Meta CEO Mark Zuckerberg laid out a sweeping vision for AI in a roughly 6,500-word manifesto, arguing that one of the greatest dangers from advanced AI is not simply the technology itself but allowing governments or a handful of companies to control it.
In a blog post titled “The Future Is for Everyone,” Zuckerberg argues that increasingly capable AI should ultimately be distributed broadly, with individuals controlling personalized “superintelligent” agents that understand their goals and work continuously on their behalf. He envisions those agents helping people manage everything from careers and finances to health, relationships and everyday tasks.
“The defining questions of our age are who will have access to superintelligence and what will we direct it towards. Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone?” the CEO of Meta wrote.
The manifesto provides the philosophical underpinning for Meta’s renewed push into open-weight AI. Zuckerberg said open source can prevent concentrations of AI power while encouraging innovation and competition. He said Meta will release some models openly. Early on in the LLM race, Meta had focused on releasing open AI models under the Llama family as a counterweight to proprietary models from OpenAI and Google.
On Threads, Zuckerberg said Meta is open-sourcing the weights for Muse Glimmer, a 30-billion-parameter model that can run locally, and plans to release the weights for its latest foundation model, Muse Spark 1.2.
Zuckerberg also casts AI as a potential economic equalizer. He predicts highly capable personal agents could give individuals access to expertise and capabilities previously available mainly to wealthy people and large organizations, while AI could enable more people to create businesses and inventions.
But the essay also acknowledges risks involving cybersecurity, biological threats, job disruption and the enormous infrastructure required to support widespread AI deployment.