"The biology of suffering in intelligent agents is a deep underlying source of existential risk – and one that can potentially be overcome."

“My own sense of how to behave in a simulation has more traditional roots in the theory of perception. I've long believed that each of us lives in an egocentric simulation of the world run by the mind/brain. Since the zombies of each simulation have sentient real world counterparts, one should treat them as though they were real. Nonetheless as an angst-ridden teenager, my dawning acceptance of an inferential realist theory of perception made me feel as if I'd been condemned to solitary confinement for life. The sense of loneliness was indescribable. Naïve realism is better for one's mental health.”

“Suppose we encounter an advanced civilization that has engineered a happy biosphere. Population sizes are controlled by cross-species immunocontraception. Free-living herbivores lead idyllic lives in their wildlife parks. Should we urge the reintroduction of starvation, asphyxiation, disemboweling and being eaten alive by predators? Is their regime of compassionate stewardship of the biosphere best abandoned in favour of "re-wilding"? I suspect the advanced civilization would regard human pleas to restore the old Darwinian regime of "Nature, red in tooth and claw" as callous if not borderline sociopathic.”

“Getting rid of predation isn't a matter of moralising. A python who kills a small human child isn't morally blameworthy. Nor is a lion who hunts and kills a terrified zebra. In both cases, the victim suffers horribly. But the predator lacks the empathetic and mind-reading skills needed to understand the implications of what s/he is doing. Some humans still display a similar deficit. From the perspective of the victim, the moral status or guilty intent of a human or nonhuman predator is irrelevant. Either way, to stand by and watch the snake asphyxiate a child would be almost as morally abhorrent as to kill the child yourself. So why turn this principle on its head with beings of comparable sentience to human infants and toddlers? With power comes complicity.”

“It is easy to romanticise, say, tigers or lions and cats. We admire their magnificent beauty, strength and agility. But we would regard their notional human counterparts as wanton psychopaths of the worst kind.”

“rue hedonic engineering, as distinct from mindless hedonism or reckless personal experimentation, can be profoundly good for our character. Character-building technologies can benefit utilitarians and non-utilitarians alike. Potentially, we can use a convergence of biotech, nanorobotics and information technology to gain control over our emotions and become better human beings, to cultivate the virtues, strength of character, decency, to become kinder, friendlier, more compassionate: to become the type of human beings that we might aspire to be, but aren't, and biologically couldn't be, with the neural machinery of unenriched minds. Given our Darwinian biology, too many forms of admirable behaviour simply aren't rewarding enough for us to practise them consistently: our second-order desires to live better lives as better people are often feeble echoes of our baser passions.”

“If we don't address the genetic causes of suffering we will find ourselves in 500 years enjoying material abundance via nanotech, living in a perfect democracy, colonizing space, and still sitting around wondering "Why are we miserable so much of the time? Why can't we all just get along? Why are we not all happy?”