Scientists at BitEnergy AI created a method called "Linear-complexity multiplication" (L-Mul) that replaces complex floating-point multiplications in AI models with simpler integer additions.

A new Harvard study draws parallels between large language models (LLMs) and crowdsourcing. The research shows why AI systems often provide correct answers on general topics but tend to make mistakes ...

Qualcomm says the software will act as a go-between for personal information, apps, and on-device AI models. It's designed to combine data like favorite foods, workout routines, and frequent contacts ...

Researchers at Johns Hopkins University have developed an AI model called RATIONALYST that improves the reasoning capabilities of large language models through implicit rationales. The team used a pre ...

A large-scale study involving over 100 natural language processing researchers reveals that AI-generated research ideas are considered more novel than those from human experts. However, the ...

A new study has uncovered significant gaps in the reasoning abilities of AI language models, especially smaller and cheaper ones. These models struggle with chained elementary-level math problems.

According to a recent study, people view headlines labeled as "AI-generated" as less trustworthy and are less likely to share them - even if the content is accurate or was created by humans. They ...

Reflection 70B, touted as the "world's best open-source model," has failed to meet its promised performance in independent tests. Developer Matt Shumer admits mistakes and plans to continue working on ...

A new Gartner study finds that 80% of software developers will require additional training in generative AI by 2027. The study predicts that companies will increasingly need specialized AI engineers ...

Brooks stated on X that he will be working on video generation and world simulators at his new position. "Can't wait to collaborate with such a talented team," Brooks wrote. He thanked OpenAI for ...

At the core of Movie Gen is a 30-billion-parameter transformer model for video and image generation. It produces videos up to 16 seconds long at 16 frames per second, with support for different aspect ...

A math professor at the University of Pennsylvania produces a complex mathematical proof using OpenAI's GPT-o1-mini. However, the path to this result is anything but straightforward. Robert Ghrist, ...