Science Fiction writer Greg Egan has a short story online at http://www.netspace.net.au/~gregegan/MISC/SINGLETON/Singleton.html that deals with some of the ethical implications of the Many Worlds Interpretation of QM. It describes the motivation for the invention of the "Qusp", a device Egan has also used in other stories. A Qusp is a quantum computer suitable for running a mind, but in a special way. It is designed to avoid splitting into multiple decoherent parts as usually happens in quantum evolution. The idea is that if your mind is running on a Qusp, and you make a decision, it only goes one way. You don't decide in different ways in different worlds. This is supposed to be more satisfying for people of a certain philosophical bent.
This story provokes several thoughts which I could expand upon but which I will state only briefly here. The first is the question of whether an appropriate response to a demonstration of the MWI (or the more radical notions of multiple universes we consider here) is a sense of existential despair, as one realizes that any decision and choice in life was made differently by a counterpart in another universe. I don't think this is reasonable. Your decision process should be seen as altering the measure or proportion of universes in which you do one thing or the other. In Egan's story he mostly ignores measure. I think that the proper philosophy is to consider measure as being of crucial importance. Your job in life is to bend the measure of the branches to maximize the welfare of whatever is most important to you. I wrote some time back about how this perspective could play out in some commonplace situations. The example I gave was driving too fast and unsafely down a residential street. A child could have run out in front of you and you wouldn't be able to stop. In fact you didn't hit anyone and you give a sigh of relief. But you should realize that in some universes you weren't so lucky. Rather than thinking in terms of outcomes in particular universes, you need to take a larger view and look at the multiverse as a whole. Understand that your action reduced the measure of the lives of the children who live on that street. Even though you can't see or measure the reduction, it was real and your action did affect the multiverse. You caused actual harm by driving too fast. A related source of existential worries is simply that the MWI is deterministic. Some philosophers have claimed that free will is inconsistent with a deterministic universe, and have seized upon the supposed randomness of QM as offering a place for free will to enter. I don't agree with this because free will based on quantum randomness may be "free" but it is not "will"; you don't choose the outcomes of quantum measurements. But in any case, MWI takes away the illusion of quantum randomness; the evolution of the Schrodinger equation is fully deterministic when we look at the multiverse as a whole. Fortunately there is a pretty substantial philosophical literature on compatibilism, the doctrine that free will can exist in a deterministic universe, so people who are concerned about this can study philosophy for a while. One final comment about Egan's Qusp: it's not clear that it really makes sense, that there is a problem that needs solving and that the Qusp can solve it. Egan is worried about universe splitting. But ordinary, classical computers don't normally split. Give them a certain input and they will always produce the same output. Of course, if the input itself comes from the universe, that is split already, but Egan admits that there is nothing he can do about that. He just wants his computer running his mind not to split any more than what is forced by coupling to the outside environment. So the point is that the Qusp doesn't add any capabilities that an ordinary classical computer won't have. The only way to make a computer which itself splits is to give it a source of quantum randomness, a true random number generator. Now, it is very likely that our own brains work that way. The messy chemistry that is involved with our neural firing is probably chaotic enough that quantum variations get amplified and lead to decoherence. A human brain left to itself would split into multiple decoherent branches, I strongly suspect. But I doubt that splitting is fundamental to the brain's operation. We could simulate a brain on a classical computer and it would no longer split, but it would still be just as functional. So this is all Egan would have to do for his Qusp, and it seems to me that the non-Qusp AIs would probably work that way. I don't see why they would go out of their way to include a source of quantum randomness in the other AIs, and that would be the only way to make them split. More fundamentally, though, I think that splitting due to the environment, which Egan admits can't be prevented, will still be more than enough to raise the same existential issues he was trying to solve with the Qusp. You will still decide things both ways, because tiny environmental differences will get amplified. I don't see that running on a Qusp will really give you the kind of assurance he is hoping to get. Anyway, it is an interesting story, in the first half at least. And it is good to see some thoughtful discussion of the philosophical issues raised by the MWI. Hal Finney