Meteuphoric

Crossposted from world spirit sock puppet.

I’ve argued that the AI situation is not clearly an ‘arms race’. By which I mean, going fast is not clearly good, even selfishly.

I think this is a hard point to get across. Like, these people are RACING. They say they are RACING. They are GOING FAST. If they stop RACING the other side will get there first. How is it not a RACE??

Which is a fair response.

It’s like if I said “this isn’t a chess tournament” gesturing at a group of chess champions aggressively playing chess. How could it not be?

Well, maybe all the prizes and recognition available in the circumstances are based on winning at checkers. That would make it, in a very important sense, not a chess tournament. They can play chess all they like, but it doesn’t make the incentive structure into that of a chess tournament. If they want to win at a tournament, their strategy is just badly mistaken.

It’s true that many people are trying to build AI very fast. But many people building AI very fast is different from being in a game where going very fast is the best selfish strategic move.

And this becomes important when “it’s really important to win at the race” becomes justification for a) moving fast at very high costs to other people, and b) giving up instead of trying to coordinate other players not to move fast, since other players are presumed to be immovably committed to winning the race due to that being so incentivized.

These justifications both require the structure of incentives to actually be a race, not just for people to be racing.

‘Is AI really an arms race or are people just racing?’ might sound like an abstract question. But if someone is saying they need to risk your family’s lives to fuel their quest to win an extremely high stakes chess championship, it’s very concretely important whether they are really in a chess championship!

While this is a basic point, my guess is that the distinction between what people are doing and what it is in their interests to do is too subtle and non-memorable to be tracked in the conversation.

So I propose an image I think might keep the incentives and the behavior separate more intuitively: AI as a Trojan horse race.

Various groups are working really hard to get various wooden horses through their own gates, resolute on doing so before their enemies pull in such a prize and outclass them with the contents. It’s an open question whether each horse contains fantastic treasure or a bunch of enemy agents. (This time in history we are even pretty confident that it includes a bunch of agents of some sort, and not at all confident of their loyalty..)

Is it enough to know that other cities are pulling horses through their gates? Are you satisfied then to have the biggest one pulled into your own town square?

Crossposted from world spirit sock puppet.

There is a particular flavor of suffering I fear: where something is not just unpleasant, but is requiring active effort from you to continue having the unpleasant thing happen, and so you have to not only suffer the suffering, but also the constant thinking about whether maybe you should stop right now—and so are also having to dip peripherally into questions of free will and will power and who you are and if you will ever do anything and if you are fundamentally bad, and all this while you are already quite taxed by the original suffering.

The epitome of this kind of suffering to my mind has traditionally been running. What everyday activity was less pleasant than running? Better to be lightly tortured by someone else, than have to do the inflicting as well. (No, I’m probably not a very athletic person.)

But that was years ago. These days running is often one of the most joyous things I do.

(I still don’t do it nearly enough, but often when I do I think “oh wow this is so good, I should do this much more often” rather than “can I stop? can I stop? I’m stopping.. no, oh god, when is it over?”)

What changed?

The first thing that happened—which I’d guess is not crucial but did help me get started—was that a person I had a crush on started inviting me to go on runs. This helped me get a tiny bit better at running, because I was willing to withstand almost arbitrary amounts of suffering to spend time with him. This probably got my running skill from “really wants to stop running within about twenty steps” to “can run for a block or two before hating it”. By the time he stopped inviting me (since he actually wanted to run far and fast) I think I still found running basically unpleasant, but had more of an affordance for doing it for non-negligible stretches.

The real change was from running alone and altering my running protocol.

Here is how to enjoy running, in my experience:

1. Get yourself some good running music. This is key. It’s like the difference between having fuel in your vehicle and not. Ideally you want a playlist consisting entirely of songs which if they came on at a party would send you leaping up and scrambling for the dance floor. My first playlist for this was called “corny”, and my most recent one is a variety of 90s pop punk.
2. Put on shoes. Put on music. Start running.
3. As soon as you don’t feel like running—even if it’s after five steps—walk.
4. As soon as you feel like running again, run. This may be because the music hits a bit that demands it, or the street is sloping downwards, or walking just feels a bit slow, or you regained your energy and bounding along in the sun would feel good.
5. As soon as you feel like leaping, or skipping, or balancing on a low wall with your arms out, do that.
6. Repeat steps 3-5 in any order until feeling like running stops occurring ever.
7. Wander home.
8. Repeat another day, and probably find yourself walking a tiny bit less, and enjoying yourself running a tiny bit more.

I guess the crucial elements are:

• a) There’s a huge experiential difference between running when you don’t feel like it and running when you do feel like it.
• b) Music is compelling, and in particular can compel you to move your body enjoyably (most classically observed in the phenomenon ‘dance’).
• c) If a thing is enjoyable at least sometimes, then you can enjoy it 100% of the time you are doing it by just not doing it when you aren’t feeling it.

Some additional modifications that might help:

1. Be cringe. Dance at stoplights. Smile at strangers. Think grandiose thoughts.
2. Use a fitness device where you can watch your heart rate in real time—it’s somewhat compelling to control it by running when it drops relatively low (and that is coincidentally when you may feel like running again).
3. Use a fitness device where you can track general progress in amount of exercise.
4. End up somewhere you can buy a delicious coffee or something.
5. Instead of slowing down as soon as you feel like it, pick a tree a little way further down the road to make it to first.
6. If you aren’t feeling a song, aggressively skip it.

To be clear, I have not become so good at running as to give up walking for large parts of it. But going for a forty minute walk/run in which half of the time you are running and loving it seems like a huge improvement in my life.

I have no idea how well this is likely to work for other people. I might be unusually compelled by music or unusually horrified by using willpower. (I’m also aware there are many people who just naturally enjoy running.) If you try something like this, I’m curious to hear how it goes.

Crossposted from world spirit sock puppet.

Crossposted from world spirit sock puppet.

Under present norms, if Alice associates with Bob, and Bob is considered objectionable in some way, Alice can be blamed for her association, even if there is no sign she was complicit in Bob’s sin.

An interesting upshot is that as soon as you become visibly involved with someone, you are slightly invested in their social standing—when their social stock price rises and falls, yours also wavers.

And if you are automatically bought into every person you notably interact with, this changes your payoffs. You have reason to forward the social success of those you see, and to suppress their public scrutiny.

And so the social world is flooded with mild pressure toward collusion at the expense of the public. By the time I’m near enough to Bob’s side to see his sins, I am a shareholder in their not being mentioned.

And so the people best positioned for calling out vice are auto-bought into it on the way there. Even though the very point of this practice of guilt-by-association seems to be to empower the calling-out of vice—raining punishment on not just the offender but those who wouldn’t shun them. This might be overall worth it (including for reasons not mentioned in this simple model), but it seems worth noticing this countervailing effect.

Prediction: If consortment was less endorsement—if it were commonplace to spend time with your enemies—then it would be more commonplace to publicly report small wrongs.

Crossposted from world spirit sock puppet.

Under present norms, if Alice associates with Bob, and Bob is considered objectionable in some way, Alice can be blamed for her association, even if there is no sign she was complicit in Bob’s sin.

An interesting upshot is that as soon as you become visibly involved with someone, you are slightly invested in their social standing—when their social stock price rises and falls, yours also wavers.

And if you are automatically bought into every person you notably interact with, this changes your payoffs. You have reason to forward the social success of those you see, and to suppress their public scrutiny.

And so the social world is flooded with mild pressure toward collusion at the expense of the public. By the time I’m near enough to Bob’s side to see his sins, I am a shareholder in their not being mentioned.

And so the people best positioned for calling out vice are auto-bought into it on the way there. Even though the very point of this practice of guilt-by-association seems to be to empower the calling-out of vice—raining punishment on not just the offender but those who wouldn’t shun them. This might be overall worth it (including for reasons not mentioned in this simple model), but it seems worth noticing this countervailing effect.

Prediction: If consortment was less endorsement—if it were commonplace to spend time with your enemies—then it would be more commonplace to publicly report small wrongs.