Banger paper from Meta FAIR. They introduce Autodata, an agentic data scientist that builds high-quality training and evaluation data autonomously. The headline result: on a CS research QA task, an Agentic Self-Instruct loop produces a 34-point gap between weak and strong
RESEARCH
-
AI Becomes Primary Engine for Mapping Galaxy’s Planetary Architecture
By
–
5/ The team released open catalogs so any researcher can pick targets for follow-up with ground telescopes and ESA's upcoming PLATO mission. AI is no longer just assisting astronomers. It is becoming the primary discovery engine for mapping the galaxy's planetary architecture.
-
RAVEN AI Pipeline Validates 118 Planets from TESS Mission Data
By
–
3/ Enter RAVEN: an AI pipeline trained on hundreds of thousands of simulated planets and fake signals. It detects, vets, and confirms candidates in one go. Result: 118 validated planets, 31 brand new, and over 2,000 strong candidates from TESS mission data.
-
AI Discovers 100+ Hidden Planets in NASA Star Data
By
–
1/ Holy: Astronomers just pointed an AI at NASA data from 2.2 million stars. It found over 100 hidden planets, including worlds so extreme they shouldn't even exist according to current theory. I love it. Lets break it down and explain what it means :
-
Retracted AI Education Paper Prompts Discussion of Meta-Analyses
By
–
My surprise here seems warranted, this paper was retracted (There are other peer-reviewed meta-analyses of the impact of AI on education finding positive effects, like: https://
researchgate.net/publication/38
7110151_The_effects_of_GenAI_on_learning_performance_A_meta-analysis_study
… though the best evidence of AI helping is from RCTs of interventions with AI tutors) -
LLM perfectly explains tweet about prompt injection origins
By
–
Excerpt from a Claude 4.7 Research report; prompt: “Explain the origins of prompt injection.” Surreal to see an LLM perfectly explain a tweet I made specifically about text that tricked then-SoTA LLMs, accurate down to my use of doubled exclamation points:
-
SHAPE Method Rewards Reasoning Progress Over Verbosity in LLMs
By
–
What if LLMs could reason smarter, not just longer? Researchers from Huawei Taylor Lab, Peking University, and Shanghai University of Finance and Economics introduce SHAPE. The method rewards actual progress in reasoning — not verbosity — by using a two-level system: a
-
Proposing real-world benchmarks for medical AI models
By
–
We need more real-world benchmarks for models. Saying that a model scored "87.4% on the MMLU AQuA-RAT" is useless for anyone who's not a researcher. How about we test for: -ER diagnosis (accuracy + time)
-Radiology reads (scan → diagnosis)
-ICU management (decisions → -
Québec’s Sovereign AI Flagship: Frontier. AI-First. Sovereign.
By
–
QUÉBEC ♡ AI / AI http://
QUEBEC.AI
Frontier. AI‑First. Sovereign. Québec’s sovereign AI flagship. https://
quebec.ai #QuebecIA #QuebecAI #IASouveraine -
LLMs Show Extreme Uneven Competence Like Skilled Specialists
By
–
A surgeon who can perform a complex procedure but can't reliably do basic arithmetic isn't a fraud. It's just a specific and unusual kind of competence. LLMs are the extreme version of that.