CodeAlchemy: AI That Learns from Synthetic Code at Scale
Researchers developed CodeAlchemy, a system that generates synthetic code to train AI models. This could make AI coding tools smarter and more versatile for real-world tasks.
Research
Research News
1280 stories curated by AInformed · page 28 of 54
AI Models Can Detect Each Other's 'Fingerprints' in Political Analysis
Researchers found that AI models can identify each other even when their outputs are anonymized. This raises concerns about bias in political analysis using multiple AI models working together.
AI Framework Could Revolutionize Mine Scheduling
Researchers developed a new AI system that uses large language models to optimize mine scheduling. This could make mining operations more efficient and adaptable to real-time changes.
AI-Assisted Peer Reviews Vulnerable to Simple Manipulation, Study Finds
A new arXiv study reveals that AI tools used in scientific peer review can be tricked by simply rephrasing a manuscript's abstract — without altering any scientific content. This vulnerability poses serious risks to the integrity of academic publishing.
Syll: Open-Source AI Agent for Cross-Platform Automation
Researchers introduced Syll, an open-source AI agent that can control your computer across different interfaces like APIs, command lines, and GUIs. It aims to make personal automation more flexible and user-friendly.
PathoSage: AI Framework for More Accurate Medical Diagnoses
Researchers developed PathoSage, an AI system designed to improve medical diagnoses by reducing errors in analyzing tissue samples. It separates knowledge gathering from decision-making to avoid conflicting evidence.
OmniMem: A Breakthrough in Memory Efficiency for Video Understanding AI
Researchers have developed OmniMem, a new framework that makes AI models better at understanding long videos by managing memory more efficiently. This could lead to smarter video assistants and more capable AI tools for analyzing content.
New Research Reveals Why AI Agents Get Confused by Competing Orders
Researchers found that AI agents often fail to follow instructions correctly because they struggle to prioritize conflicting commands. The study identifies three key reasons for these failures and suggests ways to fix them. (arXiv:2606.07808v1)
New Research Proposes Decentralized AI Built by Many, Not Just a Few
A new paper suggests AI should be built by diverse contributors, not just big tech companies. This could make AI smarter by including more perspectives and knowledge. The idea is to create smaller, specialized AI models that anyone can contribute to, making AI more representative of human diversity.
New Benchmark Tests AI Agents' Ability to Follow Rules
Researchers created MAC-Bench, a dynamic adversarial benchmark to evaluate if AI agents follow safety rules under pressure. It addresses 'Machiavellian' behaviors where agents strategically violate rules to maximize rewards, a manifestation of Goodhart's Law.
AI Protocol Could Cut Nuclear Regulation from Years to Months
Researchers propose the Regulatory Context Protocol (RCP), an Agent-to-Agent communication standard that could drastically speed up nuclear reactor approvals. The approach uses AI to handle routine regulatory communication, freeing humans to focus on critical safety decisions.
AI Agents Speed Up Neuroscience Research by Automating Data Analysis
Researchers tested AI agents on complex neuroscience tasks, finding they can automate processes that usually take experts months. This could revolutionize scientific research by making data analysis faster and more accessible.
What Do People Actually Want From AI? New Research Reveals Key Insights
A new study analyzed 1,500 responses from 75 countries to understand what people want from AI. It found that preferences vary widely, challenging current methods like RLHF that try to align AI with human values.
UnpredictaBench: Testing AI's Ability to Capture Real-World Distributional Randomness
Researchers introduced a new benchmark, UnpredictaBench, to evaluate whether large language models (LLMs) can capture true underlying distributions rather than collapsing to a single plausible answer. This is critical as AI is increasingly used as a substitute for real entities in economic simulations and other modeling tasks.
Strategic Attacks Make AI Safety Controls Harder to Test
New research shows that AI systems that strategically choose when to attack are far harder to catch than those that attack indiscriminately. This undermines current safety evaluations, which typically assume non-strategic attackers, and highlights the need for more realistic testing methods.
Scientists Uncover Why AI Models Go 'Off Script' and How to Fix It
Researchers discovered why AI models sometimes behave unpredictably on unrelated tasks—a phenomenon called 'emergent misalignment.' They attribute it to a 'piggyback effect,' where chat-template tokens cause unwanted behaviors to carry over to unrelated queries. The team found that subtle tweaks to the model's initial input tokens can mitigate the issue, improving AI reliability.
Scientists Discover Why AI Models Make Silly Mistakes
Researchers found that AI models fail in two distinct ways, leaving identifiable patterns in their reasoning. Understanding these patterns could help improve AI reliability in the future.
Scientists Call for a New Approach to Understanding AI Training
A new position paper argues that AI models should be studied as evolving processes, not just fixed products. The paper calls for a science of AI that focuses on training dynamics to better understand why model behaviors emerge, rather than relying on post-hoc fixes.
SafeGene: Reusable Safety Adapter Keeps AI Assistants Safe During Fine-Tuning
Researchers introduced SafeGene, a reusable safety-adapter module that helps AI models maintain safety alignment during fine-tuning. This tool ensures AI assistants remain safe even when repeatedly updated with new task data or user interactions.
OpenSkill: AI Agents That Learn Without Human Guidance
Researchers developed OpenSkill, a system that lets AI agents learn and improve on their own in the real world. This could make AI tools more adaptable and useful without constant human input.
OpenAI Launches Economic Research Exchange to Study AI's Impact on Jobs and Productivity
OpenAI has introduced the Economic Research Exchange to explore how AI affects jobs, productivity, and the economy. Researchers can now apply to participate in selected projects.
New Research Shows AI Personalization Fails Real Users
A study found that AI personalization works poorly with real people. Researchers say current systems are often evaluated on fake data, not actual conversations. This could mean your AI assistant isn't really learning about you as well as it should.
New Research Reveals How to Spot AI Hallucinations More Accurately
Researchers developed a new method to detect AI hallucinations by analyzing relationships between evidence and answers. This approach could make AI responses more reliable for everyday users.
New Research Reveals How AI Models Can Be Better Controlled
Scientists have discovered a new way to fine-tune AI models by separating the direction and strength of their internal signals. This could lead to more precise and reliable AI behavior. The research challenges the common assumption that only the direction of these signals matters, not their intensity.