AI · Theory

Turing Test

Theory

18/215

Turing Test

Turing Test: a judge chats with A (human) and B (program) without seeing them. If the judge cannot tell who is who, the program “passes”. It tests conversational indistinguishability, not wisdom, not morality, not “true mind”.

Limits: a script can fool a weak judge. A useful medical model might fail chat and still save lives. Say the test, then one limit. ELIZA in the 60s already fooled some people with cheap tricks.

Turing Test — output — three lines. Indistinguishability, not wisdom.

Exam tip

Setup + one criticism.

Example

# Turing Test
print("Judge chats with A (human) and B (program).")
print("If judge cannot tell who is who → test passed.")
print("Passing ≠ real understanding.")

Turing Test — output: three lines. Indistinguishability, not wisdom.

Short notes

DefJudge cannot tell human vs program in chat.
TrapPassing ≠ real understanding.

AI · Theory

Turing Test

← All stacks

Theory

18/215

Turing Test

Limits: a script can fool a weak judge. A useful medical model might fail chat and still save lives. Say the test, then one limit. ELIZA in the 60s already fooled some people with cheap tricks.

Turing Test — output — three lines. Indistinguishability, not wisdom.

Exam tip

Setup + one criticism.

Example

# Turing Test
print("Judge chats with A (human) and B (program).")
print("If judge cannot tell who is who → test passed.")
print("Passing ≠ real understanding.")

Turing Test — output: three lines. Indistinguishability, not wisdom.

Short notes

DefJudge cannot tell human vs program in chat.
TrapPassing ≠ real understanding.

Turing Test

Example

Short notes

Questions

Turing Test

Example

Short notes

Questions