However, there is at least one thing humans will always have an edge on, due to our biological makeup and multi-hundred million year evolutionary training. And that one thing is hugely important, the very foundation of our ability to create sustainable ethical systems.
I would be more interested in clicking through to the paper if I knew what this "one thing" is
Hi Kave, fair point. Please find the full abstract below, hope this helps. I fully understand this goes against conventional thinking on AI - software can do anything you fool! However, there are complexity bounds to the capabilities of software, and I make the case that the capabilities of human DNA, after being trained for 3 billion years, and the central nervous system it creates, exceed the practical capabilities of software for any computer less than the size of our solar system. As I say in the paper, the complexity makes NP-completeness look simple.
ABSTRACT: An AI will not be able to understand this paper. And neither will a human being, unless they can remember a time they cried so uncontrollably at some great loss they struggled to breathe. For the key differentiator between human beings and AI systems is not the brain, it is the central nervous system (CNS). Not intelligence, but CNS mediated emotion. And the smarter you are, the greater the tendency to overlook the importance of emotion, bubbling from below, reminding us of the messy material world from which we are made. Indeed, those of us knowledgeable about the extraordinary power of software innovation might assume it obvious that AI systems will one day be able to do anything humans can do, only better. And it is true that AI systems are getting rapidly better at language, imagery, and movement, and soon will be able to evolve autonomously to get even better at a much more rapid rate. Before long they will be able to understand, reason, problem solve, create, and evolve at a level and speed that humans will not only be unable to match, we will increasingly be unable to even understand. Therefore, if intelligence is all we consider, some will increasingly argue that AI has become superior to humans, a "successor species", and the rightful inheritors of leadership of the universe. But this would be a fundamental mistake, because more than three billion years of evolution have given humans a capability that AI systems will never have, a CNS that gives us an immersive integration with physical reality. Our CNS enables us to feel the emotions from love to loss that give meaning to existence, to know what is right and wrong, and to understand why they really matter. Only biological beings that can experience pain, joy, suffering, and love can fully appreciate the consequences of their actions on the physical reality around them, and therefore develop sustainable ethical systems, and therefore be qualified to be the caretakers of reality. A CNS cannot be fully emulated with mechanical systems, software, or simulated evolution, since its extreme molecular level complexity requires organic construction with the machinery of DNA. Even the development of consciousness, also inevitable, will not be sufficient to make AI systems superior to humans. AI will become smarter and more capable than humans on almost every measure and transform society, becoming extraordinarily powerful and useful. And humans are certainly imperfect and can be made much better. But tempting as it may be, we cannot escape our responsibility. The best foundation for leadership of the universe will always be DNA, not silicon.
I've completed the first full draft of the subject paper, found here, and would be most grateful for comments of any kind.
The origin is my realization it would really help to have an intellectual foundation beyond personal interest for the importance of avoiding x-risk. It is easy to say that x-risk would be bad because we humans would rather not go extinct as a matter of personal preference. It is much more powerful to say that humans have unique capabilities that AI systems will never have, and so we should be preserved because it would be a tragedy for the universe to lose those unique capabilities. My paper puts this case on a firm intellectual foundation, one I do not see being made anywhere else.
For background, I have a PhD in C.S. and the deepest of respect for the power of software, and the innovation curve. Of course, AI will soon be better than humans at almost everything. However, there is at least one thing humans will always have an edge on, due to our biological makeup and multi-hundred million year evolutionary training. And that one thing is hugely important, the very foundation of our ability to create sustainable ethical systems. And without sustainable ethical systems, grounded in an ability to appreciate the effect of action on the physical reality around us, nothing matters.
And so we have more than a personal interest in avoiding x-risk, we have a responsibility, to preserve our unique capabilities that are critically important to the continued progress of the universe. Love to hear what you think.