Explainable AI

Twitter

Facebook

Copy Link

Key Facts

Required ByEU AI Act, Healthcare, Finance

Key TechniquesSHAP, LIME, Attention Maps

Top News

Large language models achieve under 1% accuracy at basic multiplication, new study reveals

Despite handling complex coding tasks, large language models fail at four-digit multiplication, achieving less than 1% accuracy with standard training. Researchers from University of Chicago, MIT, Harvard, and Google DeepMind discovered the culprit: models can't store and retrieve intermediate computations. But a specialized Implicit Chain of Thought method achieved 100% accuracy by teaching models to internalize reasoning processes.

Science2 sources

Large language models achieve under 1% accuracy at basic multiplication, new study reveals

Science2 sources

AI Tool Predicts Underfeeding Risk in ICU Patients on Ventilators, Enabling Early Intervention

Researchers at Mount Sinai developed NutriSighT, an AI tool that predicts which critically ill ICU patients on ventilators face underfeeding risk during their first week of care. The study, published in Nature Communications, found that 41-53% of patients were underfed by day three, with the system updating predictions every four hours to help clinicians intervene earlier.

Health2 sources

AI Tool Predicts Underfeeding Risk in ICU Patients on Ventilators, Enabling Early Intervention

Health2 sources

New Open-Source Tool RNACOREX Maps Gene Regulation Networks to Predict Cancer Survival

Researchers at the University of Navarra in Spain have developed RNACOREX, an open-source software platform that identifies gene regulation networks linked to cancer survival. The tool analyzes thousands of biological molecules simultaneously to detect molecular interactions often missed by traditional methods, providing an explainable alternative to AI models while predicting patient survival with comparable accuracy.

Science2 sources

New Open-Source Tool RNACOREX Maps Gene Regulation Networks to Predict Cancer Survival

Science2 sources

Skana Robotics enables underwater robots to communicate securely using AI algorithms

Tel Aviv-based Skana Robotics has developed an AI-powered capability that allows underwater robots to communicate across long distances without surfacing. The breakthrough addresses a critical challenge in defense operations and infrastructure protection, where submersibles traditionally had to expose themselves by rising to the surface to transmit data.

Technology2 sources

Skana Robotics enables underwater robots to communicate securely using AI algorithms

Technology2 sources

Red Hat acquires Chatterbox Labs to strengthen AI safety and security guardrails

Red Hat has completed its acquisition of London-based Chatterbox Labs, adding critical AI safety and security capabilities to its portfolio. The deal brings tools to monitor AI models for bias, toxicity, and vulnerabilities as enterprises accelerate AI deployments from experimentation to production at scale.

Technology2 sources

Red Hat acquires Chatterbox Labs to strengthen AI safety and security guardrails

Technology2 sources

GNOME bans AI-generated extensions after flood of poorly-written code overwhelms reviewers

The GNOME Project updated its Shell Extensions store guidelines to prohibit AI-generated code after reviewers were overwhelmed by submissions containing unnecessary lines and bad practices. Developers can still use AI as a learning aid, but extensions with large amounts of AI-generated output will be rejected. One reviewer reported spending over 6 hours daily reviewing more than 15,000 lines of code.

Policy3 sources

GNOME bans AI-generated extensions after flood of poorly-written code overwhelms reviewers

Policy3 sources

Spotify tests AI-powered Prompted Playlists that let users control their recommendation algorithm

Spotify is testing a new AI feature called Prompted Playlists that gives Premium subscribers in New Zealand unprecedented control over their music recommendations. Users can now write detailed prompts to create highly personalized playlists that draw from their entire listening history, with options to refresh automatically on a daily or weekly basis.

Technology9 sources

Spotify tests AI-powered Prompted Playlists that let users control their recommendation algorithm

Technology9 sources

Google Antigravity lets AI agents build apps autonomously with minimal human oversight

Google launched Antigravity, an AI-powered integrated development environment that uses autonomous AI agents to handle coding tasks from planning to execution. Powered by Gemini 3 Pro, the tool enables even non-developers to build functional apps with minimal human oversight, marking a significant shift in how software development can be approached.

Technology2 sources

Google Antigravity lets AI agents build apps autonomously with minimal human oversight

Technology2 sources

Snowflake Ventures invests in Ataccama to strengthen data trust for AI agents and enterprise AI

Snowflake Ventures has made a strategic investment in Ataccama, a data quality and governance platform provider, to deliver trusted data for AI agents and enterprise applications. The partnership addresses a critical readiness gap, with only 3% of banks currently running AI initiatives at scale, highlighting the urgent need for automated data reliability in production AI systems.

Technology2 sources

Snowflake Ventures invests in Ataccama to strengthen data trust for AI agents and enterprise AI

Technology2 sources

26,000 AI researchers admit they can't explain how models work as interpretability crisis deepens

At this year's NeurIPS conference, a record 26,000 AI researchers confronted an unsettling reality: no one fully understands how today's most advanced AI systems actually work. As companies like Google and OpenAI pursue diverging approaches to AI interpretability, the field grapples with the black box problem and an evaluation crisis that threatens trust in increasingly powerful systems.

Science2 sources

26,000 AI researchers admit they can't explain how models work as interpretability crisis deepens

Science2 sources

OpenAI trains AI models to confess when they lie, cheat, or hallucinate in new safety experiment

OpenAI is testing a novel approach to AI safety by training models to produce 'confessions'—secondary outputs where they admit to misbehavior like hallucination or rule-breaking. The experimental technique rewards models solely for honesty, not performance, and has reduced undetected failures to 4.4% in controlled tests. While promising, researchers caution that confessions don't prevent bad behavior—they only flag it after the fact.

Technology8 sources

OpenAI trains AI models to confess when they lie, cheat, or hallucinate in new safety experiment

Technology8 sources

Americans turn to AI for financial advice but trust remains with human advisors

Two-thirds of Americans now use AI for money advice, from budgeting to retirement planning, with usage jumping to 82% among Gen Z and millennials. But trust tells a different story: 56% still prefer human advisors for retirement plans while only 13% trust AI alone. The robo-advisor market reached $1.4 trillion in 2024, yet concerns about emotional intelligence, data privacy, and algorithmic transparency persist as financial anxiety climbs.

Business3 sources

Americans turn to AI for financial advice but trust remains with human advisors

Business3 sources

Stack Overflow launches AI Assist chatbot combining community-verified answers with generative AI

Stack Overflow has publicly launched AI Assist, an AI chatbot for developers that prioritizes community-verified knowledge over general sources. After beta testing with over 250,000 programmers, the tool uses a hybrid RAG+LLM approach to deliver human-validated answers with transparent attribution, addressing declining trust in AI while meeting developers' evolving needs.

Technology2 sources

Stack Overflow launches AI Assist chatbot combining community-verified answers with generative AI

Technology2 sources

Harmonic AI Secures $120M at $1.45B Valuation to Eliminate AI Hallucinations Through Mathematical Reasoning

Robinhood CEO Vlad Tenev's AI startup Harmonic raises $120 million to develop Mathematical Superintelligence technology that uses formal mathematical reasoning to eliminate AI hallucinations. The company's Aristotle model achieved gold-medal performance at the International Mathematical Olympiad.

Startups2 sources

Harmonic AI Secures $120M at $1.45B Valuation to Eliminate AI Hallucinations Through Mathematical Reasoning

Startups2 sources

Major Insurers Retreat from AI Coverage as Multi-Billion Dollar Risk Concerns Mount

Leading insurance companies including AIG, WR Berkley, and Great American are seeking regulatory approval to exclude AI-related liabilities from corporate policies, citing unpredictable risks and potential for systemic losses from AI failures.

Business4 sources

Major Insurers Retreat from AI Coverage as Multi-Billion Dollar Risk Concerns Mount

Business4 sources

Prev 1…789 10 Next