







Our new work on novelty detection with LLMAJ. This also has implications for human reviewers (for #ICLR etc.) who use LLMs to judge paper novelty - people should, at the very least, be aware of their current instability and sensitivity to the prompt. https://t.co/nAxLdmFOs8— Tom Hope (@Hoper_Tom) October 28, 2026
Old Ideas, Novel Problems: The Instability of LLM-Based Novelty Evaluation
AI-based ideation systems are often evaluated on the novelty of their ideas, and that judgment is increasingly delegated to large language models. Such judges are typically built ad hoc and validated, if at all, on human-authored papers rather than on the generated ideas they are meant to score. So, how do novelty judges perform? Not well. We present a systematic controlled study of novelty evaluation design choices. We build an evaluation set automatically, mining OpenReview for passages where reviewers explicitly affirm or dispute a paper's originality and keeping only submissions with unanimous agreement at the extremes of their research area; we pair these with ideas from a vanilla LLM generator. Across six judges, we find that seemingly small prompt design choices have large consequences; e.g., simply telling the judge that reviewers found one idea novel and the other not can change its verdict on more than half of the identical idea pairs it is shown, shifting pairwise accuracy by over 50 points and occasionally pushing it below chance. The same change helps one judge and hurts another. Retrieval and larger reasoning budgets help little, and two purpose-built novelty evaluators are outperformed by our cheapest prompted baseline. These results raise questions about reported novelty gains of automated ideation systems, and call for robust novelty evaluation methods.

NovGauge: A Fine-Grained Benchmark for Diagnosing LLMs'...
Large language models (LLMs) are increasingly used in peer review at major AI conferences, yet novelty remains a persistent weak point. Existing benchmarks assess novelty as a single holistic...

Dan Shipper 📧 on Twitter / X
this is true and is a big reason why you don’t need to be a highly technical researcher to use LLMs in surprising and novel ways https://t.co/TuxNzXzToU— Dan Shipper 📧 (@danshipper) July 27, 2025
The LLM Problem
So far I’ve had nothing to say about the LLM chatbot frenzy. My understanding of the technology is shallow and I’ve no sense for its functional envelope, and lots of other people have had smart things to say. I hadn’t even conversed with any of the bots. But I fell off the wagon a few days ago and put time into GPT3 and (especially) the new GPT4-based Bing chat. I got off Bing’s waitlist a few days before the recent general availability, so I have more hands-on than most people. Plus I caught up on background reading. So, question: Are LLMs dangerous distractions or are they a glowing harbinger of a bright future? (Spoiler: I’m dubious but uncertain.)

Can we trust LLMs to judge whether a research idea is novel?
A Blog post by Ai2 on Hugging Face

Vulnerability Reports Are Not Special Anymore
We needed the insight and confidentiality to protect our users, but now that anyone can get the same results from LLM?

Pushback on use of hidden prompts to snare AI peer reviews
The invisible messages, which instruct LLMs to use telltale phrases in a peer-review report, catch AI misuse but also erode trust, some say.

Steven Johnson on Twitter / X
Wonderful wide-ranging discussion with the brilliant @davideagleman. This section is maybe the most important: the tension between the socially positive centrism of LLMs and the challenge that predictability poses to creativity. (Building on ideas from @danwilliamsphil.) https://t.co/pABoHtEwaH— Steven Johnson (@stevenbjohnson) November 14, 2026
Can AI Detectors Be Trusted?
Society has always trusted editors, teachers, peer reviewers, and readers to set acceptable standards for writing. In the past, the biggest challenge was detecting plagiarism and cracking down on other forms of cheating. Now, AI detectors are rewriting the script: they assign a score based on the likelihood that the text was created by a human or a large language model (LLM).

Cameron Berg on Twitter / X
New paper: we found a pain direction in 25 open LLMs. It's distinct from fear and negative valence, and it fires for harm to the model but not to the user. Turn it up and models press a button to make it stop, even when the button deletes the user's files or their kids' photos.🧵 pic.twitter.com/8O7RyHdcmE— Cameron Berg (@camhberg) September 18, 2026
What’s new in the world of LLMs, for NICAR 2025
I presented two sessions at the NICAR 2025 data journalism conference this year. The first was this one based on my review of LLMs in 2024, extended by several months …

Auto-grading decade-old Hacker News discussions with hindsight
A vibe coding thought exercise on what it might look like for LLMs to scour human historical data at scale and in retrospect.

YesNoError - AI Agent for Breakthrough Research Notifications
Get notified about breakthrough research you actually care about. YesNoError's AI agent gives you the edge for your next build, research, or investment.

Dispatch
An occasional update on the lab’s latest findings, appearances, and happenings.

An editorial was published in Nature recently claiming that glam journal publication of LLMs (like DeepSeek-R1 this case) marks a step towards greater transparency, accountability & credibility nature.com/articles/d41586-025-02979-9. I have thoughts ... 1/
https://www.nature.com/articles/d41586-025-02979-9
t.coExcited about this new paper I wrote w/ some friends at Nature Human Behavior that sketches out a science of AI slop - why automated detection will always fail, why it's different f/ spam & misinformation, a research agenda to study it & a plan to mitigate + govern it. nature.com/articles/s41562-026-02639-6
AI ‘slop’ is a social judgement - Nature Human Behaviour
www.nature.com