Turochamp EmulatorChess · History · Artificial intelligence
An idea from 1948.
A different way to play.

1950 / The imitation game

Playing well. Thinking well?

Chess gives intelligence a scoreboard. Conversation makes the question less tidy.

A question about evidence

Turing’s 1950 paper Computing Machinery and Intelligence approaches machine intelligence through an imitation game. Instead of beginning with an agreed definition of thinking, it asks us to consider an interaction and the judgements made by an interrogator.

The familiar human-versus-machine reading focuses on text conversation: can the interrogator distinguish the machine from a person? Turing’s original presentation is more involved than the popular shorthand, so the paper remains worth reading for itself.

What a successful performance would show

Convincing conversation is evidence about conversational performance under particular conditions. It does not, by itself, establish consciousness, feelings or reliable understanding in every other setting. The test’s interpretation depends on the participants, the questions and the procedure.

Why chess belongs beside this discussion

Chess gives us relatively clear rules and outcomes. Conversation does not offer the same tidy scoring system. Putting the two side by side helps separate an impressive performance from broader claims about a mind.

What would you need to observe before calling a machine intelligent?

Turochamp is especially useful because its decisions can be exposed. You can inspect the calculation and still find yourself describing the engine as cautious, greedy or adventurous. Such language can be helpful, but it adds a human interpretation to a mechanical procedure.

The connection to ELIZA

A conversational program can invite a sense of personality through a very different mechanism from a chess engine. ELIZA offers a companion topic for exploring that response. Here, the aim is to keep the mechanism visible: what inputs were used, what procedure ran, and what the result does or does not demonstrate.

Try a small thought experiment

Watch one Turochamp move without opening its explanation. Describe why you think it chose the move. Then inspect the candidate scores. Does your explanation match the calculation, or have you supplied an intention that the program never represented?

Sources & further reading
  1. Turing — Computing Machinery and Intelligence, Mind (1950) ↗

External sources require an internet connection.