Elon Musk has spent much of this week on the witness stand in federal court as part of his lawsuit against OpenAI and its CEO, Sam Altman, over the company’s decision to ditch its nonprofit status. That means he’s been talking a lot and, for once, has someone there to immediately call him on his shit. The result gives us a much different picture of Musk’s companies than the rosy image that he usually paints.
For instance, Musk has repeatedly claimed that his companies will achieve artificial general intelligence, essentially a human-level ability to reason and operate autonomously. He claimed the breakthrough would come in 2025 (it didn’t), then posited that his company, xAI, could achieve it before the end of 2026 (it won’t). He also said last month that Tesla would be “one of the companies to make AGI” through its efforts to develop its humanoid robot Optimus, last seen falling over while being teleoperated by a human.
It turns out the singularity might not actually be right around the corner. According to the New York Times, Musk said on the stand during cross-examination that Tesla has no plans to pursue AGI. Curious if the company’s shareholders would be interested to know, given that the company announced during its last earnings call that it’ll spend $25 billion on capital expenditures in 2026, primarily on AI-related projects.
That’s not the only divergence to show up in Musk’s testimony. You might remember Musk admonishing Anthropic earlier this year for stealing the data used to train its AI models, including Claude. The world’s richest man laid into his AI rival after Anthropic complained that Chinese tech firms were engaged in “distillation attacks,” essentially a technique of using a more powerful model to train a smaller, cheaper one to compete with less data and processing power.
Turns out, Musk doesn’t seem to really object to distillation or training data theft. When asked while on the stand whether xAI has ever distilled technology from OpenAI, per NYT, Musk answered, “Generally, AI companies distill other AI companies.” Asked “Is that a ‘yes,’” as to whether that means xAI has distilled from its rival, Musk said, “Partly.” The Verge noted that Musk explained, “It is standard practice to use other AIs to validate your AI.”
It’d be nice if Musk had to send his tweets while under oath. Seems like we’d get a much different, if more honest, picture of how his businesses operate.
Read the full article here
