and it strings together dependent clauses that force the reader to hold multiple caveats in memory simultaneously. A clearer version might read: "The model does not outperform the best baseline at any of six operating points, nor does it beat a permuted placebo at any of three. The maximum observed gain was approximately 1.5, though this result was not statistically significant." That version lets the reader see the failure immediately. The original version, by contrast, disguises the failure as complexity. We should call that out as bad practice, not as "AI writing."
The temptation is to focus on detection, is this human or machine?, but that misses the point. The community question Bobrodsky raises is about standards. When we Discover how AI can reveal insights about your own potential, we expect those insights to be communicated clearly. The same expectation should apply to academic peer review. If a paper's writing makes it hard to assess the actual contribution, then the review process must flag that as a flaw in communication, not just in style. The concrete takeaway is this: when you encounter dense, hedging prose in technical writing, ask whether the author is using complexity to mask a weak result. If the answer is yes, say so directly. That critique holds whether the author is human or an LLM, and it's the only way to keep the conversation focused on substance.