The Alignment Problem: Machine Learning and Human Values
Christian, B (2020)
One-Sentence Thesis
The type of agent, human or AI, influences moral judgments in complex trade-off situations, highlighting the need for a nuanced approach to value alignment in AI systems.
Argument Outline
- 1Introduction to the Hard Problem of AI Alignment
- 2Background on value alignment and moral dilemmas
- 3Discussion of top-down and bottom-up approaches to value alignment
- 4Presentation of empirical findings on agent-type value forks
- 5Analysis of the implications for value alignment and AI ethics
Key Distinctions
Top-down vs. bottom-up approaches to value alignment
Human vs. AI agents in moral decision-making
Key Terms
Value alignment
Moral trade-offs
Agent-type value forks
Flashcards
17 cardsRelated Questions
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", non-consequentialist moral judgment contrasts with which of the following?
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", Agent type influences which of the following?
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", Victor Tadros defines which of the following?
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", Joshua Gert defines which of the following?
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", Non-Consequentialist Moral Theorizing explains which of the following?
In Christian, B's "The Alignment Problem: Machine Learning and Human Values", Malle et. al. contrasts with which of the following?