JGWeissman comments on General purpose intelligence: arguing the Orthogonality thesis - Less Wrong

20 Post author: Stuart_Armstrong 15 May 2012 10:23AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Article Navigation

Comments (156)
Tags: efficiency rationality intelligence motivation orthogonality ai

Comments (156)

You are viewing a single comment's thread. Show more comments above.

Comment author: Wei_Dai 16 May 2012 08:39:57PM 5 points [-]

The primary point of my comment was to argue that an agent that has a goal in the strong sense would not abandon its goal as a result of philosophical consideration.

Such an agent would also not change its decision theory as a result of philosophical consideration, which potentially limits its power.

Philosophy probably has to play some role in clarifying our "pseudo-goals" as actual goals, but so does looking at our "pseudo-goals", however arbitrary they may be.

I wouldn't argue against this as written, but Stuart was claiming that convergence is "very unlikely" which I think is too strong.

Comment author: JGWeissman 16 May 2012 09:01:18PM 2 points [-]

Such an agent would also not change its decision theory as a result of philosophical consideration, which potentially limits its power.

I don't think that follows, or at least the agent could change its decision theory as a result of some consideration, which may or may not be "philosophical". We already have the example that a CDT agent that learns in advance it will face Newcomb's problem could predict it would do better if it switched to TDT.

Comment author: Wei_Dai 16 May 2012 10:15:20PM 2 points [-]

I wrote earlier

"ability to improve decision theory via philosophical reasoning" (as opposed to CDT-AI changing into XDT<CDT> and then being stuck with that)

XDT<CDT> (or in Eliezer's words, "crippled and inelegant form of TDT") is closer to TDT but still worse. For example, XDT<CDT> would fail to acausally control/trade with other agents living before the time of its self-modification, or in other possible worlds.

Comment author: JGWeissman 16 May 2012 11:30:22PM 0 points [-]

Ah, yes, I agree that CDT would modify to XDT<CDT> rather than TDT, though the fact that it self modifies at all shows that goal driven agents can change decision theories because the new decision theory helps it achieve its goal. I do think that it's important to consider how a particular decision theory can decide to self modify, and to design an agent with a decision theory that can self modify in good ways.

Powered by Reddit Powered by Reddit