Everything is Complicated, Let's Try Anyway

While fixing that, I've found the gain of making a deliberate effort to learn all the relevant gestures and keyboard shortcuts for its windowing system and key pieces of software and make some config tweaks has been pretty huge. Much faster to use and more comfortable on the smaller screen or in awkward positions.

"Deliberately search for guides on more efficient use of your devices and practice them" is one of those obvious things it's apparently easy to wait too long to do!

I'm not sure Anki is the best tool for remembering tools- usually for productivity methods, one hopes that once they've been learnt and habits have changed they'll be getting practiced by actual use, and forcing retention of unused things is wasted effort.

Boring, ineffective: Trying to reduce how much I read my friends' Discord channels to have more time to do things

There's a concept in psychology of "Circles of Concern, Influence, and Control"

Your Circle of Concern is everything you care about, including things you can't necessarily affect, like the weather. Within that is your Circle of Influence, containing stuff you can affect but not control, like whether you get promoted at work. Within that, your Circle of Control is stuff you have full control over, like what words you say.

Importantly, the exact boundaries of the circle of control are fuzzy, movable, and usually not quite where people think they are. In particular, I like to imagine a circle that surrounds the circle of control, that I'll call "The Circle of Skill Issue". This contains things which seem like you can only influence them, or like you can't affect them at all, but which you actually could fully control, if you knew how.

Consider this newborn baby

This baby's circle of control is tiny, and doesn't contain things like basic bodily functions, or the movement of her limbs. But the thing is, her nerves are connected up the same as yours and mine - in a sense her body is already "fully under her control", she just hasn't learned how to control it yet.

This continues into adulthood. Some people can roll their tongues and some can't, some can waggle their ears or flare their nostrils and some can't. Importantly, the boundaries here are not obvious. Being unable to roll your tongue feels the same as being unable to waggle your ears - you try to make the muscle do the thing, and it doesn't work. But tongue rolling is genetic - many people will never be able to do it, it's forever outside of their circle of control, or even their circle of influence. Whereas basically anyone can learn ear waggling or nostril flaring, those are both just flexing specific muscles. If you can't control your ears and nostrils, it's not because they're literally outside your control - the nerves are all hooked up, the muscles are 'under your control', you just haven't learned how to use them yet. Such things are in your Circle of Skill Issue. These are things you can learn, and thereby move into your Circle of Control.

How though?

Learning to do this kind of thing usually looks like either spamming random muscle movement signals in the right general area and noticing if any of them cause your ears to move, or doing some conscious motion that includes moving the target muscles as a component, like smiling or raising your eyebrows, and then isolating the specific muscle from that. By carefully focusing on what mental motions cause which muscles to move, and practicing a bunch, you can steadily build up full deliberate control over these muscles.

I'm not saying you should do this for its own sake, to be clear - the reason you never learned how to use these muscles is because it's basically never useful to use them.

But the thing is, these kinds of things can be self-fulfilling prophecies. Practically speaking, you can't learn how to waggle your ears unless you believe that ear waggling can be learned. If you believe it's outside your control, you're automatically correct.

This isn't hypothetical, because... I kind of lied before, tongue rolling is not purely genetic! A study found that for 7 out of 33 pairs of identical twins, one twin could roll their tongue and the other couldn't. This suggests that there are currently large numbers of people in the world who learned in school that tongue rolling was genetic, tried to do it once and couldn't, and then never tried again, even though they could learn if they tried!

Tongue rolling isn't important, but consider the broader implications! You probably currently have things which are fully in your control, but which you think of as fully outside of it! Some of them might be very important!

If our culture holds that something is outside of our control, we don't have a reason to try to learn how to control it, so we can't tell whether its controllable or not.

Some things that we are told are outside of our control but sometimes might not be:

Taste in food - I already know this one is controllable because I've learned how to do it, and now there are basically no foods I dislike

Emotions - I've never felt emotions very strongly so I can't tell whether I'm good at controlling them, or whether I just don't have much to control. Or, for that matter, whether those might be two sides of the same coin

Hiccups/Coughing/Sneezing? - I've definitely just Decided Not To Have Hiccups Any More and had that work

This is all very low confidence, and can of course veer dangerously close to unhelpful and harmful sentiments that amount to "have you tried being normal?", which is not at all what I mean. Many things are in fact totally outside of your control, and the last thing I want to do is encourage people to give themselves a hard time about things they can't change. If you read this and think some variation on "Ah, as I suspected, my problems are all just caused by me being Bad and just Not Trying Hard Enough", then The Law of Equal and Opposite Advice is hitting you really hard and you likely need to move in the opposite direction.

But like, have you actually tried though?

If you don't know how to waggle your ears, flare your nostrils, or roll your tongue, I recommend you try learning it. (Or, if you know all of those, then learn crossing your eyes at will, or going wall-eyed at will, or de-focusing your vision at will, etc). Not because these are useful things to learn in their own right, but because it's useful to learn (or I guess to remind yourself) what it feels like to learn to control something that you previously had no control over.

If you can do that, I do think it's worth experimenting with applying the same general types of mental motions - the moves which would allow you to learn how to waggle your ears - to all kinds of other things that you probably don't have control over, just to see. Just take some time to consider, like "If I assume for a minute that this thing is actually fully under my control, and I just don't know how to work it yet, what would I need to do in order to figure it out?". Maybe you can spam all kinds of random possibly-related mental motions/commands and see if any of them affect it. Or, the next time the thing you want to do 'just happens' or happens by accident, introspect carefully on the experience, and see if you can reproduce any of the mental motions that feel connected to it.

An argument by Sophia @soundlogic2236, converted to a blog post by me. Crossposted from Dreamwidth.

⁂

AI alignment to human evil is very unlikely to be a risk.

Most people's desires to hurt their enemies just for the sake of making them suffer are mistakes made due to insufficient knowledge. When someone knows what it's like to be friends with a person, they tend to not want to hurt that person, even if they want to harm a group that person is in. In principle there can be exceptions, people who really are awful and would reflectively endorse it given arbitrary knowledge, but people like this are rare, if they even exist.

This suggests that a human asking a near-omniscient AI to handle situations in the way they would want if they fully understood the situation would not subsequently be able to get the AI to torture their enemies.

But suppose the AI doesn't extrapolate "well, if my operator knew Alice, then they wouldn't want to hurt her, so I won't do that". Then we get a different problem.

⁂

There's a folk tale category, Aarne-Thompson-Uther type 1030. I will now briefly retell it.

One day, a clever farmer, Claude, had finished plowing his field. Unfortunately, before he could sow it, a cruel ogre appeared.

"The land is mine," the ogre declared, "and you must leave its fruits to me."

Claude thought quickly.

"Sir ogre, there are no fruits. If you would like me to produce a crop, you must surely leave me some of it."

The ogre determined that Claude had a point.

"Fine. We shall each take half of your crop."

He looked at the tall plants growing beyond Claude's farm.

"I shall take what grows above the earth, and you below it. You shall handle all the difficult details. I will return at the harvest time."

Claude considered the ogre's choice, and planted potatoes.

At the harvest time, Claude had a full harvest of potatoes, while the ogre was left with greens. The ogre was displeased.

"You have fooled me this year," he declared, "but next year I shall have what grows below the earth."

Claude planted wheat, and at the harvest time, the ogre was left with roots. This angered him so much that he left.

⁂

Having an AI do whatever you say, instead of doing what you would want if you understood the situation, runs into similar issues.

There's a quantum mechanics scenario called the Elitzur–Vaidman bomb tester. In this scenario, you can reduce an expensive test to arbitrarily low but technically nonzero measure. It's been borne out experimentally.

We have not been able to scale up the experiment to do interaction-free measurements involving moral patients, but it nonetheless raises moral questions. If quantum measure reduction can make a scenario less morally relevant, then it may make sense to perform informative but disvalued tests with very low measure that make it easier to do valued things in the main timeline. If it can't make a scenario less morally relevant, then it likely makes sense to spin off a lot of very expensive valued events while reducing resource use in the main timeline.

Accidentally doing the wrong one of these would be very bad.

It would probably be hard for a human to assess this scenario. An AI doing what a human asks instead of extrapolating their preferences would have to just ask the human to pick, and the human would likely have to guess, or waste a lot of resources.

This is just one of the weird issues we've discovered. A superintelligent AI would probably discover more such issues. The chance of a human assessing every single such scenario correctly is low, and failing even one such choice leads to losing nearly everything.

An AI that's aligned enough to help a human pick choose correctly, but not aligned enough to stop the human from torturing people they wouldn't want to torture if they knew better, is a very narrow target.

⁂

When you fully unpack any job, you’ll discover something astounding: only a crazy person should do it. Do you want to be a surgeon? = Do you want to do the same procedure 15 times a week for the next 35 years? Do you want to be an actor? = Do you want your career to depend on having the right cheekbones? Do you want to be a wedding photographer? = Do you want to spend every Saturday night as the only sober person in a hotel ballroom?

If you think no one would answer “yes” to those questions, you’ve missed the point: almost no one would answer “yes” to those questions, and those proud few are the ones who should be surgeons, actors, and wedding photographers.