Superintelligent AI could emerge by next year, says Elon Musk

    74
    Elon Musk

    10th April 2024 – (Texas) Elon Musk, CEO of Tesla and SpaceX, has revised his prediction on the arrival of superintelligent artificial intelligence (AI), stating that it could exist as early as next year. Previously, Musk had projected the emergence of superintelligent AI by 2029. While “superhuman” intelligence surpasses the abilities of any individual human in a specific task, “superintelligent” refers to surpassing the combined abilities of all humans across all tasks.

    During a live-streamed interview on his social network, X, Musk expressed his belief that AI smarter than any individual human would likely be developed by the end of next year. However, he added a caveat, mentioning that power demands and shortages of powerful AI training chips could restrict their capabilities in the near term.

    Musk’s concerns about superintelligence and its consequences were evident in 2023 when he stated that he would pause the development of advanced AI if possible. Highlighting the importance of avoiding a dystopian future, he launched his AI startup, xAI, with the aim of building AI technology for positive purposes.

    Recent developments show that xAI is actively pursuing the development of superintelligence. Musk revealed in a recent interview that the latest version of their chatbot, Grok AI, was on par with OpenAI’s leading model, GPT-4. However, competitors such as Anthropic’s Claude 3 Opus have already surpassed GPT-4’s capabilities.

    Musk’s predictions have been known for their ambitious nature. In the past, he has made inaccurate forecasts about autonomous driving capabilities and Mars missions. Nevertheless, his commitment to AI development remains strong. Despite his earlier involvement with OpenAI, Musk founded xAI last year and released the chatbot Grok, which is yet to be widely available.

    Developing AI with superintelligence capabilities requires substantial financial resources. Training the latest-generation systems demands tens of thousands of high-end AI chips, each priced at hundreds of thousands of dollars. Even the world’s richest person, Elon Musk, faces challenges in this regard. Reports suggest that he is seeking billions of dollars in funding from investors in the United States, the Middle East, and Hong Kong to compete with OpenAI, which is supported by Microsoft.

    However, global AI development is currently constrained by chip shortages and limited energy supply. Musk acknowledged that chip scarcity is affecting the development of the next-generation large-scale AI models, and the power supply capacity of local data centres will be a significant challenge. He anticipates that data centres’ voltage transformers dedicated to AI will become strained this year, leaving electricity supply as the remaining major obstacle.

    As AI data centres consuming substantial amounts of power proliferate across the United States, the risk of power shortages and blackouts due to overloaded power networks in many regions has escalated. According to the International Energy Agency, by 2022, the power consumed by 2,700 data centres in the United States will account for over 4% of the country’s total electricity consumption. By 2026, this percentage is projected to rise to over 6%.