Re: Algorithms and Creativity

From: HARNAD Stevan (
Date: Tue Jun 04 1996 - 09:29:24 BST

> From: "Chatwin Judy" <>
> Date: Thu, 23 May 96 15:59:00 PDT
> An initial reaction would be to say that it is not possible to have an
> algorithm for intellectual creativity. (Artistic creativity may be
> treated as a totally different concept because a computer program
> (which is an algorithm) can be written that will enable a machine to
> draw a picture - this is not the type of creativity discussed here.)

In artistry, you want to distinguish between technique and convention,
which an algorithm COULD in principle produce, and creativity in the
form either of the usual inventiveness (going beyond the current
convention or style) or "expressiveness," which is a topic in itself,
but, roughly, means adding that something to the technique that makes it
come alive and "say" something to us. Doing art by algorithm alone would
sound (or look) mechanical; it would not pass the Turing Test for not
having been done by a machine!

> The reason for this gut reaction is because it would require a method
> of procedure that is guaranteed to eventually lead to creativity. The
> possible criteria for which may be seen to be something that is
> relatively valuable when measured against what already exists and is
> unexpected. If something is unexpected how can an algorithm be
> guaranteed to find a solution for it?

By definition, if you already have an algorithm that works, then simply
applying it to a new example is not creative. Creativity would be to
produce a new or stronger algorithm, or to come up with a solutionthat
the algorithm alone could not have reached.

> Many of the factors discussed as not being creative may be produced by
> an algorithm, e.g. problem solving, learning, deduction and imitation.
> These have known boundaries and the capacity for repetition. The rarity
> of creativity makes it difficult to fulfil this criteria. However, once
> the limits of the necessary algorithms have been exceeded then the
> scope is there for creativity to exist, thereby suggesting that
> creativity can override the initial enforced limitations.

Well, nothing is guaranteed: Learning the rules, algorithms, techniques,
facts -- this is the "preparation" that Pasteur wrote about. What
happens next may be mostly chance recombinations. But they are much more
likely in the prepared mind -- and the prepared mind is a rarity in all

> The one essential ingredient for creativity appears to be preparation,
> i.e. previously learnt knowledge and skills which then maximise the
> potential for creativity through chance. Presumably, this criteria
> could be provided by an algorithm but would only provide the
> foundations for creativity; the apparent ingredient of chance cannot be
> produced by an algorithm.
> I would conclude, therefore, that there cannot be an algorithm for
> creativity although the case is not as clear cut as initially thought.

Good reply, and already a B+. To break an A, it has to be related (not
just with key words, but integrative concepts that show understanding)
to the bigger themes of the course: How the mind works, unconscious vs.
conscious processes, whether cognition is just computation, how analog
processing might participate in analogies that get you past stale
algorithms, etc.

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:43 GMT