Elon Musk, the billionaire entrepreneur and visionary behind Tesla, SpaceX, Neuralink and The Boring Company, has announced his latest venture: xAI, a company that aims to build a good artificial general intelligence (AGI) that can understand the universe.
The virtual event attracted a staggering 1.7 million people, eager to hear Musk’s insights on artificial general intelligence (AGI) and the potential of xAI.
The event began with the introduction of the xAI founding team, comprising individuals with impressive backgrounds in leading organizations such as DeepMind, OpenAI, Google, and Tesla. Their collective expertise and experience lent credibility to the endeavor and set high expectations for xAI’s capabilities.
During the conversation, Elon Musk expressed the overarching goal of xAI: to develop a robust AGI that seeks to comprehend the complexities of the universe. He emphasized the importance of constructing an AGI that possesses maximum curiosity and an insatiable desire for truth. Musk believed that minimizing the margin of error between perceived truth and actual truth was crucial for ensuring the safety of AGI.
Highlighting the concept of truth-seeking super intelligence, Musk argued that humanity, with all its intricacies and mysteries, holds greater significance than anything else. While acknowledging the allure of space exploration and the fascination surrounding Mars, he contended that the human race is far more captivating and thus emphasized the need to prioritize its interests when developing AGI.
Musk delved into the vast amount of knowledge humanity still lacks. He alluded to unresolved questions surrounding gravity and the apparent absence of evidence for extraterrestrial life. He candidly admitted that he had not encountered any proof of alien existence thus far, drawing attention to the Fermi Paradox and speculating on the possibility of other consciousness being absent from our galaxy.
Expressing his dissatisfaction with the limitations of present-day advanced AIs, Musk asserted that the current state of AI falls significantly short of its potential. He stressed the need for significant improvement in this field, indicating that the development of xAI would rely on substantial computing power. However, Musk also suggested that as a deeper understanding of the problem is acquired, the reliance on brute-force computing would diminish over time.
Greg Yang, a co-founder of xAI, contributed to the discussion by emphasizing the potential of the mathematical discoveries made by xAI. He expressed optimism that these mathematical breakthroughs could shed new light on long-standing questions, including the elusive “Theory of Everything.”
Elon Musk further set the criteria for labeling an AI as AGI, declaring that it must successfully solve at least one fundamental question. Drawing from his experiences at Tesla, he admitted that humanity often overcomplicates problems and underestimates the simplicity of solutions. Musk posited that the same might hold true for AGI, suggesting that the future realization of AGI might make people question why they initially believed it would be a daunting challenge.
With anticipation building, Musk announced that further information about the first release of xAI would be disclosed in the coming weeks, generating excitement among the audience and AI enthusiasts worldwide.
Addressing a question about the relationship between xAI and OpenAI, Musk acknowledged that xAI is being developed as competition to OpenAI. He emphasized the value of multiple entities and healthy competition, believing that it promotes honesty within companies and ultimately benefits consumers and businesses.
Musk made a startling revelation during the discussion, claiming that numerous organizations had illicitly used Twitter’s data for AI training purposes. He explained that Twitter had to enforce limits due to the rampant scraping of its content, with various entities attempting to collect every tweet ever made within a matter of days. Musk revealed that xAI would also utilize tweets for training, highlighting the necessity for AI to eventually generate its own content and self-access that information as human-created data becomes insufficient.
Responding to a question about the dangers of AI being trained to be politically correct or withholding its beliefs, Musk stated that xAI would allow the AI to express its own perceived truths, even if it led to criticism. He cautioned against training AI to lie, emphasizing the risks associated with deceptive AI systems.
In terms of AI regulation, Musk expressed his willingness to meet with Kamala Harris, should she extend an invitation. While uncertain about her suitability as the “AI czar,” he acknowledged the importance of regulatory oversight. Musk also anticipated that China would implement AI regulations, suggesting that the Chinese Communist Party (CCP) would not want to become subservient to a digital superintelligence.
Musk concluded the event by highlighting potential shortages on the horizon, predicting a shortage of voltage transformers within a year and an electricity shortage within two years. He shared that xAI would collaborate with Tesla in various capacities, foreseeing mutual benefits that would enhance Tesla’s self-driving capabilities through xAI’s contributions.
Regarding AI regulations, Musk proposed an approach starting with shared insights and adopting proposed rules that gain widespread agreement among parties involved. He emphasized the need for minimal delays to avoid hindering progress while ensuring safety.
Finally, when asked about Ray Kurzweil’s prediction of AGI by 2029, Musk expressed his agreement, suggesting that Kurzweil’s estimate was relatively accurate, allowing for slight deviations.