learn.ifs.name

Computing Machinery and Intelligence

Turing, 1950, Computing Machinery and Intelligence (Mind)

Subscriber deck — free during early access

9 cards in “7 Learning machines” · whole deck

7 Learning machines

  1. What are the atomic-pile and onion-skin analogies?
    Atomic pile: most minds (and machines) are subcritical, an injected idea dies away, but some are supercritical, one idea sets off a whole theory; can a machine be made supercritical? Onion skin: strip off the mechanically explicable functions of the mind and see whether a “real” mind remains. Turing called both “recitations tending to produce belief”, not proofs.
  2. How much storage did Turing think the imitation game would need, and why was programming the bottleneck?
    He estimated the brain at \(10^{10}\)–\(10^{15}\) binary digits and expected about \(10^9\) to suffice, practicable with foreseeable engineering. But at a thousand digits of programme a day, about sixty workers would need fifty years, so a more expeditious method was needed.
  3. What did Turing propose instead of programming an adult mind?
    Simulate a child’s mind and educate it. The child brain is like a notebook from the stationer’s: rather little mechanism and lots of blank sheets. The problem splits into the child programme and the education process.
  4. What analogy did Turing draw between improving child machines and evolution?
    structure of the child machinehereditary material
    changes of the child machinemutation
    judgment of the experimenternatural selection
    It should be faster than evolution, since the experimenter can use intelligence and isn’t limited to random mutations.
  5. What role did Turing give to rewards and punishments in teaching a machine?
    Events shortly before a punishment become less likely to repeat, and those before a reward more likely: an early statement of reinforcement learning. But rewards alone carry little information, so “unemotional” channels are needed too, e.g. teaching the machine by rewards to obey orders in a symbolic language.
  6. How can a learning machine change its rules if its rules of operation are fixed?
    The rules that change during learning are less pretentious ones, with only ephemeral validity, like amendments under a fixed constitution.
  7. Why did Turing expect a learning machine’s teacher to be largely ignorant of what goes on inside it?
    Unlike ordinary computation, where the programmer tracks the machine’s state at every step, a machine educated from a well-tried child programme will have internals its teacher can’t follow, though he may still predict its behaviour. This contradicts the claim that machines only do what we know how to order them to do.
  8. Why include a random element in a learning machine?
    Learning is a search for behaviour that satisfies the teacher, and with many satisfactory solutions random search avoids being stuck in a large region without solutions and needs no record of what has been tried, as in evolution.
  9. Where did Turing suggest starting the competition between machines and men in intellectual fields?
    He didn’t know: either a very abstract activity like chess, or giving the machine the best sense organs money can buy and teaching it to understand and speak English like a child. Both should be tried.

Source