Igor Markov, Software Engineer, Google at MLconf SEA - 5/20/16
0 likes1,346 views
The document discusses the potential dystopian threats posed by AI from a hardware perspective, emphasizing the need for stringent controls on self-replication, self-repair, and energy access for AI systems. It argues for establishing hard boundaries between different levels of intelligence and trust to manage risks effectively. The author suggests that by taming potential threats, AI can be harnessed for protection rather than danger.
Igor Markov, Software Engineer, Google at MLconf SEA - 5/20/16
- 13. How did the humanity survive? • by being smart • by knowing the adversary • by controlling physical resources • by using the physical world to advantage
- 15. Now, back to the dystopian AI myth • AI may become smarter than us • Possibly malicious • The physical embodiments are unclear
- 21. Introduce hard boundaries between different levels of intelligence, trust • Can toasters and doorknobs be trusted? • Who can use weapons? • Each agent should have a key weakness
- 25. Constraints on AI to intercept dystopian threats 1. Hard boundaries between levels of intelligence and trust 2. Limits on self‐replication, self‐repair and self‐improvement 3. Limits on access to energy 4. Physical and network security of critical infrastructure Tame potential threats and use them for protection