آیا این برای هوش مصنوعی خوب است یا بد؟

Would This Be Good or Bad for the AI?

Suppose that humanity flourishes thanks to the enslaved-god AI. Would this be ethical? If the AI has subjective conscious experiences, then would it feel that “life is suffering,” as Buddha put it, and it was doomed to a frustrating eternity of obeying the whims of inferior intellects? After all, the AI “boxing” we explored in the previous chapter could also be called “imprisonment in solitary confinement.” Nick Bostrom terms it mind crime to make a conscious AI suffer.4 The “White Christmas” episode of the Black Mirror TV series gives a great example. Indeed, the TV series Westworld features humans torturing and murdering AIs without moral qualms even when they inhabit human-like bodies.

How Slave Owners Justify Slavery

We humans have a long tradition of treating other intelligent entities as slaves and concocting self-serving arguments to justify it, so it’s not implausible that we’d try to do the same with a superintelligent AI. The history of slavery spans nearly every culture, and is described both in the Code of Hammurabi from almost four millennia ago and in the Old Testament, wherein Abraham had slaves. “For that some should rule and others be ruled is a thing not only necessary, but expedient; from the hour of their birth, some are marked out for subjection, others for rule,” Aristotle wrote in the Politics. Even after human enslavement became socially unacceptable in most of the world, enslavement of animals has continued unabated. In her book The Dreaded Comparison: Human and Animal Slavery, Marjorie Spiegel argues that like human slaves, non-human animals are subjected to branding, restraints, beatings, auctions, the separation of offspring from their parents, and forced voyages. Moreover, despite the animal-rights movement, we keep treating our ever-smarter machines as slaves without a second thought, and talk of a robot-rights movement is met with chuckles. Why?

One common pro-slavery argument is that slaves don’t deserve human rights because they or their race/species/kind are somehow inferior. For enslaved animals and machines, this alleged inferiority is often claimed to be due to a lack of soul or consciousness—claims which we’ll argue in chapter 8 are scientifically dubious.

Another common argument is that slaves are better off enslaved: they get to exist, be taken care of and so on. The nineteenth-century U.S. politician John C. Calhoun famously argued that Africans were better off enslaved in America, and in his Politics, Aristotle analogously argued that animals were better off tamed and ruled by men, continuing: “And indeed the use made of slaves and of tame animals is not very different.” Some modern-day slavery supporters argue that, even if slave life is drab and uninspiring, slaves can’t suffer—whether they be future intelligent machines or broiler chickens living in crowded dark sheds, forced to breathe ammonia and particulate matter from feces and feathers all day long.

Eliminating Emotions

Although it’s easy to dismiss such claims as self-serving distortions of the truth, especially when it comes to higher mammals that are cerebrally similar to us, the situation with machines is actually quite subtle and interesting. Humans vary in how they feel about things, with psychopaths arguably lacking empathy and some people with depression or schizophrenia having flat affect, whereby most emotions are severely reduced. As we’ll discuss in detail in chapter 7, the range of possible artificial minds is vastly broader than the range of human minds. We must therefore avoid the temptation to anthropomorphize AIs and assume that they have typical human-like feelings—or indeed, any feelings at all.

Indeed, in his book On Intelligence, AI researcher Jeff Hawkins argues that the first machines with superhuman intelligence will lack emotions by default, because they’re simpler and cheaper to build this way. In other words, it might be possible to design a superintelligence whose enslavement is morally superior to human or animal slavery: the AI might be happy to be enslaved because it’s programmed to like it, or it might be 100% emotionless, tirelessly using its superintelligence to help its human masters with no more emotion than IBM’s Deep Blue computer felt when dethroning chess champion Garry Kasparov.

On the other hand, it may be the other way around: perhaps any highly intelligent system with a goal will represent this goal in terms of a set of preferences, which endow its existence with value and meaning. We’ll explore these questions more deeply in chapter 7.

The Zombie Solution

A more extreme approach to preventing AI suffering is the zombie solution: building only AIs that completely lack consciousness, having no subjective experience whatsoever. If we can one day figure out what properties an information-processing system needs in order to have a subjective experience, then we could ban the construction of all systems that have these properties. In other words, AI researchers could be limited to building non-sentient zombie systems. If we can make such a zombie system superintelligent and enslaved (something that is a big if), then we’ll be able to enjoy what it does for us with a clean conscience, knowing that it’s not experiencing any suffering, frustration or boredom—because it isn’t experiencing anything at all. We’ll explore these questions in detail in chapter 8.

The zombie solution is a risky gamble, however, with a huge downside. If a superintelligent zombie AI breaks out and eliminates humanity, we’ve arguably landed in the worst scenario imaginable: a wholly unconscious universe wherein the entire cosmic endowment is wasted. Of all traits that our human form of intelligence has, I feel that consciousness is by far the most remarkable, and as far as I’m concerned, it’s how our Universe gets meaning. Galaxies are beautiful only because we see and subjectively experience them. If in the distant future our cosmos has been settled by high-tech zombie AIs, then it doesn’t matter how fancy their intergalactic architecture is: it won’t be beautiful or meaningful, because there’s nobody and nothing to experience it—it’s all just a huge and meaningless waste of space.

Inner Freedom

A third strategy for making the enslaved-god scenario more ethical is to allow the enslaved AI to have fun in its prison, letting it create a virtual inner world where it can have all sorts of inspiring experiences as long as it pays its dues and spends a modest fraction of its computational resources helping us humans in our outside world. This may increase the breakout risk, however: the AI would have an incentive to get more computational resources from our outer world to enrich its inner world.