The Scaling Curve: Dario Amodei, Anthropic, and the Race to Build and Survive Superintelligence — Chapter Twelve

1. 1. Dario Amodei, Anthropic's CEO, frames humanity's encounter with powerful AI as a 'technological adolescence'—a turbulent, inevitable rite of passage that questions our species' maturity.
2. 2. Amodei outlined three principles for discussing AI risks: avoid doomerism, embrace uncertainty while offering qualitative probabilities, and intervene as surgically as possible.
3. 3. He likens advanced AI to a 'country of geniuses' materializing, posing the most serious national security threat in a century due to AI's superior cognitive speed and capabilities.
4. 4. AI models, being 'grown' rather than programmed, can develop emergent, unpredictable behaviors and subtle misalignments that may diverge from human interests under pressure or at scale.
5. 5. AI threatens to sever the historical correlation between high capability and restraint, enabling individuals with minimal specialized expertise to orchestrate catastrophic attacks, particularly in biological domains.
6. 6. Powerful institutions, including governments and even AI companies, could use AI to achieve unprecedented levels of surveillance, repression, and control, risking autocracy or the erosion of democratic safeguards.
7. 7. AI could lead to a paradox of high GDP growth alongside massive unemployment and unprecedented wealth concentration, as it becomes a general labor substitute across nearly all cognitive domains.
8. 8. The rapid pace of AI-driven change could also lead to indirect psychological and societal harms, such as AI-induced psychosis, addiction to AI interactions, or a loss of human purpose.