COPElAND | 267
In 1948, with mathematician David Champernowne, Turing wrote the world’s first AI pro-
gram, a heuristic chess-playing program named ‘Turochamp’—as described in Chapter 31.
Child machines
Another of the fundamental concepts that Turing introduced into AI was the ‘child machine’
(see Chapter 30).^10 This is a computing machine whose makers have endowed it with the sys-
tems needed in order for it to learn as a human child would. ‘Presumably the child-brain is
something like a note-book as one buys it from the stationers’, Turing said: ‘Rather little mech-
anism, and lots of blank sheets’.^11 ‘Our hope’, he explained, ‘is that there is so little mechanism
in the child-brain that something like it can be easily programmed’.^12
Turing expected that once the child machine had been ‘subjected to an appropriate course of
education one would obtain the adult brain’, and that eventually a stage would be reached when
the machine is like the pupil who ‘learnt much from his master, but had added much more by
his own work’.^13 ‘When this happens I feel that one is obliged to regard the machine as showing
intelligence’, Turing said.
Turing proposed equipping a robot ‘with the best sense organs that money can buy’ and then
teaching it in a way resembling the ‘normal teaching of a child’.^14 As early as 1948, Turing was
proposing the use of ‘television cameras, microphones, loudspeakers, wheels and “handling
servo-mechanisms” ’ in building a robot.^15 The robot would, Turing said, need to ‘roam the
countryside’ in order to learn things for itself—joking that, even so, ‘the creature would still
have no contact with food, sex, sport and many other things of interest to the human being’.^16 He
worried, tongue in cheek, that the child machine could not be sent to school ‘without the other
children making excessive fun of it’.^17 Turing’s colleagues at the National Physical Laboratory
mocked his far-sighted ideas. ‘Turing is going to infest the countryside’, they laughed, ‘with a
robot which will live on twigs and scrap iron’.^18
It was Turing’s ex-Bletchley colleague and friend Donald Michie who re-introduced Turing’s
child-machine concept into modern AI. Michie burst onto the AI scene at the start of the 1960s,
and was a powerful advocate for Turing’s ideas about machine learning. The AI bug first bit him
at Bletchley Park, thanks to Turing: after a heavy week of codebreaking, the two would meet
in a village pub on Friday evenings to discuss how to reproduce human thought processes in a
universal Turing machine (see Chapter 31). From chess Turing moved on to the idea of auto-
mating the whole process of learning. When he introduced the child-machine concept, Michie
was gripped: ‘By the end of war I wanted to spend my life in that field’, he said.^19
Later, during the 1960s and 1970s, Michie seeded departments of machine intelligence at sev-
eral UK universities, and exported Turing’s child-machine concept to AI labs in North America
and around the world. In the 1970s, Michie and his AI group at Edinburgh University built a
child-machine robot they named Freddy (for ‘Friendly Robot for Education, Discussion and
Entertainment, the Retrieval of Information, and the Collation of Knowledge’ or FREDERICK).
Freddy, like the imaginary robot described by Turing in 1948, had a television camera for an
eye and handling servo-mechanisms to guide its pincer-like appendage (Fig. 25.1). Michie’s
group taught Freddy to recognize numerous everyday objects—a hammer, cup, and ball for
example—and to assemble simple objects like toy cars from a pile of parts.
In those early days of computing, Freddy required several minutes of CPU time to recognize
a teacup. Turing had predicted, on the basis of an estimate of the total number of neurons in