Meta unveils first AI model from expensive superintelligence team

Meta Platforms has unveiled Muse Spark, the first artificial intelligence model from a team it assembled at great expense last year to catch up with rivals in the race to develop AI.

The company's shares continued to rise and reached a gain of almost seven percent.

US tech giants are under pressure to show that their massive investments in artificial intelligence will deliver results. The stakes are particularly high for Meta after it hired Scale AI CEO Alex Wang last year in a $14,3 billion deal and offered individual engineers pay packages worth hundreds of millions of dollars to fill its new superintelligence team, Reuters reports.

Superintelligence refers to AI machines that could outperform humans in thinking. The Muse Spark is the first in a new series of models from the team and is part of a family of models internally known as Avocado.

The model will initially be available only on the rarely used Meta AI app and website, and in the coming weeks it will replace the existing Llama models that power chatbots on WhatsApp, Instagram, Facebook, and Meta's collection of smart glasses.

"This initial model is intentionally small and fast, but capable enough to reason about complex questions in science, math, and health. It represents a strong foundation, and the next generation is already in development," the company said in a blog post.

It did not disclose the size of the model, a key measure commonly used to compare the computational power of an AI system with the competition.

Meta CEO Mark Zuckerberg has previously tempered performance expectations, telling investors in January that he thought the team's first models "will be good, but more importantly, they will show the rapid trajectory we're on."

"I expect that we will gradually push the boundaries throughout the year as we continue to release new models," he said.

Focus on everyday tasks

Muse Spark can help users with tasks like estimating the number of calories in a meal based on a photo or placing a picture of a mug on a shelf to see how it looks in space.

Meta also introduced Contemplating mode, which runs multiple agents simultaneously to enhance reasoning capabilities. For example, while one agent is planning a family vacation, another can search for kid-friendly activities.

The company is betting that applying artificial intelligence to everyday personal tasks will help increase engagement among the more than 3,5 billion users on its social platforms, which could give it an advantage over rivals with smaller reach.