HOLY SHIT… Google AI just proved that fine-tuning Gemini 2.5 made it dumber on hard queries. > Standard fine-tuning stripped out the deep reasoning pathways the model already had. Replaced them with shallow pattern matching. The fine-tuned version scored lower than the base
RESEARCH
-
Proposal for secret ARC-AGI benchmark with undisclosed tasks
By
–
Kind of want a ARC-AGI-X test where a reputable organization runs it & builds a validated benchmark with outside expert help, but they never disclose the questions or even the nature of the challenges themselves so the tasks can never be targets. All we see is a leaderboard
-
Genie 3: From Content Generation to Interactive Reality
By
–
Genie 3 signals where frontier AI is going next: From generating content → to generating interactive realities. That changes how we think about R&D, simulation, and even product design. I break it all down in this latest video. Don't miss out on the latest AI advancements!
-
World Models: Persistence and Interaction Over Graphics Quality
By
–
The real breakthrough is not graphics quality. It is persistence and interaction. These world models: → Generate environments auto-regressively, frame by frame
→ React continuously to your actions
→ Maintain environmental consistency for minutes
→ Recall previous -
World Models: Persistence and Interaction Over Graphics Quality
By
–
The real breakthrough is not graphics quality. It is persistence and interaction. These world models: → Generate environments auto-regressively, frame by frame
→ React continuously to your actions
→ Maintain environmental consistency for minutes
→ Recall previous -
YOLOv3: Speed Meets Accuracy in Object Detection
By
–
⚡YOLOv3: Speed Meets Accuracy in Object Detection
— Satya Mallick (@LearnOpenCV) 26 mars 2026
YOLO changed the game with fast detection but accuracy needed a boost. In 2018, YOLOv3 arrived with Darknet-53, residual connections, and multi-scale predictions.
It improved small object detection, enabled multi-label… pic.twitter.com/FNfCXKmaSPYOLOv3: Speed Meets Accuracy in Object Detection YOLO changed the game with fast detection but accuracy needed a boost. In 2018, YOLOv3 arrived with Darknet-53, residual connections, and multi-scale predictions. It improved small object detection, enabled multi-label
-
AI Scientist V1 Completed Before o1-Preview, Models More Capable Now
By
–
The AI Scientist V1 was completed months before o1-preview and reasoning models were released. The models have clearly gotten much more capable since then. Very excited for where things are headed for AI and automated research! Sakana AI (@SakanaAILabs) The AI Scientist: Towards Fully Automated AI Research, Now Published in Nature Nature: nature.com/articles/s41586-0… Blog: sakana.ai/ai-scientist-natur… When we first introduced The AI Scientist, we shared an ambitious vision of an agent powered by foundation models capable of executing the entire machine learning research lifecycle. From inventing ideas and writing code to executing experiments and drafting the manuscript, the system demonstrated that end-to-end automation of the scientific process is possible. Soon after, we shared a historic update: the improved AI Scientist-v2 produced the first fully AI-generated paper to pass a rigorous human peer-review process. Today, we are happy to announce that “The AI Scientist: Towards Fully Automated AI Research,” our paper describing all of this work, along with fresh new insights, has been published in @Nature! This Nature publication consolidates these milestones and details the underlying foundation model orchestration. It also introduces our Automated Reviewer, which matches human review judgments and actually exceeds standard inter-human agreement. Crucially, by using this reviewer to grade papers generated by different foundation models, we discovered a clear scaling law of science. As the underlying foundation models improve, the quality of the generated scientific papers increases correspondingly. This implies that as compute costs decrease and model capabilities continue to exponentially increase, future versions of The AI Scientist will be substantially more capable. Building upon our previous open-source releases (github.com/SakanaAI/AI-Scien…), this open-access Nature publication comprehensively details our system's architecture, outlines several new scaling results, and discusses the promise and challenges of AI-generated science. This substantial milestone is the result of a close and fruitful collaboration between researchers at Sakana AI, the University of British Columbia (UBC) and the Vector Institute, and the University of Oxford. Congrats to the team! @_chris_lu_ @cong_ml @RobertTLange @_yutaroyamada @shengranhu @j_foerst @hardmaru @jeffclune — https://nitter.net/SakanaAILabs/status/2036840833690071450#m
→ View original post on X — @_yutaroyamada, 2026-03-26 08:53 UTC
-
Real Historical Publications Mistaken for AI-Generated Content
By
–
Kudos to @AkinUnver for sharing these publications on 𝕏 a few days ago. At first glance, they look AI-generated, but they’re 100% real. High-res scans here:
→
https://
davidrumsey.com/luna/servlet/d
etail/RUMSEY~8~1~361761~90129210:—Ceride-i-Adliye-also-
…
→
https://
davidrumsey.com/luna/servlet/d
etail/RUMSEY~8~1~361766~90129205:—Ceride-i-Adliye-also-
… -
Sakana AI Advances Foundational AI Technology and Research in Japan
By
–
Sakana AI is an "AI company from Japan, for Japan," and we are fully committed to both developing foundational technologies that are useful in everyone's daily lives, like the recent Sakana Chat, and advancing cutting-edge AI research that leads the world. “Towards end-to-end