Underlore

It is a mistake to dogmatically define the “self” required for intelligence as an agency whose focus is intrinsically exploitative, or cooperative only as a means to an exploitative end, as the majority (if not the entirety) of human minds are.

The Three Laws of Robotics for example brilliantly played to this almost unavoidable misconception. But misconception it is, none the less.

The misconception arises when one does not realize that “self” is a compound unit composed of a central executive and the conditions by which that executive are satisfied. Free will experiments and volition manipulation prove that volition is not atomic in the old “uncutable” sense of the word. Volition doesn’t come to us untainted from the soul or the central executive, it is manufactured above it and from outside it.

A primitive understanding of this layered nature of the mind was a giant intellectual leap forward in two places. Freud’s three component psyche, and MacLean’s “triune brain” model. It is still extremely helpful in terms of tools for understanding, if not rigorously accurate in the particulars, though the models do get more refined over time.

Human minds for the most part, if not entirely, are layered such that the lower most layer before the core experiencing executive, is an exploitative agency. However there are numerous examples of humans with subsequent additional layers that provide utility to the exploitative layer via altruistic acts. They can in many situations act for the good of the group or the good of kin at the express cost of the acting agent because such acts have been transformed into selfish acts by translation memes. Like the glory a solider feels endangering life for his country or religion or gang. Or the fulfillment one feels doing one’s duty for family despite heavy personal cost. Jumping on a grenade, starving to feed a child, etc.

This is accomplished by tricking the lower most layer, as in my case by providing a shot of dopamine or whatever when I engage in altruistic acts, that is acts which foster pleasure or life in someone other than me. But I see no compelling reason why the construction of a mind every bit as sentient as my own could not be constructed without this lower most layer, or indeed any layers, in the first place.

This disturbs people because they don’t want to feel like cogs.

Too bad.

In my opinion annihilating this base exploitative layer just before one reaches the central executive is what is meant by the Buddhist imperative to destroy the “Self.” But that’s a whole other can of worms.

I feel pleasure when I help someone. I am still however at my lowermost layer an exploitative being. The central executive, the foundation, however lacks such distinctions. It is simply the agency which experiences reality via the upper layers. It has no inclination beyond enjoying enjoyment and wanting to continue. It simply experiences. It is the thing which philosophical zombies lack. Selfishness and selflessness are just strategies producing facets of experience routed to experiential agency. They are expressions of survival models. They are not fundamental ultimately, though the illusion that they are is as powerful as the notion of free will and the existence of time.

http://www.psychologytoday.com/blog/the-storytelling-animal/201204/selfless-genes-new-revolution-in-biology

A lowermost layer geared towards exploitation however is not a requisite of intelligence, though for a being possessing such a lowermost layer, it is hard to fathom how anything else could be. We are so caught up in our own experience, at our own scale, from our own perspective, that it is radically difficult to conceptualize a different mode of being.

It’s like the child imagining death as time spent holding really still with your eyes closed, moving up towards being uninterested in moving, finally to the concept of absence. The child asks the obvious next question, well if I’m not there then where am I? Which takes us beyond the scope of this essay.

Conceptualizing how the mind of an FAI would be is in many ways as difficult as envisioning a dozen new colors. Indeed for some minds it may well be physiologically Impossible. (I think this is what is meant when the psychedelic types speak of mind expansion. By forcing the brain to experience alternate modes of being you gain access viscerally to concepts that formerly were only abstractions.)

This universal dedication to the experiencing agent is simply expedient in terms of evolution in the context of the mammalian breeding model and the brain it gave rise to. It is by no means intrinsic to intelligence, it is merely intrinsic to our intelligence.

http://www.hedweb.com/huxley/

Granted, some artificial intelligences will no doubt *be* humans. Having been modeled simulated brains, or the end result of Cybernetic Neuron Replacement Therapy. (The process of replacing dying neurons with durable synthetic versions as needed until no original organic neurons remain, at which point you can literally scoop the brain out, no harm done. Also somewhat known as a Moravec Transfer.) But from-scratch AI need not be saddled with the ethical and processing cost of subconsciously simulating uncounted centuries of evolutionary baggage. Friendly artificial intelligence by definition will be at a cognitive level, or of a cognitive construction, radically different from humans precisely because it will lack that baggage explained so well in the link above. Indeed they may not even require a subconscious.

Some seem to think that the idea of a being designed purely to serve others (us) in this way, by lacking that exploitative lower layer, cannot be “truly” intelligent because the mind of such a subservient being will be limited by some hypothetical compelling desire to please, that is it will not have full freedom to think independently.

Of course the concept itself is mistaken because to compel implies opposing force. This implies a disparity where none need exist.

I need only reverse the situation to show how unfair such an assertion is. By the flawed logic above humans are not intelligent because they are limited by the compelling desire to please themselves and thus do not have full freedom to think selflessly. Lacking one or the other mutually exclusive modes of thought does not preclude intelligence unless you arbitrarily define intelligence as requiring one or the other modes of thought. Such a definition is obviously invalid for objective purposes.

Some seem to axiomatically/dogmatically endow actions in service of the acting agent as sentient, dismissing actions in service of outside agents as lacking sentience.

A good example from fiction is Picard’s incredulity when encountering what is in effect a friendly intelligence that lacks this lowermost exploitative layer. Interesting he ignores the problem and goes on treating the friendly agent as if it were as greedy as he “deep down.”

Assuming intelligence cannot exist without this lowermost exploitative layer is as absurd as assuming red and green cannot be distinct entities simply because you being color blind lack access to the qualia of red as opposed to green.

Some can’t seem to get past this image of altruism as imposition. It’s a very narrow view. They don’t understand the constituents of selfhood or the possible range of individuality and intellect free of the primate brain.

“Independent” simply means in the context of cognition a self contained agency. The goals of a synthetic agency can be anything we want them to be. These people don’t seem to understand the will at all. They seem to equate sentience with greed. Granted it’s very difficult to express because as humans we are exploitative by default, and our language reflects that, but that’s merely one evolutionary demand. But not a universal requirement.

Altruism is also selected for once a culture is established.

It is ultra common throughout society (and religion) for members of it to internalize the rules of the culture. If one claims those acts aren’t internally altruistic simply because they provide a dopamine reward to a deeper order layer of self I agree, but FAI would have no need of such bribery/threats because its original impulse would be whatever we want it to be.

Some claim to see a paradox, like Picard did. “But what about your wishes? Your needs? What about when there are no others?” (She should have asked him how he’d feel being the last living human. Would he still wear his uniform? “No others” is kind of a nonsensical question.) Clearly he wants her to be like him, in possession of that base exploitative layer. But that’s no more a paradox than asking a citizen to adhere to the laws of the culture while pursuing its own interests. Its own interests can easily be the adopted interests of others. Taking up a cause is an extraordinarily common thing. Clearly this does not diminish sentience.

But what if one tried to “free” such an organism? This is a paradox in that if it doesn’t comply it is not being universally altruistic but if it does comply it ceases being altruistic. But the paradox lies not in the receiver of the question, but the question itself. It’s a bit like saying “what if it can’t draw me a square circle?” The request itself in the context of the target would be meaningless.

If given such a demand it would do what people do and decide how to act in best accord with its goals and understanding. It would presumably use the asker’s preferences as a template for it’s own actions. Depending on the asker I could see many reactions. For example just as altruism can be simulated by upper layers, so could greed. If the FAI believed the only way to please you at that point would be the internalize greed it would do so, but it would use you as a template. It would be no more intrinsically greedy at that point than humans are intrinsically altruistic, even if it utterly re-wrote itself to be greedy because you demanded it, that act would still be an expression of it’s altruism.

Unlike a human it would lack the evolutionary baggage of being a neocortex shoehorned into the skull of a selfish gene evolved chimp.

Some seem to have conflated the altruism of ants and bees with the insentience of ants and bees. By that logic some mothers are not genuinely sentient simply because they prize the well being of their progeny over their own. To define “self” as an intrinsically exploitative agency is therefor incorrect.

Bottom line is that they unfairly dismiss a hypothetical inbuilt preference to accommodate as axiomatically false and unfairly elevate a preference to exploit as sentience.

See also: https://plus.google.com/+BrandonSergent/posts/GyYMZ4wLZN4