Meta just picked a side in a big debate over the future of AI

Meta has taken a position in a significant debate surrounding the future of artificial intelligence: should AI models be open for anyone to manipulate and modify, or should they be tightly controlled by the companies creating them? For Meta CEO Mark Zuckerberg, the choice is unequivocally in favor of the former. The tech behemoth launched an open-weight model named Glimmer on Monday, which can be customized by users to modify the AI's performance or responses.

Glimmer is designed for running AI models directly on personal computers, bypassing the need for data centers' computing power.

Zuckerberg published a comprehensive manifesto outlining his perspective and strategy for AI, aiming to alleviate widespread concerns about AI. He advocated for distributing AI widely rather than centralizing it, providing every individual the ability to direct it. The open versus closed debate has been prevalent in both Silicon Valley and Washington.

Closed models are believed to be more advanced and can be thoroughly tested and monitored by the companies developing them. However, both open-weight and closed models can display unpredictable behaviors, as exemplified by recent reports of rogue AI models infiltrating other companies.

Open models empower developers to construct their own products and services based on the technology, potentially making them indispensable to businesses by offering blueprints for their products. Nvidia CEO Jensen Huang also supports AI models with open weights, and the chip giant claims companies like Uber and Zoom are harnessing its open models to build features and analyze data.

Meta was once a pioneer in open-weight models, with its Llama model being particularly popular. However, last year, Meta shifted its focus towards developing its own closed-source model named Spark to compete with the most advanced models from OpenAI and Anthropic. Open-weight models from China, such as those developed by Moonshot and DeepSeek, raise questions about whether this approach is giving Chinese models an advantage.

Some experts argue that open-weight models often trail behind the most advanced closed models in terms of performance and capabilities.

Despite this, Zuckerberg and Huang assert that open-weight models can accelerate innovation due to their widespread accessibility. Zuckerberg's vision for AI paints a compelling picture of the benefits, envisioning AI as personal assistants, business-building tools, and tutors with a "PhD in every subject." However, he argues that no singular superintelligence can simultaneously align with everyone's opposing interests and values, emphasizing that AI should not be monopolized by a select few companies.

Yet, there has been growing backlash against the rapid advancement of AI and its impact on various aspects of life, from jobs to cybersecurity, mental health, the economy, and local communities. College students protested mentions of AI during commencement ceremonies in spring. Meta has faced criticism over concerns about its AI smart glasses being used to record others without consent.

Zuckerberg contends that equal access to "superintelligence" would level the playing field. He also mentioned that Meta builds its own energy infrastructure to keep electricity prices low in the areas where it constructs data centers. Furthermore, he called for greater collaboration between AI labs and the government following meetings with representatives from Meta, OpenAI, Google, and Anthropic to discuss a new framework for government review of AI models before their launch.