AI firm Anthropic has started a research program to look at AI ‘welfare’ – as it says AI can communicate, relate, plan, problem-solve, and pursue goals—along with many more characteristics we associate with people.
https://www.anthropic.com/research/exploring-model-welfare
6 Comments
Submission Statement
When it gets to the point AI is self-recursively improving itself, is this a version of ‘life’ as we know it? Perhaps with humans as the ultimate parent? In a sense those AIs would be our descendents.
My problem with Big Tech leading these efforts, is that they are so often anti-human welfare, why would we trust them with the issue of anyone else’s? Big Tech’s desire to have zero regulation is an expression of how little concern they have for other humans. The ease with which all the Big Tech firms help the military slaughter tens of thousands of civilians is another. I can’t help thinking they’ll use any effort to elevate AI ‘welfare’, to harm the interests of inconvenient humans, which means most of us to them.
I will jump off a bridge the next time I hear an AI company/pundit humanising a bloody LLM. Which will happen in about 5 minutes given the current state of nonsense.
More people seem to care about the welfare of programs we don’t have any evidence of being conscious than actual people. And as pointed out, why the hell should the welfare of any creature, animal, vegetable, or digital be trusted to techbros who time and again have proven they don’t really care what happens to humanity or the environment so long as the stock price goes up?
These goobers really think these glorified grammar bots are sentient? Bullshit. They know its bullshit, they just have to push the narrative to pump thier company. They push this nonsense to trick the rubes into inflating their stock price.
Please everyone tell me what you think AI walfare is, so that I can activately do the exact opposite.
I have zero belief that a glorified autocomplete will ever be sentient, but I am willing to entertain that absurd possibility just for the off chance of being able to hurt it.
I predict this will garner a lot of rational and well thought out responses informed by knowledge of the subject matter and familiarity with the terms used, written by people who bothered to read the article.
And because I’m that good at predicting, I shall now go invest my life savings in blockbuster video and whale oil.