Vecr

Yes, David Chalmers.

For the article itself, phrases like "a naïve and toxic view" and "a debased version of what we are" are statements that need to be strongly backed up.

"For me, thinking is a specific and rather unique set of experiences we have." yes, okay, but what experiences?

What do you mean by "concepts"?

She talks about behaviorist reactions, but fails to back that up at all. Why does she think e.g. mechanistic interp. is behaviorist?

"there’s nothing on the other side participating in this communication." I think this is currently correct, for some definition of "nothing", but give an argument.

"new moral claims in the world" give a good physics-based account of this process in humans.

But this efficiency, Vallor continues, “is never defined with any reference to any higher value, which always slays me. Because I could be the most efficient at burning down every house on the planet, and no one would say, ‘Yay Shannon, you are the most efficient pyromaniac we have ever seen! Good on you!’” -- ever heard of E/ACC?

"Vallor tells me she once tried to explain to an AGI leader that there’s no mathematical solution to the problem of justice." -- not currently, but this is peak woolly. Maybe she doesn't know it, but a mathematical solution would need to contain incredibly horrible things. If you stay away from the math, you can avoid thinking about that sort of thing.

Mainly though, it's what the article does not contain. That's tricky to enumerate. I can try if you want.

The person who defined the Hard Problem researches LLMs and is a lot less woolly about it.

Being woolly isn't the best that can be done if others are making actual arguments, the kind that aren't in the article.

It's just woolly. Don't philosophers talk about causal models and mental states and simulations and China brains and aspects of externalism instead of just saying "I think AI researchers are wrong, I think it's like this instead"?

It's not optimal, but if you want to play it safe you can use 1 byte = 1 token, plus a few control tokens like stop, user turn, model turn, etc.

Those are positions, but you need arguments to back up your positions. I know this is an advertisement, but couldn't she put in at least a few so we can sample her reasoning quality?

I know multiple people that carefully prompt to get that done. The model outputs in direct token order, and can't turn around, so you need to make sure that's strictly followed. The system can and will come up with post-hoc "reasoning".

If she just wants it gone/the responsibility off her chest she could send it to one of the really big re-granters.

Is is a donor's responsibility if they're seeking better outcomes.

You apparently have a better sense of what the argument is than I do [0], would you explain?

[0]: https://news.ycombinator.com/item?id=42484852

The base models aren't tuned. Technically it's what the makers put in, but the data they train on isn't selective enough for what you're talking about.

"Trained on a dump of the internet" isn't literally true, but it's close enough to give you an intuition.

The base models have multiple "personas" you can prompt/stochastic sample into that each answer that kind of question in a different way.

This has been known since 2022 at least, and unless I'm missing new information it's not considered to mean anything.

Future natural selection that is.

What path of reasoning lead you to think natural selection is good? Because it's natural?

Where are the actual arguments? She states that AI is a mirror, and yeah, you put stuff in and you get stuff out, but who thinks otherwise?

There are interesting ways to argue for humans being special, but I read the entire article and unless I missed something important there's nothing like that there.

The fictional scenario has to be reasonably consistent. The version of Anthropic in the scenario has become morally compromised. Training on customer data follows naturally.

How are the Bongard Problems going?

Can't bet negative in that kind of game. If a game is expected to lose you money, don't play.

No it doesn't. We're defining the arrows here. Sure, we'll keep left-to-right to avoid confusing them, but as long as the usage of the arrows is consistent the message would still be correct if it was right-to-left.

That's the entire point of defining the arrows using entropy.

And fill them with what? How much nuclear waste are you expecting? At most you would need a few Yucca Mountain size facilities per century, because you need to reprocess at some point. There would probably be a few other facilities in other countries.

Bad idea anyway, you want geological stability and either low moisture or clay sealing. There are several good places, that are more than big enough, but putting waste everywhere would be stupid. At that point just store in dry casks like we do now.

High level nuclear waste gets less radioactive pretty quick. It gets easier further out, not harder.

It's still arrows though. Put some illustrations of (actual) arrows and spears along with the stylized arrow, all pointing in the same direction. That should cover the less advanced civilizations, and provide a good bridge to the thermodynamic depiction to others.

Two connected boxes, one with tightly packed dots in it. In the next two panels, a door is shown opening. Dots progress in a few panels to equilibrium, spread out more. Make the dynamics pretty accurate for a gas, but make sure all the details are big enough to survive the thousands of years.

You could make a stylized arrow in addition to that, made of dots. The base of the arrow would be smaller with tighter packed dots, and as you progress along to the tip they'd be more spread out. The tip end would be larger than the base.

Then you'd use that stylized arrow for everything in the message. Make sure it's used the same way in all the comics, including the entropy one.

Single bit errors = radiation. If the voltage log is fine, there's nothing else that could impact that many chips at once. These characters are morons.

The entropy arrow definition would have worked here.

Ray Cats isn't a good idea. I never figured out why the comic strip with arrows wouldn't work.

Have panels that define arrows using increasing entropy. It's really is almost universal.

Checks out with multiple RNG seeds.

It shouldn't be a problem because the RNG is advanced each run. Might save someone a check though.

What's mandated to be a single character? I'm not sure what the popular style is today.

Implementing algorithms doesn't need a computer, and it doesn't need something that's not a computer.

That said, with enough willpower why couldn't you implement a universal Turing machine in your brain? I think you would need external memory (a tape) but that's not a problem.

Quantum computers are never* going to be good at a whole lot of tasks that classical computers are already used for.

* Some people have weird ideas.

I think someone associated with Stephen Hawking once made that joke.

Stock movements aren't generally justified.