The Myth of Artificial Intelligence by Erik J. Larson

Author:Erik J. Larson
Language: eng
Format: epub, pdf
Publisher: Harvard University Press

SURPRISE!

Peirce understood the origins of abduction as a reaction to surprise:

The surprising fact, C, is observed.

But if A were true, C would be a matter of course.

Hence, there is reason to suspect that A is true.7

Surprises are out on the long tail of trouble for induction. And abductive inferences seek explanations of particular facts (A), not generalizations or laws, like induction. C, too, is a particular—a surprising fact. So abduction isn’t a generalization at all.

Inferences from particular observations to particular explanations are part of normal intelligence. If Kate, a barista, usually works at the Starbucks on Thursday but not Friday, a computer with knowledge gleaned from prior experience might not expect her on Friday, but would be confronted with a long tail problem if she’s working on Friday, after all. It might be she is working extra hours, or was called in to cover someone who is sick that day. And she might not work on Thursday, because she’s sick, or has been transferred to another store, or quit. These are all particular (surprising) facts that might explain her appearance or otherwise. They are commonsense inferences that don’t rely on generalizations or expectations. (Criminal investigations, by the way, always begin with surprising facts. Induction might tell us that young males commit most crimes, but the investigator still needs to know who in particular is responsible for this one—and the culprit need not be young or male, or even human, as we saw in the Rue Morgue.)

Peirce understood abduction as a weak form of inference, in the sense that it was conjectural—an abduction at time t might be proven wrong at time t + 1. Much inference in the real world is defeasible, that is, proven wrong or incomplete by subsequent observation or learning (say, by reading a book).

Conjectural inference is a feature, not a bug, of intelligent systems. Rosie the Robot might believe that Kate has quit Starbucks because a coworker has provided this information, but when Kate shows up for work ten minutes later, and the coworker is smiling, Rosie the Robot should retract its inference. We scarcely notice how quickly we conjecture plausible reasons for what we see (or read about), and also how quickly we drop or update such conjectures. The everyday world is a constant stream of seemingly surprising facts against a backdrop of expectations. Much of the world, like a traffic light, isn’t a constant surprise—but then, traffic lights do break, too.

The meaning of an observation itself undergoes a conceptual change with abduction, as well. Whereas induction treats observation as facts (data) that can be analyzed, abduction views an observed fact as a sign that points to a feature of the world. Signs can be thought of as clues, because they are understood from the beginning as embedded in a web of possibility that may help point to or shed light on a particular problem or question important to the observer. In rich cultural contexts like crime-solving, clues are necessary because there are too many facts to analyze, and only a few are relevant.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.