The AI giants have not yet found the winning formula for AGI, artificial general intelligence that will (soon, we are promised) be able to match human intelligence. No matter, Meta has decided to look beyond that by creating a new lab tasked with developing nothing less than "superintelligence"—an AI smarter than the poor human brain.
Billions for a cognitive chimera?
As always, when Mark Zuckerberg gets carried away by the current craze, he spends lavishly. The CEO of Meta has attracted 28-year-old Alexandr Wang, founder and boss of the startup Scale AI, to this lab. Meta reportedly invested $14 billion (!) to acquire a 49% stake in the startup, and is reportedly offering 7- to 9-figure salaries to dozens of researchers working at OpenAI and Google, according to the New York Times.
Obviously, Meta is no stranger to the AI industry. The first dedicated lab dates back to 2013, after the failed acquisition of DeepMind — which Google pocketed. ChatGPT then came along; Certainly, OpenAI's technology hardly impressed Yann LeCun, chief scientist at Meta. Nevertheless, Silicon Valley has gone headlong into the hunt for AI, AGI, and... "superintelligence."
But behind this futuristic veneer, the very notion of "superintelligence" divides researchers. For some of them, aiming for an AI smarter than humans is as much science fiction as a techno-solutionist fantasy that distracts from the very real problems associated with current AI: bias, surveillance, ecological impact.
Others see it as a major existential risk, a tipping point where an AI that has become too powerful could escape human control. Skynet is lurking... The concept remains vague, but the investments are very real.
This new lab, and the reorganization of the AI activity within Meta, are also a way for Mark Zuckerberg to regain control. The company's AI division is reportedly disorganized and suffering from internal tensions. Furthermore, products released in recent months—such as the latest Llama or Meta AI models—have reportedly not met with the expected success.
This reorganization aims to give (re-give?) Meta a position as a global leader in AI. But money isn't everything: the current political and regulatory environment, particularly in the United States, is delicate. Meta is out of the question to buy up AI startups galore: regulators are on edge. Mark Zuckerberg is therefore walking a fine line.
Source: NYT
0 Comments