Artificial intelligence and human experience

1. Anticipation and Fulfillment in Visual Perception

Reading an interview with philosopher Michael Madary, I was thinking as so many do about how far artificial intelligence (AI) can go in mimicking the human experience. Madary’s bailiwick is philosophy of mind and the ethics of emerging technologies, especially virtual reality. The interview focuses mainly on Madary’s anticipation and fulfillment model of visual perception. The basic model, it seems to me, is equally applicable to human or AI behavior; i.e., visual perception as proliferating perspectives across time. You first see the object from one limited point of view. To truly grasp it, though, you need to anticipate what it looks like from additional perspectives. You move around, double-check, and your anticipation is more or less “fulfilled” or verified. Then the point of fulfillment becomes a starting point for next stage of anticipation, etc. Visual perception is this process of constantly accumulating perspectives, ever absorbing regions of indeterminacy into determinacy, ever approximating an unreachable objectivity of perspective.

It seems AI should be very good at all of this. So what, if anything, about human reality does AI miss? Is it that there is no psychological reality corresponding to the process as AI executes it? Can the distinction of the human’s psychological reality be evidenced? Is it manifest in the idea of motivation? AI can execute the same process but has a different (or absent) motivation vis-à-vis the human counterpart? The human viewing a modern sculpture from different perspectives may be seeking aesthetic beauty or pleasure, which would seem beyond the scope of AI. Although local human actions may mimic AI processes, it may be that the ultimate motivation behind human action in general falls outside the scope of AI – let’s call that ultimate motivation happiness (following Aristotle) or the satisfaction that comes with living a good life (following Plato); is there any comparable motivation for AI?

2. Transcendental Fulfillment

Or how about this take on the AI/human difference. What humans are always seeking, the grand dream under it all, is fulfillment liberated from the anticipation-fulfillment cycle, a sense of contentment that gets us out of the rat race of endless desires and partial fulfillments. For a correlative to the visual perception model, picture yourself gazing at a painting, motionless, without any interest in increasing perspectives, just having the static fulfillment of the beauty in front of you for a certain duration of time. What elapses in that duration of time may be an expression of the thing that is inaccessible to AI. What humans really want is not a sense of fulfillment that is indefinitely deferred by endlessly proliferating perspectives – the never-ending drive for more data that might occupy the AI entity. We want THE sense of fulfillment that comes when you opt out of the cycle of proliferating perspectives, a sense of fulfillment that transcends process. So whereas the anticipation-fulfillment cycle is an end in itself for AI, for humans all such processes are instrumental; the end in itself, that end which motivates the whole process, is that which falls outside of the process, a kind of static contentment that in inaccessible to the AI.

3. Singularity, or Singularities

The concept of singularity might help to clarify the distinction between fulfillment embedded in the anticipation-fulfillment process and transcendental fulfillment, fulfillment liberated from the cycle. Transcendental fulfillment is, let’s say, a metaphysical singularity – the space of infinite oneness alluded to by many philosophers and mystics, indicating escape from the rat race. Compare that to technological singularity, the critical mass at which artificial superintelligence would eclipse all human intelligence, rendering homo sapiens superfluous. Perhaps fears about the technological singularity, about an external AI dislodging us, are misplaced. I will tentatively side with those who say that computers do not really have their own motivation, their own autonomy. The risk then is not of some external AI overtaking us; the risk lies rather in the porosity between human reality and AI. The risk is not that AI will defeat us in some battlefield Armageddon but rather that it will bleed the life out of us, slowly, imperceptibly, until we go limp. We have already ceded quite a bit of brain activity to the machine world. We used to have dozens of phone numbers in our heads – now all stored in our “external” brains.” We used to do a lot more math in our heads. (Try getting change today from a teen worker without the machine telling them how much to give.) World capitals? In the external brain. City map layouts in one’s head? Replaced by GPS real-time instructions.

“So what?” students today might say. “Memorizing all that stuff is now a waste of time?” And they may be right. But as the porosity increases between human reality and the computerized ether we live in, we cede more and more of our basic survival skills to the ether. I don’t expect malice on the part of AI (although the HAL 9000 was a cool concept), but there may come a tipping point at which we have ceded the basic means of species survival to the machine world. And in ceding more control of our inner lives to the external brain, we become more embedded in the anticipation-fulfillment cycle. Even basic human activities take on a query-and-report format. It becomes increasingly difficult to “opt out” of the processing apparatus and find that space of reflection that transcends the endless proliferation of future-directed perspectives.

4. The Historical Side: Dystopic or Utopic

All this talk about homo sapiens being bled out sounds quite dystopic, and perhaps dystopia is the endgame. But not all possible futures are grim. First of all, in structural terms, porosity is two-directional. Ever since the invention of writing, we have transferred information into the external media of books, giving subsequent generations the capacity to “upload” that information and store it in their brains when the books are removed. This prompted Clark and Chalmers, as far back as 1998, to theorize about the “extended mind,” in which the space of the mind is shared by internal processes and environmental objects that work in tandem with those processes. Another parallel is in Wittgenstein’s Blue Book example wherein we use a color chart until we “learn” our colors, and then throw the chart away. In these cases, the external device provides intermediate information storage. We use the Internet in this fashion all the time. Nothing dystopic here. But is it different when the device becomes capable of evolving its own algorithms, generating its own information, and using it to implement tasks that go far beyond mere storage? Perhaps so, but it is not yet clear that the dystopic end is inevitable.

Second of all, in terms of social implication, technology could free us up to spend less of our lives on drudgery and more of our lives in that reflective space of self-fulfillment, working out our own electives of self-realization. Indeed, this is the signature promise of technology in the age of capitalism. Ever since the early 19th-century Luddite rebellion, technology has repeatedly made this promise and repeatedly failed to deliver. Why would it be any different now? It could only be different if there were a fundamental shift in our perspective of what it is to be human.

When Madary exemplifies his visual perception theory with the example of a modern sculpture, he introduces what for me is the wild card in the anticipation-fulfillment cycle: the element of surprise.

“Recall a situation in which you moved to gain a better perspective on a novel object and were surprised by how it appeared from the hidden side … Modern sculpture can be helpful for illustrating visual anticipations because the precise shape of the sculpture is often unclear from one’s initial perspective. Our anticipations regarding the hidden side of a modern sculpture tend to be more indeterminate than our anticipations about the hidden sides of more familiar objects.” (Madary)

Whereas I started out by saying that Madary’s anticipation-fulfillment model of visual perception applies equally to AI and humans, I suspect we might handle the element of “surprise” differently. In the case of humans, “surprise” is a trigger for imagination, a less tractable faculty than might be intelligible to our future friends in the AI phylum. Sure, our AI compatriots might predict possible futures as well or better than we do (and thus they might best us at chess), but is that really “imagination”? Humans imagine not only possible futures, but also alternative presents and alternative pasts, using self-generated imagery to feed nostalgic visions of times gone by. There is something about the creative process of imagination that might separate us from AI and might make us less predictable in the event of “surprises” or disruptions in the normal anticipation-fulfillment process. Since technology has typically failed to deliver on its promise to enhance self-fulfillment time for personal development, we might anticipate another failure when technology says that AI will truly free us up from drudgery. But the result could be different this time if a rupture in conditions is great enough. Imperatives of income inequality and ecological destruction might be rupture enough. As we survey our predicament the way Madary’s viewer surveys the modern sculpture, we might on the other side glimpse the end of capitalism (which may sound dramatic, and yet all ages do end). Perhaps this might jolt the imagination to a new sensibility, a new subjective frame of reference for values like “work” and “technology” and “success” and “self-actualization” – to wit, a new definition of what it means to be fully human.

How rapidly and wholly we make that turn to a new definition of what it means to be fully human will lock in the dystopic or utopic endgame. In the dystopic version, homo sapiens is bled out by some combination of AI and economic and ecological calamities. In the utopic version, consciousness about what it is to be human evolves quickly enough to allay those calamities and to recapture AI as the servant of human ends and not vice versa.

Footnote on Kierkegaard’s 3 modes of lived experience: aesthetic, ethical, religious

The anticipation-fulfillment model of visual perception can be seen as the basic process of Kierkegaard’s aesthetic mode, sensory-based life. A whole life lived on the aesthetic level is lived as a continuous accumulation of equally ephemeral sensory perspectives on one’s own life.

The ethical life turns out the follow the same model but on the ethical level. The ethical life is a continual accumulation of equally provisional ethical perspectives.

The religious life, though, breaks the model. It concerns the absolute. It does not consist of accumulating perspectives but of a singularity; it eschews all the accumulations of visual or ethical perception for the singular relation between the religious subject and the absolute, a singularity which obliterates all mediate or quantitative concerns.

Michael Madary’s Visual Phenomenology

Richard Marshall’s interview of Michael Madary in 3:AM Magazine

BookCoverImage         year-bfly-cover       Cover png



Particles and Swarms

Does anyone know about particle swarm theory? It seems close to a unified theory of everything. Or at least like a pebble whose waves ripple through everything – biology and computer science, quantum physics and relativity, metaphysics and religion.

Basically, it says that independent particles form swarms, wherein each particle spontaneously takes advantage of the experience of the entire swarm. Examples in the natural world include fish schooling, bird flocking, and ant colonies. Swarm intelligence (SI) has apparently (I’m no expert) become increasingly important in artificial intelligence and robotics.

Can this bridge the persistent gap between the predictions of relativity and those of quantum physics? The problem as I see it is that relativity assumes a universe with physical matter of determinate location and mass. Quantum theory says that when you get down to the building block elements in the atom, units of matter no longer have such determinate values, but can only be described in terms of clouds of probability.

The relativity/quantum theory discrepancy has been scrutinized lately by “oil drop experiments” and “pilot waves.” It seems that you can drop oil on a liquid surface and as it bounces along, it interacts with its own ripple waves, creating a pilot wave that resembles the blur that quantum physicists see when they look at an electron or elemental particle – this would mean (I think) that underneath quantum physics is a stable physical reality after all.

So what if you looked at all the fundamental particles (or waves or whatever units you prefer) of the universe together as a swarm, all those pilot waves interacting, the every move of each affected by the every move of all the others, all one singular pattern of vibration? Do you get a 21st-century physics that recapitulates Leibniz’s 17th-century metaphysics of the indivisible unit, the monad? To wit, Leibniz:

“Each monad … adapts itself to all the others outside itself … This connection of all created things … the connection and adaptation of every single thing to all others, has the result that every single substance [every monad] stands in relations which express all the others. Whence every single substance is a perpetual living mirror of the universe … They are but perspectives of a single universe, varied according to the points of view which differ in each monad.”

From Leibniz, it is an easy step to the world view of the Eastern religions. This connectedness of all things, objective or subjective, expressed as material or expressed as Soul – is particle swarm theory the underpinning here also? And in that swarm lies an immanent intelligence, transcendent and mysterious to the individual, but not requiring any external or anthropomorphic god.

To shift from this synchronic view (how the swarm functions across the space of the many particles) to a diachronic view (how the swarm functions across time), the swarm is the intelligence that drives the trajectories of evolution, terrestrial and cosmic, or, more viscerally, all a singular shudder in some vast cosmic orgasm. A fifteen billion year–old orgasm, you say? Why not? From what I know of Einstein and Hawking, the universe may be one minute old from some other reference point, but only seem fifteen billion years old to us because we are near the event horizon of some black hole, where time becomes stretched toward infinity.

I am no expert in these fields, but I hope that my lateral thinking about them can stimulate a few thoughts. Even if I do nothing but stimulate streams of imagination, I hope that that in itself is no mean accomplishment.

“Imagination is more important than knowledge. For knowledge is limited to all we now know and understand, while imagination embraces the entire world, and all there ever will be to know and understand.” (Albert Einstein)