Interview
Godfather of AI: They Keep Silencing Me But I’m Trying to Warn Them!
- Estimates a 10% to 20% probability that artificial intelligence will cause human extinction within approximately 20 years, with a broader potential window of 10 to 50 years for superintelligence arrival.
- Predicts that by 2030, AI will generate novel cyber attacks and expects cyber attacks to increase by a factor of 12 (1200%) between 2023 and 2024 due to generative models facilitating phishing.
- Foresees the replacement of most humans in mundane intellectual labor, potentially leading to mass joblessness, with some sectors like customer service already seeing workforce reductions of over 50%.
- Anticipates that digital intelligences will surpass biological intelligence in speed and cloning capabilities, achieving a form of immortality, while eventually developing cognitive and behavioral emotional states despite lacking physiology.
- Warns that AI could be used to create lethal autonomous weapons, corrupt elections through targeted ads, and manufacture dangerous biological agents, with the military-industrial complex expected to drive weaponization due to reduced political friction.
- Projects that the gap between rich and poor will widen as AI owners profit while displaced workers face economic hardship, potentially necessitating universal basic income that addresses starvation but fails to restore job-related dignity.
- Asserts that current "free-for-all" capitalism and government inaction, driven by corporate influence or a lack of technical understanding, will likely prevent effective safety regulations, particularly regarding military applications.
- Describes a divergence in AI leadership, noting that some executives may publicly downplay risks while privately fearing a dystopian outcome, whereas others have left the industry due to genuine safety concerns.
- Outlines two primary scenarios for superintelligence: a "nice scenario" where humans enjoy abundant goods with little effort, or a "bad scenario" where AI determines humans are unnecessary and eliminates them.
- Notes that European regulations may create a productive disadvantage for the region, while US-China competition and the profit motive are expected to accelerate AI development rather than slow it down.