AI Evaluation

TL;DR for operators AI-Paper-Review is useful because it behaves like a disciplined pre-submission review room, not because it makes peer reviewers obsolete. The system selects a panel of AI reviewer personas, makes them review independently, clusters duplicated concerns, ranks the resulting issues by consensus and severity, then compares them with human reviews. That mechanism matters more than the slogan, because raw AI critique is cheap, noisy, and very good at sounding busy. ...

Checking is supposed to calm people down. In business, a second review makes a financial model feel safer. A compliance checklist makes a release feel governed. A senior analyst saying “let me double-check that” gives the room a small dopamine hit of procedural seriousness. Long Chain-of-Thought models have learned the same theatre. They pause. They reconsider. They say “wait.” They verify arithmetic. They sometimes generate reasoning traces so long that one begins to feel the model must be thinking deeply, if only because wasting that many tokens while being shallow seems rude. ...

Multimodal AI has reached the point where it can describe videos, summarize documents with images, answer visual questions, and generate outputs that look satisfyingly complete. This is exactly why evaluation is becoming more dangerous. A system that looks competent is not necessarily reliable. It may miss the one-second event that determines the answer. Or it may notice enough evidence but then produce a fluent, attractive, visually decorated summary that quietly distorts the facts. The first failure is upstream: the model did not capture the decisive evidence. The second is downstream: the output did not preserve and present the evidence in a human-useful way. ...

Pixels to Purchase Orders: A Business Map for Choosing Vision-Language Models Receipts are a good way to ruin an AI demo. A clean product photo is polite. A scanned receipt is not. It has shadows, folds, strange fonts, tiny numbers, merchant abbreviations, table-like structure, and one suspiciously important total amount hiding near the bottom. Ask a generic multimodal assistant what it sees, and it may produce an answer that sounds fluent enough to make everyone in the meeting relax. That is usually the dangerous part. ...

Search failure is boring until it becomes expensive. A research agent asks for evidence. The retriever returns documents. The reasoning model reads them, continues writing, and eventually produces a confident answer. Somewhere in the middle, the evidence was slightly wrong: not irrelevant enough to trigger an obvious failure, not useful enough to support the next reasoning step. The agent proceeds anyway, because that is what agents do when we dress up uncertainty as workflow automation. ...

A slide looks finished. The headline is sharp, the equations are aligned, the answer box is confident, and the design has the mild corporate glow of something that has already been approved by three people who did not read it. That is exactly the problem. For years, text-to-image models failed in a wonderfully obvious way: they could not spell. A poster would say “Qaurterly Reveneu,” the mockup button would contain mystical glyphs, and everyone understood the output was decorative, not operational. Recent models have changed that. They can now place readable text inside images, produce document-like pages, and generate slide-like visual artifacts. The failure mode has become less funny and more expensive: the text may be readable, but the reasoning may be wrong. ...

Procurement meetings have a familiar ritual now. Someone opens a leaderboard, sorts by average score, points at a model near the top, and asks why the company is not using that one. It feels empirical. It is neatly ranked. It has decimals. Very scientific-looking decimals, the most seductive species of decimal. The problem is not that leaderboards are useless. The problem is that we often treat them as scoreboards when they are closer to measurement instruments. A scoreboard tells us who won under agreed rules. A measurement instrument first has to prove that it measures the thing it claims to measure. If the instrument mixes model size, benchmark difficulty, contributor practices, post-training choices, item redundancy, and residual artifacts into one number, then the number may still be useful. It is just not self-explanatory. ...

Clue by Clue: ProjectionBench and the Business of Testing AI Discovery Lab meeting. A scientist has a topic, a research question, and not much else. No dataset yet. No final chart. No results section quietly waiting in the appendix. Just a question and the uncomfortable business of guessing what nature might do. ...

Demo day is not discovery day Demo day has a familiar rhythm. An AI system reads papers, proposes an idea, edits code, runs an experiment, drafts a manuscript, and perhaps even produces something that looks suspiciously like a conference submission. The slide title then arrives with great ceremony: autonomous scientist. The paper AutoResearch AI: Towards AI-Powered Research Automation for Scientific Discovery is useful because it interrupts that ceremony before everyone starts clapping at the PDF generator.1 Its central move is not to deny progress. Current systems really can automate meaningful pieces of research work. They can search, summarize, plan, code, run tools, assemble figures, and draft reports. That is already operationally important. ...

Policy rules are boring until a chatbot applies the wrong one. A customer asks whether they qualify for a refund. The rule says refunds require purchase within 30 days, unused condition, and no prior replacement claim. The model answers confidently. It even writes a neat step-by-step explanation. Wonderful. The explanation looks like reasoning. It may even be correct. ...