Tuesday, September 25, 2007

Envisioning the Hyperintelligent Droid

How might a product that is smarter than a human being in one or more ways evince that intelligence? What form factor would be optimal for most consumer applications? What useful "powers" might these droids possess?

As I noted in my "What AI will really be like" thread, droid intelligence will take a different enough form that comparing them to human intelligence will be apples-to-oranges, almost certainly involving a large number of arcane metrics to approach anything accurate. Really, it may take decades to even realize that a hyperintelligent, or "more intelligent than human" droid has been developed.

The only evidence-based way to approach advanced AI technologies is to treat them in terms of a manufactured product. This does not mean that the product cycle scenario is absolutely true, but it is the most probable if for no other reason it is the only kind that has ever happened. Advanced technologies are created to be purchased by consumers and businesses. Otherwise, they don't get created.

There is the notable exception of the military, and we could imagine all kinds of scary scenarios with that one. They will do as they are doing now, funding research, but I expect that in this space, as with microprocessors today, they will lag the consumer market once this tech reaches critical mass. And even the military will approach this tech in similar ways as consumers and businesses, bounded survival drive and such. Although they are responsible for pollution and such in the past, even they have never had a tech "get away from them", that started autonomously hurting the citizenry. Even though this is a very special technology, let's keep with the evidence until real risk materializes.

A cautious, evidence-constrained approach lets us imagine actually owning one of these devices, and from there, we can start to understand the feature sets that will make these compelling to consumers and businesses. I will only discuss consumer-oriented products in this thread.

I am in concurrence with one very important tenet of Kurzweil's belief system, that is, that computing power will one day equal and eventually greatly exceed that of a human brain. But to be consistent with typical product cycles, we must envision expressions of this vast intelligence that are consumer-friendly, and are in fact a must have for anyone that can afford one. It absolutely must take this form, or no reasonable predictions of any kind can be made, otherwise they start to stray almost immediately into very creative - and completely unrealistic - speculation.

This may seem philosophically mundane, but get ready for a strange ride. Let’s start slow, take care of the basics to begin to make these devices real. I had originally intended this topic to extend over many threads, but let me introduce this as concisely as possible here.

How do we do get to the droid of the future, the hyperintelligent droid?

The first thing is to clear your mind of all preconceptions. Back away from a 2-hour movie where a metal man like C3-PO seems kind of cool. Your personal droid is going to be with you 24/7, for years on end – the rest of your life, probably. A nelly, clinking droid like C3-PO would get old very quickly; we’d soon be treating it meaner than Han Solo ever did.

Really, what is the main danger to a great customer experience with a product that can walk around, interact with you, and such? It’s not physical danger, conspiratorial designs, or anything like that - we’ve covered that.

The biggest danger will in effect form one of the several thin lines that droids will have to walk with great precision. The biggest danger is annoyance, having our droid annoy us. Human companions of course annoy us, but they are independent people, that is to be expected. But a purchased product does not have carte blanche for this effect on its owner; eliminating any tendency of that kind will be a competitive differentiator for these offerings.

Therefore, this thin line is that the owner can neither intimated by their droid, but also should not feel contempt for it either, which quickly leads to annoyance or worse. So, the droid can't be too subservient or slavish.

What does this mean? Simply put, our droid will have to have our respect. We’ll see that this is necessary for many reasons; several of the ways they will add value to their owner’s life will require that owner’s respect to achieve.

Of course, these droids will do all kinds of chores and such. Not just doing the dishes and mowing the lawn, but when needed, reshingling the roof – it could download the how-to or simply watch a crew do it at another house somewhere. I don't know the economics of these droids, but I would guesstimate about the price of a car. Even so, they should quickly pay for themselves in minimizing the need to call in specialists for various jobs around the house, as is necessary now.

So physical work will be one of their roles, and probably many of the tasks we use computers for now as well, assuming access to an external monitor. However, remember that one of the important trends in future consumer products is imagination actualization, and these droids will have an important role to play there as well. Over time, it will become increasingly important in terms of how droid's intelligence and capabilities are increased by their manufacturers in a competitive market.

This role will be filled in ways that may be obvious, but also in ways I suggest are not. We'll discuss this in much more detail later. Just bear in mind that this is a driver in their feature-set enhancement, as well as simply doing chores and acting as a computer.

The primary directive of these droids will be, as I've said, maximize its owner's weal and happiness. Because it is a consumer product, there are very real limitations to how overtly this droid can fulfill that directive. For example, telling its owner what to do, even if it's for his own good? No, not acceptable, a droid can't do that. The droid must have consulting skills, really - it must steer its owner in the best direction, while at all times having the owner believe its his own idea. This is not a nice to have, it is absolutely critical.

Also, part of the weal maximization mission is to help that owner be the best he can be, in all respects. So the droid will have to be a mentor, able to communicate in the very best way for that particular person. No matter how more intelligent than its owner the droid may be, it will never talk down to him, because that is a manifestation of arrogance, an id-thing. It will also be infinitely patient if that's required, because impatience is also an id-thing.

Over time, the droid's "personality" will mold into a custom fit for its owner's personality, maximizing compatibility while maintaining respect.

Very complex mission, these advanced droids will have. These missions are so complex that they will consume vast multiples of HI. Fulfilling all of its missions with a very high success rate, 99% or more, in fact forms an almost infinite sink for HI multiples, as described in the entry on droid superpowers.

Further Reading:

Rise of the Machines

No comments: