Very interesting topic.

I think the way you treat something has to be determined by its capacity to suffer. When it comes to artificial intelligence, we tend to talk about what a robot can "do" (i.e., show creativity, show emotion). This is the basis of the Turing Test.

Say a robot is good enough to fool everyone, even for long periods. But then you bring out a team of programmers who can explain the "tricks". If the robot only cries because they've programmed it to do so, using a algorithm to identify what people are likely to find sad - with a defensive subroutine to say something like "I don't want to talk about it" or "I'm just feeling hormonal" if it misfires and starts crying at a false positive - then there's no harm in making the robot cry. It's just an accurate simulation of suffering, rather than real suffering.

We may understand how the human brain has evolved and some of how it works, but we don't understand how it can give rise to the subjective experience of existing - the awareness that it has of its own existence. In terms of natural selection, the human brain would be just as "fit" if it did exactly what it does do, but completely on autopilot - just taking input from the senses, bouncing it around, and producing responses in a purely mechanical way (like our crying robot, or the Chinese Room for that matter).

But if my own brain is anything to go by, it doesn't. It is actually aware of what it's doing (to an extent!). It knows it is here. Cogito ergo sum.

Personally, I think a robot of the future could have the same experience, provided it was based on a neural network (a proper synthetic brain), rather than a series of programmer-mandated rules that help it produce human-seeming behaviour. A true artificial intelligence would have to start out like a baby and develop a personality over time, in response to experience, with just a small number of hard-coded rules about stuff like trial and feedback.

If a robot produced strikingly human-seeming behaviour that had arisen over time from a synthetic neural network, and produced signs of suffering without being directly trained or programmed to do so, then I would err on the side of assuming that it could suffer.

That's my two penn'orth anyway

I agree with subjective experience comment.
Subjective experience, or "qualia" (sp?) is evidently a mind/body connection.

Consider a beautiful robot companion and I looking at a beautiful sunset.
My subjective experience gives the sunset meaning. My robot lover (I'll take Rachel of Blade Runner) would not have a subjective moment. Nor could she share a true subjective reaction to it. She could define it only from a database of symbols, or appropriate linguistic responses she'd calculate as appropriate for the moment.

Could I have the subjective experience of what it's like to be a cat? A dog? A bat? A rat?
Could my female companion, in turn, surpass my inability to not know what it's like to be a cat, by knowing what it's like to actually be a woman? Then we are now talking about having created something greater than us. That's really beyond even discussing AI.

While we are at it, why don't we talk about zombies. They are more or less fully functional but they too would lack a subjective consciousness. Are zombies entitled to "rights?"
We are assigning rights to objects that can not have subjective experience.
Where does it end?

The smartest computers we have now are not even close to the level of a rat's brain, let alone, say, a chimpanzee.

The reverse is also true. The smartest humans alive are not even close to the level of ability computers have in certain areas.

Can AI be considered sentient even if it lacked emotion? Could an AI be considered sentient even if it did not experience suffering? These two characteristics seem to be byproducts of an evolved intelligence. But a crafted intelligence does not need these capacities for behavioral control. I would wonder at the morality of crafting an intelligence that was capable of suffering, especially if there were other means to arrive at the end result.

The painting fool.

http://www.thepaintingfool.com/about/index.html



This is a computer system that someone is trying to get to paint creatively.

What this amounts to is machine learning.

http://en.wikipedia.org/wiki/Machine_learning

whereby to teach a computer what the letter "A" is, and recognize it in all it's forms, they show the system hundreds, thousands of examples of the letter "A" in different fonts and as written by numerous hands. The computer builds a model of the structures which comprise "A" and thereafter will continue to be able to recognize that symbol even when done in a way which it has not previously observed.

In a lot of ways, this is how our own brains work. You learn to recognize patterns by seeing them repeated and building a causal relationship

If this programmer succeeds, and you might say that he has already done so to a degree, in getting TPF to paint scenes from it's own "imagination", rather than from a digital image, does that satisfy your criteria for what creativity is?

If not, how would this example differ from our own creativity?

These two characteristics seem to be byproducts of an evolved intelligence.

On what basis?

On the basis that a connotation of the definition of intelligence doesn't require them. I would ask for reasoning as to why these characteristics are required in any definition of intelligence. I don't see that they are. It would be alien intelligence, but intelligence all the same.

The smartest humans alive are not even close to the level of ability computers have in certain areas.

Define "smart"

He's talking about directly comparable abilities.

The most accomplished human at math will perform more slowly than powerful number-crunching machines.

There are other comparisons.

http://io9.com/5886178/why-mass-effect- ... generation

Sci-fi universes let us explore interesting questions. in the above link someone makes the case for "mass effect", a sci-fi video game universe, where many interesting social questions and human suppositions can be questioned.

One angle on thie whole thread is the treatment of "the other". You could substitute sentient AI for clones, or synthetic life, or aliens, and the questions are largly the same.

how do we treat the other? Why do we insist on human privilage? Mass effect sets up a world that subverts our assumptions of human privilage, and the consequence is very interesting subtle social commentary.

From the link:

Mass Effect is the first blockbuster franchise in the postmodern era to directly confront a godless, meaningless universe indifferent to humanity. Amid the entertaining game play, the interspecies romance, and entertaining characters, cosmological questions about the value of existence influence every decision. The game is about justifying survival, not of mere intelligent life in the universe, the Reapers are that, but of a kind of intelligence. Therein the triple layered question – What value does galactic civilization bring to the universe; What value does humanity bring to galactic civilization, and What value do I bring to humanity – forces the player to recontextualize his or her participation in the experiment of existence.

Michio Kaku has been saying for a while now that moore's law is doomed to fall apart soon due to quantum uncertainty.

But these guys seem to think otherwise.

http://www.abc.net.au/news/2012-02-20/t ... or/3839524

Atomic transistors.