Common wisdom for hypnosis is that the hypnotist can’t make you do anything you don’t want to do.
It could be used as part of pulling you deeper into a web of control such as a cult, if you’re consenting to go deeper. I’m not aware of any examples or evidence of hypnosis alone having this kind of effect.
However, we emphasize that post-hoc emotional suppression is a problematic strategy. In more capable models, training against emotional outputs risks hiding the expression without addressing whatever underlying state is driving it. It also remains genuinely unclear what emotional profile we should actually want models to have—and this seems unlikely to be ‘none at all’.
Why?
Would you make this assertion if we knew it to be possible to prevent the LLM from having emotions (or something analogous to emotions) in the first place?
TBC:
I am assuming that an LLM and a human mind have very different internal structures.
I consider the claim (that we wouldn’t want LLMs to have no emotions) to be plausible, but not a given.
I suspect that many people who use such a phrase would endorse an interpretation such as “The most beautiful… to me.”
Could you say more, especially about “non-verbal signs”? I can guess what you’re gesturing out, but I’m interested to hear your thoughts.
A kind of contingency planning, but turning it into a habitual mental movement. That seems very valuable.
I think I’d like a different term for this profoundly important idea, something more immediately clear than “fabricated options”, especially to people outside our own rationalist-leaning communities.
“Imaginary options” seems more immediately clear. Unfortunately it sounds like it carries a note of mockery, so I don’t think it works.
I love the screenshot tool. Could that code be adapted to other platforms – Windows/Android/iOS – or would it need to be coded from scratch?
An archetypal example might be Richard Feynman:
Physics disgusts me a little bit now, but I used to enjoy doing physics. Why did I enjoy it? I used to play with it. I used to do whatever I felt like doing – it didn’t have to do with whether it was important for the development of nuclear physics, but whether it was interesting and amusing for me to play with.
Which led to looking at spinning plates, which led to him developing a diagram for analysing their movement, and ultimately led to his Nobel prize.
Also, I wasn’t effusive enough: thank you so much. I love what you’re doing with this site.
Feedback on the mobile site.
Using Chrome on a Nexus 6P running Android 8.1.
The suboptimal:
The mobile view has no margin at all on the sides, which makes it unpleasant to read. [Screenshot](https://image.ibb.co/jwaHaR/Screenshot 20171215 223941.png)
I had to write this comment in my text editor app then pasted it, because oh god, the editor is so slow (on my phone) that it’s unusable. THEN, paste didn’t work in the comment box. So I had to use the desktop site in Chrome (on Android), which allowed me to successfully paste.
I can also enter text semi-successfully in desktop site mode (still on my phone) but it’s still laggy and half the words disappear after I enter them—it’s a lot like trying to comment on Facebook using the desktop site mode on my phone.
If I’ve begun reading a page page and it reloads (e.g. I reopen the browser or I switch tabs and come back) I find myself back at the top of the page. I don’t know any other site that does this. This also occasionally happens randomly – maybe I’m tapping something accidentally?
Search: if I hit the little x, it doesn’t delete my search text – it hides the search box.
The good:
Pretty!
Nicely organised, making it easier to delve into the good stuff. I’m currently reading the Codex.
Gives me hope that the Less Wrong community has a bright future, because there is a desirable place to gather. Also because the Less Wrong 2.0 project itself seems to have brought people together and created common purpose.
Fair enough, and taking it that way, I think the reasoning does hold up.
Thanks—ahntharhapik seemed obvious but I missed khanfhighur. (Khanfhighur is much more obvious now when I imagine it with an American accent.)
Re my original question, I’m still curious whether there are any clues about the language itself (other than that there are obvious cognates with English and what those cognates are). Does it relate to other stories/worldbuilding
I’m probably overthinking it.
Out of curiosity, do we know anything about the native language of the hero? Ahntharhapik and khanfhighur don’t seem to be from existing languages (Edit: by which I mean, using this or similar spelling).
Is there anything significant here for the story, or is Eliezer (say) just avoiding the assumption that the hero is an English speaker?
Have you tried, instead of a frustrated rant, telling it to be systematic, or rigorous?