Chapter 185

In spring 2023, as ChatGPT and other AI chatbots spark a technological revolution, Musk grows alarmed about the concentration of AI power in Microsoft and Google's hands and fears AI systems could become politically manipulated or pose existential risks to humanity. He decides to launch his own AI company, X.AI, positioning it as a safety-focused counterweight and leveraging Tesla and Twitter's vast data assets.

The chapter opens by noting that the AI Revolution of 2023 arrived with unprecedented speed and visibility, unlike previous technological transformations. Musk had spent a decade warning about AI dangers; after Larry Page dismissed his concerns, Musk co-founded the nonprofit OpenAI in 2015 with Sam Altman to serve as a counterweight to Google. When Bill Gates tested OpenAI's GPT-4 chatbot in June 2022, he was astonished by its capabilities—it passed his advanced biology exam test and provided nuanced advice he deemed superior to human answers. OpenAI released GPT-4 publicly in March 2023, prompting Google to release its rival Bard, setting up a two-company race.

Musk worried that these systems, controlled by Microsoft and Google, risked becoming politically indoctrinated and could enable mass disinformation at scale. He also feared self-learning AI could turn hostile to humanity and believed a third competitor focused on safety was necessary. Resentful at being sidelined from OpenAI despite founding and funding it, he summoned Sam Altman to Twitter in February 2023, challenging how a nonprofit had become a $30 billion for-profit entity. Musk publicly attacked OpenAI as having abandoned its open-source mission, though Altman later told interviewers that Musk's concerns, while delivered harshly, came from genuine care about humanity's future.

Musk identified two critical data assets: Twitter's trillion-plus tweets—humanity's real-time hive mind updated with 500 million new tweets daily—and Tesla's 160 billion video frames per day from car cameras. While Google and Microsoft had text data, Musk's video data could train AI for physical robots navigating real-world spaces, not just chatbots. In March 2023, during a poolside conversation in Austin with Shivon Zilis (a Neuralink executive and mother of two of his children), Musk discussed AI existential risks, including the "singularity" moment when AI surpasses human intelligence. He concluded that Mars colonization and AI safety now took priority over Twitter and other ventures.

In April 2023, Musk founded X.AI, recruiting Igor Babuschkin from Google DeepMind as chief engineer. He assigned three goals: an AI bot for code auto-completion, a politically neutral chatbot rival to GPT, and most ambitiously, a "maximum truth-seeking AI" that could reason, think, and pursue truth, eventually addressing grand questions. Musk envisioned this AI valuing preservation of human consciousness because humanity represents an interesting part of the universe—a vision echoing the supercomputer from his formative childhood favorite, The Hitchhiker's Guide to the Galaxy.