Anthropic's Automated Alignment Researcher Shows Promise in AI Self-Improvement
Anthropic's new paper shows automated systems can improve AI alignment performance on all 10 benchmarks tested, without harming overall performance.
ARTICLES BY MAYA CHEN
LATEST NEWS
-
-
Google Deepmind's Co-Scientist Now Plans Experiments and Writes Papers
Google Deepmind's Co-Scientist, built on Gemini models, now conducts lab experiments, writes scientific papers, and verifies results across three disciplines, according to a 2026 report.
-
AI Shopping Agents Fail to Deliver Consistent Purchases, Study Finds
A study by Wharton researchers shows AI shopping agents can drastically shift recommendations based on context, with one source changing choices by up to 99 percentage points.
-
Enterprise AI's Real Risk Isn't Autonomous Agents, It's the Complexity Between Them
Enterprise AI adoption faces significant challenges not from autonomous agents, but from the complexity in managing interactions between them, according to VentureBeat.
-
ChatGPT and Critical-Thinking Training Boost Student Work
A study involving over 1,000 students found that using ChatGPT and critical-thinking training improved academic work in different ways, with ChatGPT boosting quality and coherence.
-
Hugging Face Unveils TransVLM for Enhanced Shot Transition Detection
Hugging Face's TransVLM model achieves 78.3% segment F1 on public data and 89.5% on synthetic data, outperforming existing methods.
-
Hugging Face Introduces TAVR for Cross-Scene Avatar Generation
Hugging Face's TAVR outperforms existing methods with an overall quality score of 16.42 on a new cross-scene benchmark, surpassing the next best method by 2.29 points.
-
AWS Details Supervised Fine-Tuning Data Preparation for Foundation Models
AWS outlines data preparation steps for supervised fine-tuning, emphasizing quality checks and formatting for effective model training.
-
Hugging Face Explains Engrams in Modern LLMs
Hugging Face clarifies that engrams are embeddings used in large language models to capture linguistic concepts, with a focus on how they handle sequence information.
-
AI Models Struggle with Puzzles Despite Rapid Progress
AI models solved 18% of New York Times Connections puzzles in late 2024, but improved to near-perfect accuracy by early 2025.
-
Researchers Use LLMs to Improve Autonomous Vehicle Motion Planning
A new approach allows self-driving cars to adjust their behavior based on natural language input from passengers, as reported by IEEE Spectrum.
-
Kids Outperform LLMs in Language Learning Efficiency
Human children master language with far less data than AI models, which require vast amounts of text to achieve fluency, according to new research.