OpenAI, the company behind ChatGPT, is investing in research on “superalignment” to address the risks of artificial superintelligence. While superintelligence has the potential to solve important global issues, it also poses a threat to humanity. Elon Musk’s xAI project aims to understand the true nature of the universe, but its claims have been taken too seriously, diverting attention from more important AI issues. It is argued that AI models, including language models like LLMs, are limited by human understanding and can only reflect our current knowledge of the universe. The concept of Truth is seen as a human construct, shaped by our interests and perception. Despite their limitations, AI models have the potential to impact various fields, but it is important to address real AI challenges like bias and misinformation.
https://cpwalker.substack.com/p/the-myth-of-ai-omniscience-ais-epistemological