tech
The View From Inside the AI Bubble
Secret parties, lavish buffets, and talks of annihilation at one of the largest AI-research conferences
TL;DR
- Max Tegmark believes Artificial General Intelligence (AGI) could lead to human extinction and released an AI-safety index where no company scored above a C+.
- The idea of superintelligence has gained acceptance despite a lack of clear definition, and is promoted by the AI industry alongside other questionable narratives.
- NeurIPS, a major AI research conference, has seen a significant increase in attendance, with researchers and corporate sponsors present.
- The AI industry engages in lavish recruiting events, with high starting salaries and bonuses, and companies like Lambda advertise themselves as 'Superintelligence Clouds'.
- Few papers presented at NeurIPS mention AGI in their titles, and a survey indicated over a quarter of researchers didn't know what AGI stands for.
- Sociologist Zeynep Tufekci warned that the focus on superintelligence distracts from understanding current AI technology and its immediate dangers, such as addiction and the undermining of truth.
- While public conversations are dominated by existential threats, many AI developers are focused on tangible problems, though AI figures like Yoshua Bengio express concerns about future misuse of powerful AI for political advantage.