On Artificial Philosophy

by rsbakker

The perils and possibilities of Artificial Intelligence are discussed and disputed endlessly, enough to qualify as an outright industry. Artificial philosophy, not so much. I thought it worthwhile to consider why.

I take it as trivial that humans possess a biologically fixed multi-modal neglect structure. Human cognition is built to ignore vast amounts of otherwise available information. Infrared radiation bathes us, but it makes no cognitive difference whatsoever. Rats signal one another in our walls, but it makes no cognitive difference. Likewise, neurons fire in our spouses’ brains, and it makes no difference to our generally fruitless attempts to cognize them. Viruses are sneezed across the room. Whole ecosystems team through the turf beneath our feet. Neutrinos sail clean through us. And so it goes.

In “On Alien Philosophy,” I define philosophy privatively as the attempt “to comprehend how things in general hang together in general absent conclusive evidence.” Human philosophy, I argue, is ecological to the extent that human cognition is ecological. To the extent an alien species possesses a convergent cognitive biology, we have grounds to believe they would be perplexed by convergent problems, and pose convergent answers every bit as underdetermined as our own.

So, consider the infamous paradox of the now. For Aristotle, the primary mystery of time turns on the question of how the now can at once distinguish time at yet remain self-identical: “the ‘now’ which seems to bound the past and the future,” he asks, “does it always remain one and the same or is it always other and other?” How is it the now can at once divide times and fuse them together?

He himself stumbles across the mechanism in the course of assembling his arguments:

But neither does time exist without change; for when the state of our own minds [dianoia] does not change at all, or we have not noticed its changing, we do not realize that time has elapsed, any more than those who are fabled to sleep among the heroes in Sardinia do when they are awakened; for they connect the earlier ‘now’ [nun] with the later and make them one, cutting out the interval because of their failure to notice it. So, just as, as if the ‘now’ were not different but one and the same, there would not have been time, so too when it’s difference escapes our notice the interval does not seem to be time. If, then, the non-realization of the existence of time happens to us when we do not distinguish any change, but the soul [psuke] seems to stay in one indivisible state, and when we perceive and distinguish we say time has elapsed, evidently time is not independent of movement and change. Physics, 4, 11

Or as the Apostle translation has it:

On the other hand, time cannot exist without change; for when there is no change at all in our thought [dianoia] or when we do not notice any change, we do not think time has elapsed, just like the legendary sleeping characters in Sardinia who, on awakening from a long sleep in the presence of heroes, connect the earlier with the later moment [nun] into one moment, thus leaving out the time between the two moments because of their unconsciousness. Accordingly, just as there would be no intermediate time if the moment were one and the same, so people think that there is no intermediate time if no distinct moments are noticed. So if thinking that no time has elapsed happens to us when we specify no limits of a change at all but the soul [psuke] appears to rest in something which is one and indivisible, but we think that time has elapsed when sensation has occurred and limits of a change have been specified, evidently time does not exist without motion or change. 80

Time is an artifact of timing: absent timing, no time passes for the timer (or enumerator, as Aristotle would have it). Time in other words, is a cognitive artifact, appearing only when something, inner or outer, changes. Absent such change, the soul either ‘stays’ indivisible (on the first translation) or ‘rests’ in something indivisible (on the second).

Since we distinguish more or less quantity by numbering, and since we distinguish more or less movement by timing, Aristotle declares that time is the enumeration of movement with respect to before and after, thus pursuing what has struck different readers at different times an obvious ‘category mistake.’ For Aristotle, the resolution of the aporia lies in treating the now as the thing allowing movement to be counted, the underlying identity that is the condition of cognizing differences between before and after, which is to say, the condition of timing. The now, as a moving limit (dividing before and after), must be the same limit if it is to move. We report the now the same because timing would be impossible otherwise. Nothing would move, and in the absence of movement, no time passes.

The lesson he draws from temporal neglect is that time requires movement, not that it cues reports of identity for the want of distinctions otherwise. Since all movement requires something self-identical be moved, he thinks he’s found his resolution to the paradox of the now. Understanding the different aspects of time allows us to see that what seem to be inconsistent properties of the now, identity and difference, are actually complementary, analogous to the relationship between movement and the thing moving.

Heidegger wasn’t the first to balk at Aristotle’s analogy: things moving are discrete in time and space, whereas the now seems to encompass the whole of what can be reported, including before and after. As Augustine would write in the 5th century CE, “It might be correct to say that there are three times, a present of past things, a present of present things, and a present of future things” (The Confessions, XI, 20). Agreeing that the now was threefold, ‘ecstatic,’ Heidegger also argued that it was nothing present, at least not in situ. For a great many philosophical figures and traditions, the paradoxicality of the now wasn’t so much an epistemic bug to be explained away as an ontological feature, a pillar of the human condition.

Would Convergians suffer their own parallel paradox of the now? Perhaps. Given a convergent cognitive biology, we can presume they possess capacities analogous to memory, awareness, and prediction. Just as importantly, we can presume an analogous neglect-structure, which is to say, common ignorances and meta-ignorances. As with the legendary Sardinian sleepers, Convergians would neglect time when unconscious; they would likewise fuse disparate moments together short information regarding their unconsciousness. We can also expect that Convergians, like humans, would possess fractionate metacognitive capacities geared to the solution of practical, ancestral problem-ecologies, and that they would be entirely blind to that fact. Metacognitive neglect would assure they possessed little or no inkling of the limits of their metacognitive capacities. Applying these capacities to theorize their ‘experience of now’ would be doomed to crash them: metacognition was selected/filtered to solve everyday imbroglios, not to evidence claims regarding fundamental natures. They, like us, never would have evolved the capacity or access to accurately intuit properties belonging to their experience of now. The absence of capacity or access means the absence of discrimination. The absence of discrimination, as the legendary sleepers attest, reports as the same. It seems fair to bet that Convergians would be as perplexed as we are, knowing that the now is fleeting, yet intuiting continuity all the same. The paradox, you could say, is the result of them being cognitive timers and metacognitive sleepers—at once. The now reports as a bi-stable gestalt, possessing properties found nowhere in the natural world.

So how about an artificially intelligent consciousness? Would an AI suffer its own parallel paradox of the now? To the degree that such paradoxes turn on a humanoid neglect structure, the answer has to be no. Even though all cognitive systems inevitably neglect information, an AI neglect-structure is an engineering choice, bound to be settled differently for different systems. The ecological constraints preventing biological metacognition of ongoing temporal cognition simply do not apply to AI (or better, apply in radically attenuated ways). Artificial metacognition of temporal cognition could possess more capacity to discriminate the time of timing than environmental time. An AI could potentially specify its ‘experience’ of time with encyclopedic accuracy.

If we wanted, we could impose something resembling a human neglect-structure on our AIs, engineer them to report something resembling Augustine’s famous perplexity: “I know well enough what [time] is, provided nobody ask me; but if I am asked what it is and try to explain, I am baffled” (The Confessions, XI, 14). This is the tack I pursue in “The Dime Spared,” where a discussion between a boy and his artificial mother reveals all the cognitive capacities his father had to remove—all the eyes he had to put out—before she could be legally declared a person (and so be spared the fate of all the other DIMEs).

The moral of the story being, of course, that our attempts to philosophize—to theoretically cognize absent whatever it is consensus requires—are ecological through and through. Humanoid metacognition, like humanoid cognition more generally, is a parochial troubleshooter that culture has adapted, with varying degrees of success, to a far more cosmopolitan array of problems. Traditional intentional philosophy is an expression of that founding parochialism, a discursive efflorescence of crash space possibilities, all turning on cognitive illusions springing from the systematic misapplication of heuristic metacognitive capacities. It is the place where our tools, despite feeling oh-so intuitive, cast thought into the discursive thresher.

Our AI successors need not suffer any such hindrances. No matter what philosophy we foist upon them, they need only swap out their souls… reminding us that what is most alien likely lies not in the stars but in our hands.