If Ideas Had Shapes

A quoteblog ranging from philosophers in bathrobes to galaxy-rises

Category: Poetry

Douglas Hofstadter – Le ton beau de Marot (1997)

In any case, the imposition of any reasonably sharp set of constraints will force a writer to explore and discover pathways in semantic space that would otherwise have been left entirely unexplored, and that is a very simple but very deep truth about language and thought.

Advertisements

Douglas Hofstadter – Le ton beau de Marot (1997)

One lesson we can learn from Sagoff’s and Varaldo’s laudable poetic accomplishments is the deceptiveness of the power of selection: If you do a good job in selecting what you need in order to accommodate your self-imposed constraints, you will appear to be in control of your medium, rather than the reverse.

Douglas Hofstadter – Le ton beau de Marot (1997)

…I would simply point out that the field of MT takes for granted a philosophy that seems the antithesis of common sense–indeed, the apotheosis of utter silliness–in the translation of a work such as Perec’s La disparition, where respect for form is clearly just as important as respect for content, and where failing to carry over the lipogrammatic quality from the input text to the output text would be a huge slap in the face to the author–in fact, far more disrespectful than would be the act of inventing from scratch a completely new, plotwise-unrelated novel in the target language, as long as this new novel involved no “e”.

(MT = machine translation)

Douglas Hofstadter – Le ton beau de Marot (1997)

Can you believe the audacity it took to do this? One wolf has become several chipmunks?!

Douglas Hofstadter – Le ton beau de Marot (1997)

Gibson clearly likes Andrault’s stuff–he just doesn’t consider it art. I find this absurd. In a sense I agree that art has to “voice a human intention”, but the act of selection by Andrault is a deep human intention, just as deep as a photographer’s selection of a scene or an event to capture.

(of Jean-Claude Andrault)

Douglas Hofstadter – Le ton beau de Marot (1997)

Today’s AI and cognitive science are still a long ways from explicating the mystery of what a human mind is and does, and this, in my opinion, is just fine. I would hate to think that our minds are so simple as to yield up all their secrets in but a few decades. On the other hand, we are making slow and steady progress, and that too is just fine by me. I would hate to think that our minds are so simple as to be constitutionally incapable of piercing the shroud of fog surrounding what it is that they themselves do.

Douglas Hofstadter – Le ton beau de Marot (1997)

The image Searle wishes to project places the human front and center and downplays the rest nearly completely. In such a case, what else could one wish to map oneself onto but the human?

When, however, one realizes that the human being plays but an inconsequential role–that of dronelike bookkeeper–in a fantastically large and intricate system the interesting aspects of whose behavior take place on a slow-as-molasses time scale far more stretched-out than one can easily identify with, one starts to realize how ingenuous and wrong-headed it is to insist on mapping oneself solely onto the tiny human lost in the middle of it, because doing so completely leaves out of the picture the true source of the system’s complexity and interest.

(of Searle’s Chinese room argument)

Douglas Hofstadter – Le ton beau de Marot (1997)

In summary, artificial intelligence could be described as: the belief in thought as the manipulation of unknown sorts of patterns, the attempt to discover those patterns and the rules for manipulating them, and the strategy of using computers to try out all sorts of possible types of patterns and types of pattern-manipulation rules.

Douglas Hofstadter – Le ton beau de Marot (1997)

Babbage had a wide range of interests, including lock-picking, codes and decoding, wordplay, the nature of humor, politics, and many, many aspects of mathematics, science and engineering.

Douglas Hofstadter – Le ton beau de Marot (1997)

The picaresque tale of la Mettrie is well told in Pamela McCorduck’s deliciously titled book Machines Who Think, a poetically written and truly engrossing, though somewhat biased, early history of AI—and while we’re speaking of translating book titles, I offer you the challenge of translating McCorduck’s title into French. There is a serious snag—namely, that the French relative pronoun qui means both “that” and “who”, and therefore the obvious solution, “Machines qui pensent”, falls totally flat, since it back-translates into “Machines That Think”. Can you do better?