A Review of ``Computing Machinery and Intelligence'' by Alan Turing

The version of this paper I read has minor transcription errors, so I won't currently provide links.

This paper is well-known, little-read, but reasonably interesting; what Turing calls ``The Imitation
Game'' is now known as ``The Turing Test''.  Originally from considering whether machines can think,
he proposes the game of a machine imitating a human being through the written word as a well-defined
substitute for the original question.  While interesting, the paper doesn't hold up very well today.

After describing discrete machines followed by digital and universal computers are several rebuttals
of arguments to the original question.  This is generally uninteresting and consumes roughly half of
the paper's length.  The ninth and last argument very curiously involves ``extrasensory perception''
and is written in a way which betrays the appearance of Alan Turing believing in that nonsense.  The
other eight arguments presented and rebutted shouldn't be very new to anyone living in modern times.

The paper brought many modern issues to my mind as I read it, particularly concerning communications
over the Internet.  It may not be fair, but I found it odd how Alan Turing fails to mention machines
that merely convincingly pretend to be human using far less intelligent methods, as are now commonly
used for disrupting Internet discussions.  It's likely that simple experience of such techniques has
made them seem more obvious than they're truly, but it was still disappointing to not see mention of
basic textual templates, copying the other participant's answers, or the modern method of making the
humans behave so stupidly that verbatim texts can convincingly imitate them, to give three examples.

The final section of the paper discusses how to develop a human-like mind from seeds comparable to a
child's mind.  It's not clear to me if Alan Turing hadn't seen the issues later researchers faced or
not.  It's interesting regardless, that his proposed methods would lead to a machine perhaps capable
of holding intelligent conversation, which current methods will inevitably fail to ever achieve, for
he describes a symbolic and mathematical model of a mind, and not that modern incomprehensible mess.
In a way, the section describes the expert's system, a proven method of making artificial reasoning;
those can be compared against all other methods, far too concerned with vague problems of sensation.

This glimpse into the past is too brief to recommend against reading, despite any issue it may have.