Machine learning, LLMs, autonomous systems and the latest in AI industry.
A breakthrough tool from MIT and IBM predicts AI workload power use in seconds, not days, helping operators slash energy waste without sacrificing performance.
A new open dataset of 30,000 proof-based math problems from 143 competitions worldwide is now available to researchers and students, addressing decades of fragmented access and uneven AI performance in mathematical reasoning.
Most advanced AI models answer every question with absolute certainty, even when guessing. New research from MIT reveals how this flaw emerges and introduces a training method to make models humbler—and more trustworthy.
Explore how rethinking work’s role in society could reshape well-being, ethics, and collaboration in the age of AI. A leading philosopher at MIT argues for balanced integration of labor and leisure.
As artificial intelligence reshapes industries and daily life, MIT’s School of Humanities, Arts, and Social Sciences argues that technical skills alone won’t solve humanity’s greatest challenges. Discover why balancing STEM with humanistic inquiry is the key to future-proof education.
Underwater power cable failures are costly and time-consuming to locate. A new MIT project pairs autonomous robots with human divers to pinpoint and repair faults faster than traditional methods allow.
AI-powered protein design tools are now accessible to non-coding biologists through a no-code platform. Discover how OpenProtein.AI is bridging the gap between cutting-edge AI and practical biological research.
Two MIT faculty members—Jacob Andreas and Brett McGuire—have been awarded the 2026 Harold E. Edgerton Faculty Achievement Award for their groundbreaking work in natural language processing and astrochemistry. Their research spans AI innovation and the discovery of complex molecules in space, reshaping how we understand language and the cosmos.
AI models are accelerating materials science by predicting synthesis pathways. MIT’s new tool can propose reaction conditions in seconds, cutting down years of trial-and-error experiments.
OpenAI updates its Model Spec to prioritize teen safety with new Under-18 Principles, ensuring age-appropriate guidance and support
