Musk’s Mind Palace: The AI Architect’s Rent-Free Residence

As the Musk v. Altman trial enters its second week, key figures from the tech world have taken the stand, including Greg Brockman, Jared Birchall, and Elon Musk. However, one significant player on the sidelines is Demis Hassabis, the CEO of Google DeepMind.

Hassabis, the mastermind behind Google’s AI lab, founded DeepMind in 2010 as an independent startup before selling it to Google for a reported sum between $400–650 million. Since then, he has spearheaded several groundbreaking AI research projects at Google, such as AlphaFold. Hassabis has risen through the ranks to lead Google Gemini (formerly Google Brain) and the for-profit DeepMind spinoff Isomorphic Labs.

From the outset, OpenAI was designed as a counterforce to Google. Musk revealed that a conversation with Larry Page, where Page seemed unconcerned about AI’s potential impact on humanity, inspired him to establish OpenAI. The court documents and testimonies shed light on the fear that Google, specifically Hassabis, instilled in Musk and Altman’s inner circle.

During Brockman’s testimony, he disclosed Musk’s fixation on Hassabis, with Musk frequently discussing him in the early days of OpenAI. At an AI-focused dinner with Altman and Musk, the first question Musk posed was whether Demis Hassabis was “evil.”

A dinner meeting with Demis before OpenAI’s inception left Musk feeling alarmed. In an email to Brockman and Ilya Sutskever, another OpenAI co-founder, Musk expressed concern about being outmatched by Google, likening their positions to playing in the Super Bowl versus the Puppy Bowl. This prompted Musk to emphasize the need to elevate OpenAI’s performance significantly.

Issues with Google’s approach to AI surfaced shortly after OpenAI’s launch. Hassabis disagreed with Musk’s advocacy for open-sourcing AI, labeling it as potentially dangerous rather than a solution to the AI problem. This disagreement led to tension between the two tech giants, with Google fearing OpenAI’s stance against closed-source AI.

As time passed, Musk’s concerns about OpenAI’s ability to compete with Google DeepMind grew. Musk expressed worry that OpenAI, as a non-profit entity, might not be adequately equipped to keep pace with DeepMind’s advancements. This prompted discussions about merging OpenAI with Tesla to bolster resources and capabilities.

Shivon Zilis, an OpenAI board member, urged Musk to take action against Hassabis, emphasizing the need to slow down his progress. Zilis highlighted the urgency of the situation, stressing that failing to intervene could have dire consequences. Musk, feeling despondent about the situation, acknowledged the challenges of countering Hassabis effectively.

As the rivalry between OpenAI and Google intensified, Musk lost confidence in OpenAI’s ability to rival Hassabis and DeepMind. He contemplated leveraging Tesla’s substantial financial resources to counter Google’s AI dominance. Musk’s pessimism about OpenAI’s relevance against Google’s might was evident in his emails, where he expressed doubts about OpenAI’s future without significant changes.

By early 2019, Musk had resigned to the fact that OpenAI might not be the answer to challenging Google’s AI supremacy. Plans were underway to explore alternative strategies through Tesla. The enigmatic reference to Hassabis in a message from Altman to Musk hinted at ongoing discussions about countering Google’s AI initiatives.

As the trial unfolded, the narrative surrounding Hassabis underscored the intense competition between OpenAI and Google, with Musk’s continuous efforts to outmaneuver Hassabis and DeepMind. The evolving dynamics between these tech giants showcased the high stakes involved in the race for AI supremacy.