Foundations of AI: Philosophy — Beginner-Friendly Notes

1. Why Philosophy Before Code?

Philosophy asks the big questions AI tries to answer: What is intelligence? Can machines think? What is knowledge and reasoning? Without this, we cannot define what "intelligent" even means.

2. Five Philosophical Questions That Shaped AI

Q1: Can machines think? (Turing, 1950)

• Turing dodged defining "think" and proposed a test: if you cannot tell human from machine in conversation, call it intelligent — the Turing Test.
• Takeaway for you: AI is judged by behaviour, not by having a soul.

Q2: Is the mind just computation? (Computational Theory of Mind)

• Idea: thinking = information processing (like software on brain-hardware).
• If true, mind can be reproduced on computers.
• Counter-view: Searle's Chinese Room — manipulating symbols ≠ understanding. A person in a room following English instructions to reply in Chinese appears to know Chinese but understands nothing.
• Exam point: This debate is why we distinguish weak AI (acts intelligent) vs strong AI (truly understands).

Q3: What is knowledge and how do we reason? (Epistemology)

• Philosophy gave AI: logic, deduction, induction, belief vs truth.
• AI borrows rationalism (knowledge from reasoning — logic, rules) and empiricism (knowledge from experience — learning from data).
• Modern AI mixes both: rules + data.

Q4: What is rational action?

• Philosophy + economics define a rational agent — chooses actions that best achieve goals given what it knows.
• This is the current definition of AI: not "human-like" but rationally ideal.
• You will use this in PEAS and agent design.

Q5: What are the limits? (Gödel, free will, ethics)

• Gödel showed formal systems have limits (some truths unprovable) — hints AI reasoning also has limits.
• Free will vs determinism asks: if brain is physical, can decisions be predicted/computed?
• Ethics asks: what should AI be allowed to do?

3. Weak vs Strong AI (often asked)

• Weak / Narrow AI: does one task intelligently (all real AI today) — e.g., translator, chess engine.
• Strong / General AI (AGI): human-level understanding across tasks — does not exist yet, debated if possible.
• Superintelligence: beyond human — purely speculative.

In this subject you study weak AI; strong AI is philosophy.

4. Dualism vs Materialism — One Paragraph

• Dualism (Descartes): mind and matter are different — machine (matter) cannot truly think.
• Materialism: mind is what brain does — so a machine that does the same thing can think.
• AI as a field assumes materialism is workable enough to engineer.

5. How This Connects to Your Syllabus

• Logic & reasoning (Module 3) comes from philosophy.
• Agents & rationality (Module 1 next topics) is the philosophical answer to "what should AI do?"
• Ethics (Module 5) is applied philosophy — bias, responsibility, impact.

6. Common Beginner Mistakes

• Confusing Turing Test with proof of consciousness — it only tests observable behaviour.
• Saying Chinese Room proves AI is impossible — it only argues syntax ≠ semantics; weak AI still works.

7. One-Line Summary to Write in Exam

Philosophy gave AI its core questions — can machines think, what is knowledge and rationality, and what are the limits — leading to the modern view of AI as the design of rational agents, with the weak vs strong AI distinction framing what is achievable today.

8. Quick Self-Check

1. What does the Turing Test actually test?
2. Summarise the Chinese Room argument in 2 lines.
3. Why does AI today aim for "rational" rather than "human-like"?