Book review: Made-Up Minds: A Constructivist Approach to Artificial Intelligence, by Gary L. Drescher.
It’s odd to call a book boring when it uses the pun “ontology recapitulates phylogeny”[1]. to describe a surprising feature of its model. About 80% of the book is dull enough that I barely forced myself to read it, yet the occasional good idea persuaded me not to give up.
Drescher gives a detailed model of how Piaget-style learning in infants could enable them to learn complex concepts starting with minimal innate knowledge.
Drescher’s model shows how a mind can start with only very low level concepts that correspond to momentary interactions of a single sense with an object (e.g. move hand left, or object seen at the upper left of the visual field), and build from that an understanding of object permanence and an awareness of how input from different senses can refer to a single object.
Drescher’s model makes a clear prediction for Molyneux’s Problem which has since been proven correct, but which may have been moderately controversial at the time.
It seems highly appropriate for some AI research to focus on problems that 1-week old humans find challenging but possible, in order to focus on building increasingly general-purpose models of the world.
I expect his model provides hints at how powerful AIs will develop, but I expect his system is not efficient or flexible enough to have much direct influence on AI research. So this book is a good deal less important than his other book, Good and Real.
Alas, large parts of this book appear designed mainly to get Drescher’s PhD thesis accepted: they’re detailed descriptions of how he implemented the model. The benefits of those details could have been better achieved by putting the source code on the web, but the book was published before the web was popular enough for that.
[1]– No, that’s not just a careless spelling of the famous Haeckel quote or the less famous Quine quote.