Musk v. Altman trial highlights OpenAI leaders’ fears of DeepMind CEO

In short: Evidence shown in the Musk v. Altman trial suggests OpenAI leaders were strongly focused on Demis Hassabis and Google DeepMind as a top rival.

What happened

Testimony and documents in the ongoing court fight between Elon Musk and OpenAI CEO Sam Altman repeatedly point to Demis Hassabis, the CEO of Google DeepMind. DeepMind is Google’s main AI research group.

OpenAI president Greg Brockman testified that Musk brought up Hassabis “many, many times” in OpenAI’s early years, and described Musk as “fixated” on him. Brockman also recalled a dinner where Musk asked, “Is Demis Hassabis evil?”

Emails and messages shown in court add more detail. In a 2016 email, Musk compared DeepMind to the “Super Bowl” and OpenAI to the “Puppy Bowl,” meaning he saw DeepMind as far ahead. Other messages show worries about whether OpenAI, as a nonprofit at the time, could move fast enough to compete.

The documents also include internal concerns about concentration of power. In 2017, Brockman and co-founder Ilya Sutskever warned Musk that creating a structure where one person could control OpenAI could mirror the kind of “AGI dictatorship” they feared from competitors. AGI is shorthand for “artificial general intelligence,” a future kind of AI that could do many different tasks like a human.

Why it matters

This trial is not only about personalities and corporate control. It is also revealing how top AI leaders think about competition, secrecy, and safety, and those choices can shape the tools the public ends up using.

Source: The Verge AI