Zuckerberg outlines vision for AI future in new letter
Meta CEO Mark Zuckerberg has outlined a vision for the future of artificial intelligence centered on giving individuals broad access to powerful AI systems rather than concentrating them in the hands of governments, companies or a small group of institutions, Qazinform News Agency correspondent reports.
In a post published this week, Zuckerberg said the development of superintelligence could allow people to create businesses, develop new ideas, discover scientific breakthroughs, learn new skills and improve their health. He argued that the key questions are who will have access to such systems and how they will be used.
Zuckerberg proposed 3 principles for what he described as a positive AI future: individual empowerment, invention and a balance of power. He said superintelligence should be distributed widely so people can use it to pursue their own goals instead of being controlled by a small number of institutions.
Meta plans to develop personal AI agents capable of understanding users’ goals and helping with areas such as relationships, health, careers, finances, education and hobbies. Zuckerberg said the agents should offer strong privacy and security protections and eventually be accessible through different devices, including smart glasses.
He also expects AI to expand people’s ability to create businesses and develop new products, while providing personalized tutors and scientific tools. Zuckerberg predicted that greater access to AI could lead to more economic growth and employment rather than widespread job losses.
At the same time, Zuckerberg acknowledged risks including job displacement, AI misuse, cybersecurity threats, biological risks, government surveillance and the possibility of losing human control over advanced AI. He argued that concentrating superintelligence in a small number of hands could create greater risks and that distributing AI capabilities could provide checks and balances.
Zuckerberg also called for close cooperation between frontier AI companies and governments, including sharing intermediate model training checkpoints and technical resources to help secure critical infrastructure. He said the United States should accelerate AI infrastructure development and maintain leadership in both advanced and open-source AI.
Meta, he said, will focus on delivering personal superintelligence to billions of people and small businesses, including through free or affordable access. The company also plans to resume releasing some open-source AI models and introduce a fully private mode for personal agents.
Earlier, Qazinform News Agency reported that Meta unveiled Muse Code, its first AI coding agent.