Thanks for the link James. I might be able to work this into my study (the guillotine).
The Brian Smith book tries to drill down what counts as a "judgment" in AI. I feel like he didn't really find the answers, but a lot of his argument rests on embedding -- the more common sense a machine acquires by embeddedness, the more effective its judgment. Short of that the AI is stuck in a mere "reckoning" purgatory. I'm trying to put together an event which covers some of the traditional decision making techniques, like satisficing, to more contemporary, like perhaps "LLM as a Judge." The best definition I've ever heard about "objective" is a simple test. If something is truly unique to your beliefs it is not objective. If everybody holds about the same opinion it's probably objective. It seems like practically that's as close as you can get. A true judgment might not be unique to an AGI. On Thu, Jan 25, 2024 at 8:05 AM James Bowery <[email protected]> wrote: > > > On Wed, Jan 24, 2024 at 3:32 PM Mike Archbold <[email protected]> wrote: > >> James, >> >> Thanks for the lead. I know the general nature of AIXI but haven't read >> the paper. Basically what you are arguing, I think, is that everything done >> by a machine is a judgment, since ultimately it's only subjective. >> > > I provided two senses of "subjective" and #1 contains within it the only > notion of "objective" that may reasonably be applied to anything -- and it > corresponds to Solomonoff Induction. So if you want to argue that you are > doing something "objective" it must comport with that notion as formalized > within AIXI. > > >> So, we cannot readily distinguish "fact" from "judgment" in a machine >> > > Yes we can in the sense of Hume's Guillotine and "objective" as described > in: > > https://github.com/jabowery/HumesGuillotine > > and the point is argued by Brian Smith in "The Promise of AI Reckoning and >> Judgment." >> > > I don't know who Brian Smith is. I strongly suspect he's "in the weeds" I > originally described. > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T5edfab21647324f7-M9eabdcce266f9d1654074c6c> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5edfab21647324f7-Me5e4e6bfa1954ff2120f020b Delivery options: https://agi.topicbox.com/groups/agi/subscription
