Economics of Android Companions

Prior to college, I wondered as an intellectual curiosity, how hard it would be to quantify a human mind, and to make it digitally reproduceable. It was fairly clear from concepts such as transcendental idealism (which I learnt via the Kantian interpretation, though the concept exists in other intellectual traditions), that there could be limits to the variety of experiences available to the mind, though we may not easily discern what they are. Possible limits, not necessary ones. In college, I spent a lot of time deconstructing my use of language, and sense of cognition, and gradually mapped most of it to quantifiable structures in space and time.

Since graduation from college, I haven't seen people primarily, "as people." I have looked at people, and seen information systems. Perhaps the easiest way to explain it, is that I view people in much the same way that most engineers might look at their iPads and phones, thinking, "I don't know how this works, but it wouldn't be too complicated to replicate if I studied its replication all day, with a resourceful team, and lots of financial resources." The notion of personhood hasn't been a critical component of my daily operations for a rather long time - though this could change, of course, if I changed my modus operandi, by choice, through illness, or through other (un?)intentional modifications of self.

Now, ten years later, I've worked in commerce for the bulk of that time, and I try to communicate some of these ideas to commercial audiences that are only interested in market value. Of late I've been thinking about it again, and so here is an attempt to articulate (again possible, but unnecessary) economic value that may result from progress in robotics.

Demand: Software

Human being exhibit a mental architecture that depends on social activity, aural conversation, in non-trivial ways. At this point in history, demand for social activity is fulfilled by the supply of social activity by other human beings. Nevertheless, supply is clearly limited, as the variety of demands is broad, and complex. This implies that for the same quality of interaction (when android interaction is indifferentiable from human interaction), android social interactions will hit the market as soon as their price drops below the price of comparable human interactions. This market is worth trillions - though it's not entirely clear how quickly the transition will happen.

Demand: Hardware

The same organic architecture creates a demand for haptic conversation, in significant quantities also. This is typically displayed in the markets for pets and sex. The market for this seems less urgent. But with the potential for each human being on the planet to obtain a customised android partner, indistinguishable from a human, for a price, it seems that this market too will encounter non-trivial growth once the quality of android surrogates rises to approximate or exceed the standards currently set by human bodies.

Supply: Software

There are a few ways to reduce the cost of research in robotics. The commodification of augmented reality devices and software development kits provides one avenue for any researcher in human-robotic interaction to test robot physiology and psychology, without the need to touch hardware research. In the future it may be that the virtual companion will always be available at a lower price than the mechanically present android.

The android as a conversational companion can only gain utility when it begins to be indistinguishable from organic alternatives. Robotics has, to-date, advanced from producing personalities of reptilian personality, to producing mammalian personality. We have yet to see in the open market, personalities which are indistinguishable from those of a human infant or toddler - but it seems that this not more than a year or two away, and probably already extant in private research. The development of artificial personalities that graduate through such a life-cycle of "degrees of humanness," can only proceed thereafter to be benchmarked by what counts as an artificial personality that is indistinguishable from humans of a particular age. Progress will be measured by increases in the perceived human-age of the android. Once adults approximately 18 years of age are reproduceable, the vector for progress then shifts, from the dimension of age to the dimension of education and culture. This is the hard part, as personality remains a domain of human study that is perceived as unquantifiable by a larger quantity of people. (My own hoary intuition about this is that, given how bored I am with people of many cultures, it won't be that much of a problem once machines of a particular scale lay their analytical methods upon it.)

In terms of technical skill, android personalities which exceed our ability to differentiate themselves from human artisans in any field will progress from amateur benchmarks, to beyond the limits of human professionals. This is the easy part.

Supply: Hardware

A few types of technology will be key to the production of indistinguishable android surrogates.

We must achieve, materials for android bodies that are of sufficiently quality to be indifferentiable from human tissues: (i) these could be in-organic, though it seems that an alternative would be, (ii) to grow organic human bodies and to discover how to interact between silicon brains and organic bodies. Going the way of bio-tech seems a bit harder, since we know so little about molecular biology at the cellular level, at this point in history.

We must achieve, mechanics for these materials that are of sufficient quality to be indifferentiable from human movement. Numerous engine types and seem to have potential here, though the leading candidates appear to be those which consume electricity. The positioning of actuators and mechanisms for force transfer, must achieve mimicry of human musculoskeletal structures.

We must achieve, power sources that are of sufficient quality to be indifferentiable from human endurance. At this point I start wondering about a robot companies that need to be plugged in during physical activity... power cords might find novel positionings, and might even be detachable such that androids would know how to reposition their cords based on changes in body orientation. Cords might be attachable at multiple alternative locations on the android body.

We must achieve, heat dispersion mechanisms that are of sufficient quality to ensure human-like body temperatures. Android sweat, or a liquid cooling system that runs close to power-cords might be the tangible alternatives in the event that fans don't get the job done.


I just wanted to paint this out there, so that I'd have it as a reference, when explaining this sort of future, to other people.

No comments :

Post a Comment