A new study shows that large language models make trade-offs to avoid pain, with possible implications for future AI welfare.

This new approach, based on natural selection, dramatically improves the reliability of large language models for practical tasks like trip planning. Here's how it works.

A demo car ad created with Veo 2 from Google DeepMind is so realistic that it's hard to spot the amazing locations, intriguing people and fast cars are all AI.

Talk of AI agents is everywhere in Davos. AI pioneer Yoshua Bengio warned against them.

Can AI pass Humanity’s Last Exam? Discover the bold benchmark redefining artificial intelligence and its potential.

This story incorporates reporting fromPC Gamer, BGR, MIT Technology Review, Computerworld, TechRadar and newsbytesapp.com.OpenAI has released Operator, a largely autonomous AI tool designed to execute tasks on the internet based on simple text prompts.

With "Mind Evolution" LLMs can use search and genetic algorithms to generate and combine different solutions and find the optimal one.

The director of the corporation’s scientific laboratory points to genomics, nuclear fusion and the development of new materials as future areas of exploration for his team

The Chinese startup DeepSeek has released a new AI reasoning model that appears to rival the abilities of a frontier model from OpenAI, the maker of ChatGPT.

Wall Street is always musing about alternatives to Nvidia Corp. One analyst just suggested a big name that's not often in the conversation.

The researchers subjected nine large language models to a series of twisted games to see how they would respond to the idea of 'pain' and 'pleasure'