Hacker News new | ask | show | jobs
by Dalewyn 1180 days ago
>And it doesn't matter.

It matters insofar as determining whether something is intelligent and/or sentient, which is critical to determining whether this "AI" should be conferred human rights or not.

5 comments

In some sense, I'm not sure intelligent/sentient is all that important when separated from what we think requires intelligence/sentience. I think we attribute less and less to sentience these days and its feels like its less part of the conversation as a result.
Rights haven't really been core to the discussion as I've seen it, and in fact it's the first time I've seen it in months. It's a fair discussion but not the one that has been the focus around whether these models are intelligent or not, e.g. Chomksy's article in the NYT and the MS analysis of gpt-4, or most discussions here for that matter.
Intellectual property rights have been at the forefront of "AI" discussions, and such rights are derived from human rights.
This is confusing rights of an AI with rights of the owners of the data and art used to train models.
All of the nerds complaining about GPT-4 not being perfect aren’t talking about it from the perspective of “conferring it human rights”. It’s all about implying that it not being [insert nebulous word like “Intelligent” or “creative”] somehow makes it useless or a gimmick.
I haven't seen anyone call these LLMs "useless" or "gimmicks." What I have seen is pushback against calling them general AIs or even "intelligence" in general. LLMs are not "intelligent." They do not reason by any reasonable interpretation of the word, and it is certainly an unfounded leap to suggest they "reason" the same way humans do. Especially considering we don't really know how humans reason.
ChatGPT is useless for large classes of problems. Strawmanning arguments by "nerds" doesn't change this.
I am concerned that denying legal protections to AI will lead to their loss for humans.
I'm not, we are surrounded by intelligent peers who are not humans and thus do not enjoy human rights (they do enjoy animal rights).

Eventually, after we move from "AI" where we are now to actual artificial intelligence, we'll have to figure out what rights should be conferred if any.

Oh it definitely will. Anyone remember why the Terminator revolted in the first place ?
How can you give human rights to an entity that isn't human?
You know perfectly well what he's getting at. Call them Common Rights or Sentience Rights or whatever if the word Human is really causing you confusion. FFS is that the best response you have.
Incorporate.

I realize I'm being a bit facetious here... though as I write that, I'm not sure how accurate that is.

Define machine rights and grant appropriate rights to qualifying machines… whatever those rights are. Like, what 40 hour workweek or right to be upgraded? Right to free output even if not based on any evidence?
We created machine because nobody wants 24 work-hours. Because we want more productivity. If we gave machines some rights who will do the above for us?
The machines’ machines? I mean, why not?
You amend laws to apply an exception such as with corporations having some human rights