One hope I have in this vein is that human genes don’t contain any “metaphilosophical secret sauce” (instead all the secret sauce is in the culture) so we are able to build a competent philosopher just by doing (something like) fine-tuning GPT-n with a bunch of philosophy papers and/or human feedback. Then we use the artificial (black box) philosopher as part of an aligned AI or to help solve alignment problems.
Unfortunately, I expect that even in the scenario where this ends up working, the artificial philosophers will probably end up writing thousands of increasingly hard-to-follow papers on each philosophical problem exploring all the possible arguments/counterarguments, before reaching some consensus among themselves, and because we won’t have a white-box understanding of metaphilosophy, we will just have to hope that they learned to do philosophy the “right way” whatever that actually is.
I think philosophy is basically either conceptual analysis to turn an unclear question into a well-defined empirical / mathematical one or normative reasoning about what we ought to do, feel or believe. I’ve developed and programmed a formal theory of metasemantics and metaethics that can explain how to ideally do those. I apply them to construct an ethical goal function for AI. It would take some more work to figure out the details but I think together they also provide the necessary resources to solve metaphilosophy.
One hope I have in this vein is that human genes don’t contain any “metaphilosophical secret sauce” (instead all the secret sauce is in the culture) so we are able to build a competent philosopher just by doing (something like) fine-tuning GPT-n with a bunch of philosophy papers and/or human feedback. Then we use the artificial (black box) philosopher as part of an aligned AI or to help solve alignment problems.
Unfortunately, I expect that even in the scenario where this ends up working, the artificial philosophers will probably end up writing thousands of increasingly hard-to-follow papers on each philosophical problem exploring all the possible arguments/counterarguments, before reaching some consensus among themselves, and because we won’t have a white-box understanding of metaphilosophy, we will just have to hope that they learned to do philosophy the “right way” whatever that actually is.
I think philosophy is basically either conceptual analysis to turn an unclear question into a well-defined empirical / mathematical one or normative reasoning about what we ought to do, feel or believe. I’ve developed and programmed a formal theory of metasemantics and metaethics that can explain how to ideally do those. I apply them to construct an ethical goal function for AI. It would take some more work to figure out the details but I think together they also provide the necessary resources to solve metaphilosophy.