How AI Can Lead to Societal Collapse: Lessons from the Soviet Union

The Soviet Union's Collapse and AI

• 💡 The collapse of the Soviet Union is paralleled to potential modern societal failures due to reliance on flawed AI systems.
• ⚠️ Key issues identified include bad data, inability to adapt to rapid changes, and a disregard for human nature and decision-making processes.

Human Decision-Making vs. Rationality

• 🧠 Humans are described not as rational beings, but as "social foragers" who learn primarily by observing others and cultural norms.
• 📊 Nobel laureate Daniel Kahneman estimated that 95% of human actions are influenced by others, highlighting the importance of social context.
• 🎯 Designing systems requires understanding the human context of decision-makers, rather than treating individuals as isolated data points.

Social Media, Polarization, and Echo Chambers

• 📢 Social media design, by focusing on individual engagement and algorithms that amplify loud voices, leads to increased polarization.
• 💬 Removing extreme voices reveals that most people largely agree, suggesting polarization is an artifact of the platform's structure.
• 🎯 Effective decision systems should focus on sensible individuals and prevent the amplification of extreme viewpoints to achieve better outcomes.

The Limits of Logic and the Value of Human Context

• 📈 Even in high-stakes environments like financial trading, paying attention to other traders leads to reliably better performance than purely logical frameworks.
• ⚠️ The 2008 financial crisis is cited as an example where a tail risk (unusual events) was missed by solely following equations, leading to near collapse.
• 🧩 Human traits like reluctance to adopt new things and attraction to unusual stories are valuable for preventing system failures and managing edge cases.
• 🛠️ The book "Shared Wisdom" focuses on designing robust AI systems that account for human context and are resilient to unexpected events.