Eliezer Interview Question Suggestions
Please include your questions here, along with your signature. Thanks.
Would it be a good idea to list the questions that are answered in CFAI, and convert them to a FAQ? Some answers are short enough to be put directly in the FAQ, with a link to the section the answer came from in CFAI. Other answers that are too long to be put directly in the FAQ could be summarized, and the summary can be included in the FAQ with a link to the section of CFAI where the full version of the answer can be found. There are some other questions whose answer answer is in CFAI, but which might not be possible to even summarize for a FAQ, and for these questions it may be worthwhile to list the concepts that must be understood before you can understand the answer, and then list what to read (not necessarily in CFAI) to learn these concepts, then provide a link to the section of CFAI that contains the answer. It may also be worth listing some questions that CFAI doesn't answer, but should be answered somewhere, like "What is the actual definition of Friendliness?", and provide links to attempted answers. This FAQ could be put on a page called SL4 Guide To CFAI. I realize that many of these suggestions might not be worth all the effort.
-- Anonymous
Anonymous, I personally like the idea, since there is benefit in a guide to CFAI, in light of the document's large size, though could you give a few worked-out examples of what you have in mind for such a guide? Best, -- Anand
"What kinds of complexity does LOGI's levels of organisation reduce, and how does it fit into a general strategy of complexity reduction in problem solving (specifically, implementing humane moralities)?" "What additional complexity is introduced by various classes of multilayed systems, compared to a single layed system, in particular LOGI levels?"
These can probably be unified in some way:
"Moralities conflict, how can a Species AI resolve this in a desirable way? What's the question I'm trying to ask here?" "How non-optimal is a Species-AI compared to a Me-AI (one and the same, given the right person) in general? How many different ways can an AI be humanely representative of a species? Why does this matter?"
-- Nick Hay
proto-questions:
"You have expressed several times an admiration for the philosopher Socrates, once going so far as to suggest that our local universe be named Socrates to distinguish it from similar universes within Bayesia. Socrates seems an unusual choice as a principal figure to fix upon, given both that he is known largely through the works of his more famous student, Plato, and that Western Civilization has been more obviously impacted by his philosophical grandchild, Aristotle. What factors lead you to choose Socrates as such a pivotal figure, and what implications does this have on your personal philsophy. Also, do you believe it is significant that Socrates is the only one of the better-known Greco-Roman philosophers to be condemned and driven to death?(Socrates escaped execution for impiety by drinking hemlock)"
"What's up with the Fermi Paradox? Do you have personal values for the Drake Equation? Does the presence of such a large question-mark have implications in other areas of human endeavor?"
"Related to the above, the modern idea of Singularity suggests that the older futurist theories of civilization growth may be outdated, and that a species/civ could in effect explode from a Type-0 society to a Type-3+ society at near-lightspeed(or worse). Does this affect the valuation of extraterrestrial risks in general?"
"Given that you are a general altruist, why aren't you a vegan?"
"Using the Extropians list as a general indicator of transhumanist interests(perhaps a dangerous assumption to make), transhumanists seem to have settled into either fast-singularity AI proponents or slow-singularity proponents of various approaches. Are you aware of any hard-takeoff singularity theories that involve unFriendly AI and/or nonAI approaches? And are any such projects a threat to successful transition?"
"Who is the most dangerous person in the world you are aware of?"
"Who is the most beneficial person in the world you are aware of?"
"As a pragmatist my evaluation of possible actions is ends-oriented, with the only comparison being to my goal-system. Being an Altruist, do you find it difficult to appropriately weight collateral risks against primary benefits?"
"Does the corruption of an information resource constitute a moral failure?"
"Has your estimation of your own reliability increased or decreased as you progress in meta-rationality?"
"Is privacy an acceptable risk? What about Transparency?"
"Do you believe that Zen enlightenment would be useful, assuming it could be attained in a timely manner?"
"Do you believe there is a teaching deeper than information-theory?"
"Do you plan to run ancestor simulations of yourself?"/"Would you consent to an ancestor simulation?"
"Do you believe that the illusion of continuation of self is neccesary or incidental?"
"What do you try to give to your future self?"
"Are there any realistic circumstances that would lead you to abandon Friendly AI research/development?"
"Some people have posited that certain technologies exist beyond a 'complexity barrier' an area beyond which concepts are too complicated for humans or human agencies to manipulate effectively. Do you have an opinion on the existence of a complexity barrier? If so, where do you think it would be?"
"Does the idea of publishing information about yourself or your ideas strike you as a frivolous waste of your time? Or do you see the enabling or enlightening capability of such information as valuable to you or your goals?(or being an altruist, do you see the benefit of others as sufficient enough to outwiegh time spend on your macroscale attempts to benefit others?)"
Is the Fermi Paradox finally off your top 4 list of things you don't understand? (answered in Hard Problems?)
--observer
