cousin_it
Right. The last important battle before the end will be whether AIs are available to all (with dangers like “rogue actors”) or only to the powerful (with dangers like “eternal tyranny”). Most people on LW have been in favor of the latter, which I think is catastrophically wrong.
I’ve been critical about your “theory of change” in the past, but regardless of that, I think the arguments in this post are completely right. Things really are this bad. And yeah, joining big labs to work on alignment is especially not helping.
A lot of people have pointed out the CS Lewis quote which is perfect for the occasion:
“At first, of course,” said Filostrato, “the power will be confined to a number—a small number—of individual men. Those who are selected for eternal life.”
“And you mean,” said Mark, “it will then be extended to all men?”
“No,” said Filostrato. “I mean it will then be reduced to one man.”
Well, it’ll certainly help rule territories without people’s consent. And whether AIs or human elites will do the ruling isn’t much comfort.
Sometime ago I came up with a debate protocol that might be a good fit for this question, let me know if you’d like to try.
honesty without mentioning the ticklish issue of it maybe killing us all
So, dishonesty.
Failing that, we need to help them align it.
This thread is going in circles, let me restart cleanly.
If outright misalignment (AI straight up kills everyone) is averted, the next most likely outcome is alignment to power. This is what the question “alignment to whom” is about, see my first comment. For most people in the world, including me (a non-American, etc), this means permanent subjugation which is as bad as death or worse. Right now Anthropic’s answer to “alignment to whom” is unsatisfactory, see my second comment. So no, nobody should help them align it, unless they straighten the “alignment to whom” story.
But something convinced them to split from OpenAI and start a more alignment-focused lab in the first place. So there must’ve been convincing arguments (to them) then. The simplest explanation why they’re racing to RSI now is that they got closer to money and power.
Aren’t xAI who fails to care about alignment and China who trains models to censor themselves perfect examples of such bad actors?
Well, racing to recursive self-improvement without solving alignment kills everyone. A company can’t justify that by pointing to other “bad actors”, at that point they’re a bad actor themselves.
And even before killing everyone, there’s other stuff that needs mentioning. Anthropic, like the other US labs afaik, has agreed that its models can be used by the US government for blanket surveillance of non-Americans. The US has a history of supporting nasty regimes abroad (see Operation Condor in South America) and sending them data to help political repression (see the Indonesia massacre).
How the alignment problem gets solved—or not—in this future is something we are least certain about. … But if a slowdown simply lets the least cautious actors catch up technologically, it could leave everyone less safe.
Another crazy text in these crazy times. “We don’t know how to solve the alignment problem, but we’re going to race ahead anyway, because otherwise less cautious actors will win.” Which less cautious actors, Anthropic?
And also seconding Oliver’s question. What about power concentration, Anthropic? Your CEO has said literally this: “Anthropic has much more in common with the Department of War than we have differences.” Alignment to whom?
What about public transit in dense urban areas? Well, Ed Glaeser says:
Forty years of transportation economics at Harvard can be boiled down to four words. Bus good, train bad.
Simply maintaining roads, optimizing bus stops, and using congestion pricing to maximize throughput will produce a cost effective transit system. Ride-sharing autonomous vehicles and busses are about to make the case for roads even stronger.
Visit a big city subway at rush hour and estimate how many buses it would take to get the same throughput. Or how many single-person autonomous vehicles it would take to get the same throughput and how wide the roads would have to be.
I think his anti-rat views like this one
As everyone knows, the most important truth rationalists have uncovered with their superior powers of induction is that the robot uprising is coming. ChatGPT will shortly turn into a sphere of paperclips expanding through the galaxy at the speed of light. I think they’re wrong here, but it’s not impossible. What’s strange is what they’ve actually done with this belief. Rationalists don’t just think AI will kill us all; they’re significantly overrepresented among the people who are actually building AI.
are quite reasonable, and shared by many people here, including me =)
Yeah, I can’t check right now, but my guess is some electrical effect when changing fan speed.
He’s been writing a lot of great stuff. Last year he had a post on AI art, and the Neanderthal story which is my favorite thing he’s written.
Maybe eventually rationalists will drag him into the orbit (he wrote about being at a party with Scott once?)
Was it indoors or outdoors? Like I said, electric lights often have a flicker frequency (50 or 60 Hz depending on country, or some multiple of that).
The same thing can happen with musical instruments btw. When I play guitar indoors, sometimes I can see the string “wobble”, much slower than the sound frequency ought to be. I’m pretty sure that’s because of the interplay between string frequency and light flicker. In sunlight it doesn’t happen.
Have you ever looked at a spinning fan where it seemed like you could see the blades, rotating at a speed much lower than they should be? That’s aliasing. Your eye doesn’t sample an image often enough
I don’t think this is right. The effect can be seen in video recordings due to shutter speed, and under some kinds of electric lights which flicker at high frequency. But if you look with your own eyes in daylight, a spinning fan will just look blurry. Human eyes have no shutter and don’t take point samples.
The same goes for sound. With sampling, a signal above the Nyquist frequency will “wrap around” and seem like it’s lower frequency. But with human ears, no. A sound above the cutoff frequency of your ears simply won’t be heard.
Here’s another fun discrepancy between Fourier/Nyquist/etc and what our senses actually do. Imagine a sawtooth wave at 1Hz. As a trigonometric series, it’s a sum of many harmonics. But of course your ears won’t hear that sound as a uniform hum of many notes at once. They’ll hear it as a periodic click once per second.
Due to things like this, it takes a little bit of practical intuition to apply Fourier/Nyquist/etc to signal processing intended for humans.
For me at least, a good way to develop taste in X is to intentionally look at many examples of X. When I was learning web design, at some point I got an idea to just look at hundreds of random websites and listen to my feelings, without thinking of design principles at all. Really fast, flip-flip-flip. After doing it for maybe ten hours spread over several days, I found myself having a more clear design sense. It was much more effective than reading lots of articles about design principles.
Haven’t done this with clothing, but I suspect that walking outside and deliberately looking at many people’s clothing choices might help someone develop a fashion sense really quickly.
I think this debate would become more interesting if equality-focused people really engaged with the IABIED argument, and alignment-focused people really engaged with the “alignment to whom” argument. As it stands the debate is stuck on both sides.
I think the main problem is getting enough grip, not rotation per se. It’d be a lot easier if jar lids had little wings like on a wingnut. No idea why something like that hasn’t caught on.
Listen to people younger than you, actually. Do free things for people less powerful than you.
Mainly because I don’t see top AI labs as trying to build moral AI anymore. Dario Amodei saying “Anthropic has much more in common with the Department of War than we have differences” killed that dream for me. They’re trying to build AI that’s aligned to the powerful. In that scenario, the powerful screwing over the powerless is a historical certainty and that’s death or worse for most people. It’s basically as bad as the paperclips path.
Of course the path where AI power is spread out to the masses is also very dangerous. But I have a little bit of hope that the masses can agree on a system where everyone is ok, or at least set up some hurdles so the powerful don’t just steamroll everything.