News for hackers. Stuff that matters.
Re:Is artificial consciousness achievable? Lessons from the ... by ben_w on Sunday May 19, 2024 @05:51AM (#40405682)
I know that at least one other human is conscious, otherwise the term would never have been invented.

But you have no way to tell if I am as conscious as I claim to be, or if I'm just a large language model trained by humanity :P

Re:Is artificial consciousness achievable? Lessons from the ... by golf_mike on Sunday May 19, 2024 @07:11AM (#40406064)
Can you provide proof for the first claim?
Re:Is artificial consciousness achievable? Lessons from the ... by ben_w on Sunday May 19, 2024 @08:13AM (#40406398)
Which part exactly are you seeking proof of, and to what standard?

"I know" is unprovable to others, unless you examine the wiring of my brain. (But then, what is "knowledge"?)

"at least one other human is conscious, otherwise the term would never have been invented." — it's always possible that I'm a Bolzmann brain and this was just luck.

I don't see how the term could have been invented by a mind that didn't actually have it, except with astronomical low probably random events.

Re:Is artificial consciousness achievable? Lessons from the ... by vixen99 on Sunday May 19, 2024 @07:14AM (#40406088)
In thinking about this perennial problem it's worth bearing in mind that human beings pick up and process an immense amount of data on a continuous basis, that is currently unavailable to any LLM.
Re:Is artificial consciousness achievable? Lessons from the ... by elicksaur on Sunday May 19, 2024 @11:31AM (#40407567)
And all on an estimated 100 watts!
Re:Is artificial consciousness achievable? Lessons from the ... by ben_w on Sunday May 19, 2024 @12:48PM (#40408138)
20 for the brain, 60-125 for the whole body depending on if you mean "normally" or "metabolic minimum".
Re:Is artificial consciousness achievable? Lessons from the ... by elicksaur on Sunday May 19, 2024 @07:43PM (#40410602)
Yeah there’s different ways you could slice it. I was including the whole system, since if the GPU’s cooling system breaks, it won’t be useful for long.
Re:Is artificial consciousness achievable? Lessons from the ... by anticensor on Thursday May 23, 2024 @01:56AM (#40450963)
The "metabolic maximum" is around 1250, though. And a mediocre lifting or running momentarily increases it from 125 to 700.
Re:Is artificial consciousness achievable? Lessons from the ... by vidarh on Sunday May 19, 2024 @04:03PM (#40409400)
No, we don't. We don't even know if existence has an extent in time, because our only way of "interfacing" with time is our experience of memory we can't prove is real.

For what you know, you're a lone entity confined to an infinitely short period of time, and all else is an illusion.

But of course this isn't a useful assumption in most respects.

Re:Is artificial consciousness achievable? Lessons from the ... by ben_w on Sunday May 19, 2024 @04:26PM (#40409519)
Ah, I see you're more of an A. J. Ayer fan than a Descartes fan.

I think that if an LLM has any consciousness, it would be an experience like this — one where the past was a fiction it invented to fit the prompt, and the "now" was the only moment before the mind was reset.

But I'd put that in the same basket as my… ah, nephew comment? Cousin comment? I guess you'd call it that if we have parent comments etc.?

https://news.ycombinator.com/item?id=40406398

What you say is not wrong in principle, but it's in the same "cognitively unstable" basket as a Boltzmann brain, where to accept it would mean I couldn't trust my own reason to believe it.

Re:Is artificial consciousness achievable? Lessons from the ... by vidarh on Monday May 20, 2024 @12:56AM (#40412383)
The reasoning stands entirely on its own. There's nothing you need to "believe". That you don't have a way to disprove it is close enough to proof.

The problem with it is that it's not giving us anything useful.

OK, so maybe this moment is the only one there is? Now what?

Accordingly the only reasonable recourse is for the most part to ignore it, and add an implicit "assuming my memory and senses are roughly trustworthy" to every assumption.

This is the same reason I'm a firm materialist even though we could be in a simulation, or brains in vats: absent evidence of either, the only thing giving us useful information is our senses and our introspection, and so the most useful approach is to assume they are valid until proven otherwise.

Re:Is artificial consciousness achievable? Lessons from the ... by ben_w on Monday May 20, 2024 @10:42AM (#40416052)
> The reasoning stands entirely on its own. There's nothing you need to "believe". That you don't have a way to disprove it is close enough to proof.

Not so, for it is like the liar's paradox, except with high probabilities rather than Boolean logic.

If you assume it is true, it follows that you can't trust your own state of mind, and that all statements including "I am just a thought" become suspect, as there is nothing leading up to them to give them justification.

Re:Is artificial consciousness achievable? Lessons from the ... by vidarh on Monday May 20, 2024 @11:12AM (#40416437)
No, it's nothing like that. In the liars paradox, the statement is inherently uncomputable.

In this case the statement is either true or false, and the truth or false of the statement doesn't change anything. Your belief in whether the statement is true or false changes your belief in whether the statement is true or false - there's no basis for assuming that in the general case it would alter the actual truth of the statement (it could, in as much as e.g. you could have a simulation that keeps running as long as you believe you're in an isolated moment, and freezes you and discards the rest the moment you believe time is real).

Now, it is true that you can't trust your own state of mind to truthfully represents a physical reality. That this means you also can't trust your own logic is irrelevant, because irrespective of whether you believe this to be true or false, the absence of any external source of validation of your senses or state of mind or existence in time means the statement must inherently be true irrespective of your trust in your mind.

This claim gives a straightforward condition to falsify it: Find any single source of external validation that does not depend on your senses or observation of it. If you can, then the claim above is false.

Given you absolutely can't, the logical conclusion is that the claim is true.

Re:Is artificial consciousness achievable? Lessons from the ... by rixed on Sunday May 19, 2024 @10:52PM (#40411725)
Of course, any philosophical zombie could have invented it. Which makes clear for me how loosely consciousness is bound to senses, including that of introspection. It actually makes me feel that this consciousness we can barely describe is but a faint experience of a distant, largely foreign thing. Like we know the experience of matter that we get from our senses is a grossly simplified version of the real thing (that we still don't know exactly of course, but we already know enough to see how far away our immediate mental model is from the truth)