跳到正文
DEV Community· Alvaro Montoro·· 1 小時前AI 評分35

非凡主張需要非凡問責

Extraordinary Claims Require Extraordinary Accountability

AI 導讀

Anthropic IPO 令 AI 行業受到更嚴格審視,文章主張,聲稱 AI 技術可能帶來生存風險的高管,應接受相應程度的監督與問責。文章指出,若高管相信這些風險,便應解釋為何仍要以目前步伐推進;若不相信,就不應把風險説成可能發生的結果。美國應建立 AI 問責標準,並加強監督與透明度,而非停止行業發展。

正文

As Anthropic's IPO puts the AI industry under the microscope, executives making extraordinary claims about existencial risks should face extraordinary levels of scrutiny and accountability

I understand that executives have to sell their products and make them look appealing.

But saying, "This may kill or enslave all of us within two years," is neither.

If executives in any other industry made claims like that about their products, there would be public outrage and immediate calls for regulation. Yet when AI executives say it, we're somehow supposed to treat it as a reason to move faster.

And "If we don't do it, someone else will" might be the weakest justification imaginable.

If AI leaders really believe the existential risks of their technology, they should explain why moving forward at this pace is justified. And if they don't think those risks are real, they should stop presenting them as realistic outcomes.

Either way, the public deserves more accountability than "trust us" and "someone else will do it."

A handful of executives shouldn't be the ones deciding this future on behalf of everyone else, especially when some of them are deliberately, recklessly, or both, making claims about outcomes the could have devastating consequences for society and the world.

Yes, in the past, other industries have posed existential questions too: nuclear energy, particle accelerators, cloning, biotechnology... and they have been subject to regulations, safety standards, and oversight mechanisms. Innovation didn't stop because of them. Why should AI be different?

We've seen attempts outside of the US: The EU's AI Act is far from perfect and has its critics, but it represents a meaningful attempt to establish rules and accountability. And it's definitely better than doing nothing.

It's time for the US to take a stand, lead, and establish basic standards of accountability for what AI companies say, claim, and do.

This is not a call to slam the brakes and stop the AI industry. It's a call for meaningful oversight, transparency, and clear standards of accountability.

AI should not be exempt from the rules and responsibilities that apply to other industries. If AI companies want us to take their extraordinary claims seriously, they must be willing to accept extraordinary scrutiny in return.

來源:DEV Community · dev.to