= 27d567bc2d48d9f40e9ccc20ea370b15)
Has anyone managed not to Bottom Line in their everyday thinking? I find that it's very difficult. It's so natural and it's a shortcut that I find useful more often than harmful. I wonder if it's best to flag issues where epistemic irrationality would be very bad and primarily focus on avoiding Bottom Lining at times like that. I feel that the things I'm talking about are in a different spirit than those originally intended by the article, where you're not so much emotionally invested in the world being a certain way as you are, say, relying on your intuition as the primary source of evidence for the sake of saving time and avoiding false starts.
In response to
Rationality Reading Group: Part G: Against Rationalization
Well, that Bottom Line is generated by your intuition, and your intuition is probably pretty good at weighing evidence to find a good Bottom Line (with the caveat that your intuition probably does a lot more signalling than truth-seeking). In principle, though, that means that you don't have to justify that bottom line, at least not as much; instead, it would be more productive to search for flaws. The most likely flaw is that your subconcious figured "My friends like this conclusion" rather than "This conclusion is true".
Follow your priors. The problem here is that the prior for Hell has been constructed, "artificially", to have unnaturally high probability.
I think my claim was that your example was kinda bad, since it's not obvious that the AI is doing anything wrong, but on reflection I realized that it doesn't really matter, since I can easily generate a better example.
In response to
Versions of AIXI can be arbitrarily stupid
Do you have a better proposal for how to act if you know that you are in either heaven or hell, but don't know which?
In response to
comment
by
on
(Rational) website design and cognitive aesthetics generally- why no uptake?
I think he actually means rational. He advocates to make design decisions after reading academic papers on the subject.
Whether or not that's optimal is a separate issue.
In response to
comment
by
on
(Rational) website design and cognitive aesthetics generally- why no uptake?
The rational part is reading the papers; the optimal part is doing what the papers say.
However, reading papers is not limited to design. It is part of general rationality (virtue of scholarship).
Of course, if you search for papers on cognitive biases in web design, then they would tell you about rational design.
In response to
Stupid Questions July 2015
How do you reconcile being transgender with the fact that a lot of our sexual roles are culture-specific? For instance, imagine a MTF who wants to wear a dress. You can't tell this person "stop wearing dresses"; their desire to do so cannot be changed by society telling them no. Yet if they lived in another culture that didn't have dresses at all even for women, we know that when society told them not to wear a dress they would have eagerly gone along with it.
An obvious hypothesis is that trans women follow culture-specific roles (such as wearing dresses) for the same reason cis women follow culture-specific roles.
This would mean that whatever makes us follow culture-specific roles isn't extremely stupid, so it interprets 'women should wear dresses' correctly, rather than 'you shouldn't wear dresses'.
Isn't that only a subcase of friendliness testing?
Not really. If you have an AI where you're not sure if it is completely broken or just unfriendly, you might want to test it, but without proper boxing you still risk destroying the world in the unlikely case that the AI works.
In response to
On the Boxing of AIs
Hmm, if information is still supposed to be able to go to the gatekeeper then the talking person is effectively a gatekeeper too. However it's not a bad idea to make a box within a box. I got stuck on thinking why we need the gatekeeper to receive information, but in the end it's pretty simple. We need to accept the friendly AI. But then you could say that how can you trust someone to be your ally without gaining information about their values? This might be an oxymoron ie that by definition you can't. Thus the box should be anything but black and as white as other considerations allow.
There are other reasons than to check if the AI is friendly. AI, like other software, would have to be tested pretty thoroughly. It would be hard to make an AI if we can't test it without destroying the world.
In response to
On the Boxing of AIs
By letting it talk to the Gatekeeper.
I think you missed a "not" there.
= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d){kind=logo}
The way I see it, having intuitions and trusting them is not necessarily harmful. But you should actually recognize them by what they are: snap judgements made by subconscious heuristics that have little to do with actual arguments you come up with. That way, you can take it as a kind of evidence/argument, instead of a Bottom Line - like an opinion from a supposed expert which tells you the "X is Y", but doesn't have the time to explain. You can then ask: "is this guy really an expert?" and "do other arguments/evidence outweight the expert's opinion?"
Note that both for experts and for your intuition, you should consider that you might end up double-counting the evidence if you treat them as independent of the evidence you have found - if everybody is doing everything correctly (which very rarily happens), you, your intuition and the experts should all know the same arguments, and naive thinking might double/triple-count the arguments.