Researchers Make AI Their Confidant, But Can’t Afford To Let Their Guard Down
Researchers are increasingly using AI tools to accelerate research, but unpublished work fed into these systems can create new risks around confidentiality and ownership.
The full text for this story couldn't be loaded here — read it at the verified source below.
Verified source · Inc42
Reported by Inc42. Open the original for full media and formatting.
More in Research
All news
ResearchAmazon is launching colorful new Kindles — and a case with physical page-turn buttons
Kindles don't usually change much with each new version, but Amazon is launching next-generation models today that look and feel a whole lot nicer. The base Kindle, Kindle Paperwhite, and Kindle Colorsoft are a millimeter thinner and 30 percent faster at opening books, and their…
Read at The VergeFine-Tuning Diffusion Language Models with Context Selection and Target Weighting
Supervised fine-tuning of discrete diffusion language models masks some response tokens and trains the model to recover their original values from the visible context. The masking pattern therefore determines both the context available to the model and the tokens it learns to predict. Uniform random masking does not explicitly account for the interaction between these choices. We introduce GoldiMask, which selects tokens to reveal as context by approximately maximizing a submodular objective. This objective uses model signals to balance the benefit of revealing tokens against their value as p…
Read at arXiv cs.AIAREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks
We present AREX-2, an effort to advance the self-improving capability of LLM agents, which we define as the ability to iteratively refine a solution at test time. This ability rests on two complementary capabilities: reflection, which produces a solution better than the current one, and long-horizon execution, which keeps the iteration effective over many rounds. We hypothesize that both capabilities are domain-agnostic, and can therefore be learned in scenarios that are well suited for supervision. Accordingly, we synthesize long-horizon improvement trajectories from machine learning and alg…
Read at arXiv cs.AIExamining Variation in How Guided AI Tutors Resolve Student Impasses
When a student is stuck, a tutor faces the assistance dilemma: help given too early can hinder productive struggle, while help withheld too long leaves the student in a frustrating, persistent impasse (i.e., wheel spinning). Generative AI tutors increasingly use guardrails restricting answer-giving, yet little is known about how such tutors behave once an impasse persists. We analyze 20,462 student turns from 1,260 authentic sessions with a guided LLM chemistry tutor, identifying 6,630 impasse turns of three major types: conceptual errors, expressed uncertainty, or help-seeking. We then used…
Read at arXiv cs.AI