It is now confirmed that Meta has joined the fray to make Artificial General Intelligence (AGI) a reality. Earlier, San Francisco-based AI start-up OpenAI had outlined its plans for how all of its AI projects are leading up to AGI. For the uninitiated, AGI is several steps above what we call AI today and will possess the ability to understand and apply tasks at the level of the human brain. Now, the big question is regulation, who gets to control AGI? This has been a matter of contention ever since the term AGI surfaced in the public realm.
Even as OpenAI and other tech giants are working clandestinely towards their goal of AGI, Meta chief Mark Zuckerberg seems to have laid out his plans for all to see. In an Instagram reel, Zuckerberg announced that his company wants to invest billions into building an open-source version of AI.
“Our long-term vision is to build general intelligence, open source it responsibly, and make it widely available so everyone can benefit. We’re bringing our two major AI research efforts (FAIR and GenAI) closer together to support this,” Zuckerberg said in his post.
The 39-year-old CEO said that his company is currently training its next-gen model Llama 3 and that they are building a massive compute infrastructure to support the company’s future roadmap, which also includes 350k H100s by the end of this year — and overall almost 600k H100s equivalents of compute if you include other GPUs. It needs to be noted that Zuckerberg has not revealed any timeline for the creation of AGI.
AGI is a highly capable AI that has the ability to mimic human performance across tasks, making it seem like the holy grail of AI for many companies. Perhaps, this is also the reason why several critics are uneasy, especially in terms of the risks of building such an AI with massive capabilities.
Following his announcement, Zuckerberg has been criticised for taking a rather irresponsible approach to AGI. His latest announcement has invariably raised the possibility of AGI being made available freely to the public in future. This has raised concerns among experts and other influential figures. The underlying fear here is that such an AI could evade human control and eventually take over humanity.
Dame Wendy Hall, a professor of Computer Science at the University of Southampton, told The Guardian that the prospect of open source AGI was ‘really very scary’. Hall, who is a member of the UN’s advisory body on AI, also said that it was irresponsible of Zuckerberg to even consider it. Hall said that such a technology, in the wrong hands, could do a great deal of harm.
In the same report, another Dr Andrew Rogoyski, a director of the Institute for People-Centred AI at the University of Surrey, said that such decisions need to be taken by international consensus, not in the boardroom of a tech giant. Earlier this week, Sam Altman said at the World Economic Forum that further advances in AI would be impossible without a breakthrough in energy provision, such as nuclear fusion.
At a time of guarded AI development, Zuckerberg’s audacious vision has stirred the tech industry. Unlike secretive approaches by most tech companies, Meta plans to invest in vast compute infrastructure that will be powered by Nvidia’s H100 GPU chips, which is bound to cost substantially more. Regardless of the criticism, Meta’s pursuit of AGI, which, according to Zuckerberg, prioritises transparency and inclusivity, also seeks to mitigate issues surrounding the AI capabilities.