Meta co-founder and CEO Mark Zuckerberg has once again raised the banner of more accessible and transparent artificial intelligence (AI). And, in line with this vision, he has leveled new criticisms at rivals who invest in a more restrictive concept of this technology.
His most recent statement came to light in an essay published this Monday, August 10th. In 6,500 words, he emphasizes that, in the coming years, people will be able to use "superintelligence" to create and discover extraordinary things and improve health and quality of life, among other issues.
Zuckerberg notes that, as we approach that moment, it's important to develop a philosophy about how we can best use this superintelligence to ensure it enhances our lives, our work, our communities, our freedom, and our security.
“The defining questions of our era 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?” he asks at the beginning of the text.
“Some argue that superintelligence itself, or a small group of experts who control it, should decide what is best for humanity. We disagree,” Zuckerberg continues, adding:
"The history of democracy and economics demonstrates that there is no single objective answer to how people best define life, and therefore the best approach is to let people decide what matters in their own lives," he states in another passage.
Titled “The future is for everyone – the path to a positive future for AI”, the essay was accompanied by the release of new parameters for Muse Glimmer, a new open AI model developed by Meta, which developers will be able to download and modify.
In a separate statement regarding this announcement, the company that owns Instagram, WhatsApp, and Facebook also stated that it will release the same specifications for the Muse Spark, the most powerful version of the new model, in the coming weeks.
At the same time, the company also announced a $1 billion fund for communities surrounding the locations that host its data centers in the United States, as it accelerates the construction of the infrastructure needed to support its ambitions in this space.
Alongside these announcements, in his essay, Zuckerberg also raised the stakes by launching other arguments against the concentration of artificial intelligence and its "superpowers" in the hands of governments and companies.
“Most other labs are focused on developing AI for businesses, governments, or other institutions. Therefore, if these labs take the lead, the balance of power will favor large institutions at the expense of individuals,” he wrote.
Although he doesn't mention names, Zuckerberg seems to be referencing companies like OpenAI and Anthropic , which, so far, have favored closed artificial intelligence models in their offerings. And he uses a comparison with cybersecurity issues to defend the opposite concept.
"In cybersecurity, widely deployed open-source software has proven to be more secure because more people are able to identify vulnerabilities, strengthen systems, and update them to the latest and most secure versions," he wrote.
On the other hand, Meta has been trying to champion the discourse of a more transparent AI, open to external developers. This was reinforced by its CEO, who indicated that the company will continue investing in open-source models.
“Meta’s mission, since its founding, has been to put power in the hands of the people. If our beliefs and principles guide the process, the balance of power will favor individuals and a better future for all,” Zuckerberg stated.
He highlighted, however, that Meta is implementing a new governance structure, through which the company's board of directors will have the power to decide whether the new models developed by the company meet its safety criteria.
“I don’t believe it’s in my interest, in Meta’s interest, or in the world’s interest that I or anyone else should be the sole decision-maker on how superintelligence is deployed,” he stated, reinforcing that this model should be replicated across the industry.
Next, Zuckerberg commented on the implementation of a policy by the Donald Trump administration that establishes a voluntary 30-day review period for new models. He states that this is a rather significant period of time and proposes an alternative format.
“I believe it will be more productive to have a proactive and close collaboration between leading laboratories and the government, rather than a rigid process and a review schedule to be followed in all cases.”
According to Zuckerberg, in some cases, delaying model launches by a month could compromise the United States' leadership and lead to worse outcomes. "By ensuring the government has early access, we can advance national security while minimizing delays in public launches," he added.
Zuckerberg's essay comes about two weeks after the billionaire gave an interview to the British newspaper Financial Times, where, among other things, he said that the American government should not block the use of Chinese AI models , as this approach would not be an "effective solution" in competing with these rivals.
In the conversation, he also defended what he called an "AI for all" vision, reiterating his support for open artificial intelligence models, which can be personalized and customized by users, a format widely used by Chinese developers.
