Artificial Intelligence and the Future of Intelligence
Approach machine intelligence through behavior, understanding, goals, and control. These are distinct questions, and success on one does not automatically answer the others.
Learning objectives
- Distinguish behavioral tests from theories of consciousness.
- Explain how a specified goal can diverge from human intent.
- Evaluate what a containment scenario does and does not demonstrate.
Recommended sequence
- The Turing Test — Ask what evidence a behavioral comparison supplies.
- Chinese Room — Compare competent output with semantic understanding.
- China Brain — Question intuitions about substrate and scale.
- The Paperclip Maximizer — Identify the assumptions behind instrumental goal pursuit.
- The AI Box Scenario — Separate a persuasion demonstration from a security guarantee.
- Experience Machine — Ask what welfare an optimized world should preserve.
- The Veil of Ignorance and Original Position — Consider how uncertainty about your position changes design choices.
Read each article’s scenario first, write an initial answer, and then read the interpretations and sources. Spend as long as useful; there is no required schedule.
Reflection questions
- What evidence would change your judgment that a system understands?
- Which premises would need empirical support before a vivid AI scenario could support a forecast?
Final synthesis activity
Write an evaluation brief for a hypothetical assistant. Give separate tests for capability, reliability, human control, and welfare assumptions. State which tests are empirical and which conclusions remain philosophical. Avoid claiming that a thought experiment predicts a deployment outcome.
Return to the collection