Zuckerberg names AI's biggest risk as Meta releases Glimmer open-weight model - AltcoinDaily.co
featured-image

Meta (NASDAQ: META) has released an open-weight AI model called Muse Glimmer alongside a manifesto from CEO Mark Zuckerberg, who stated that the real danger from advanced AI is a single government or company controlling the technology, not rogue machines.

Muse Glimmer is a dense model with 30 billion parameters that fits on a laptop or a single consumer GPU, and Meta is billing it as “one of the highest-performing models of its size.”

Anyone can download the model, customize it, and run it on their own hardware because the weights are open.

Meta reportedly built Muse Glimmer through distillation, training the smaller model on outputs from its larger Muse Spark system so that the compact version inherits capability from the bigger “teacher.”

According to Meta, the weights for a version of Muse Spark 1.2 will follow in the coming weeks. Both models are being referred to as the largest American open-source models to date.

Meta had initially paused its release of open-weight models, which Zuckerberg attributed to a strategy review. Now they are back at it, as Meta Superintelligence Labs is now running.

What did Zuckerberg say about AI centralization in his essay?

Zuckerberg, in a 6,500-word essay titled “The Future is for Everyone: The Path to a Positive AI Future,” pushed back on the doom-heavy tone of much AI discourse.

In an interview, he stated. “I’m personally more worried about centralization than I am about any of the specific risks others are talking about.”

According to Zuckerberg, concentrating superintelligence in a few hands “will naturally lead to outcomes that are less favorable for everyone else.”

He added that “there is no such thing as a singular benevolent superintelligence.”

Zuckerberg also proposed an alternative that rests on three ideas. They are individual empowerment as the source of prosperity, invention rather than automation as AI’s main purpose, and a balance of power as the basis for safety.

On the matter of geopolitics, Zuckerberg wrote that U.S. rules slowing American model releases by even a month could cost the country its lead while foreign models advance.

An independent board and a $1 billion community fund

To square open releases with safety concerns, Meta is adding governance. Zuckerberg said an independent board will hold the authority to approve the safety criteria that govern model releases and to review whether a given release meets them. He said that the board will help ensure open releases “meet rigorous safety standards.”

Zuckerberg also announced a $1 billion “Future is for Everyone Fund” aimed at U.S. communities that host Meta data centers. It is said that it will be directing money toward teachers, first responders, and local energy and water infrastructure.

The fund is being seen as an effort to blunt the growing local backlash against data-center construction in the U.S.

If you’re reading this, you’re already ahead. Stay there with our newsletter.