CONCEPT
Comprehension as a Continuum
Dennett's dissolution of the binary that has organized the AI debate—the all-or-nothing divide between ‘real’ understanding and ‘mere’ computation—in favor of a graduated scale on which every system occupies some position, and the interesting question is always where, not whether.
The sharpest intuition in the AI debate is also its most disabling: the conviction that comprehension is binary, that a system either understands or it does not, that the light is either on or off. From this binary everything else follows: the dismissal of machine output as “mere pattern matching” (pattern matching cannot be understanding; therefore machines do not understand); the appeal to a magical ingredient that humans possess and machines lack; the Cartesian Theater in which consciousness performs for a unified inner audience.
Daniel Dennett spent fifty years arguing that the binary is wrong, that it is an artifact of Cartesian thinking rather than a discovery about the world, and that replacing it with a continuum changes everything about how the AI moment should be understood. The thermostat has a sliver of comprehension. The dog has a good deal more. The human has more still. And a
large language model has—well, that is