It seems to me that this article is the one prevaricating between "ethics" and "safety". The latter is of course a narrow subset of the former, as there are many ethics issues that are not safety issues.
You might not be aware of the context (actually the author of the article might not either). There has in fact been a big push by major AI companies to focus on quote safety unquote while marginalizing (not citing, giving attention to, etc) people focusing on what those companies call quote ethics unquote.
For example, from Timnit Gebru:
> The fact that they call themselves "AI Safety" and call us "AI Ethics" is very interesting to me.
> What makes them "safety" and what makes us "ethics"?
> I have never taken an ethics course in my life. I am an electrical engineer and a computer scientist however. But the moment I started talking about racism, sexism, colonialism and other things that are threats to the safety of my communities, I became labeled "ethicist." I have never applied that label to myself.
> "Ethics" has a "dilemma" feel to it for me. Do you choose this or that? Well it all depends.
> Safety however is more definitive. This thing is safe or not. And the people using frameworks directly descended from eugenics decided to call themselves "AI Safety" and us "AI Ethics" when actually what I've been warning about ARE the actual safety issues, not your imaginary "superintelligent" machines.
There has been a big push by major AI companies to focus on "safety", which they understand to refer to the novel types of harm that a powerful AI model might cause.
It's true that some people are confident there's no such novelty, and it's impossible for an AI system to cause a problem which can't be analyzed within the frameworks we've developed for human misbehavior. Some of those people do say that if you don't agree with them it must be because of "eugenics". But both of these positions make so little sense to me that I'm not sure how to engage with them.
Your second paragraph doesn't match the dialogue around the topic that I've encountered.
I generally agree with your first paragraph. My summary of the critique on "safety vs ethics" is that the push to focus on "novel" types of harm has come with dismissing and glossing over of AI reproducing and amplifying existing harms. These are well documented in machine learning from the pre-LLM era (e.g. books like Weapons of Math Destruction).
Hmm, that's a very good distinction. But I think there's still a large overlap in which safety can include the prevention of intentional harm, not only accidents. For example, traffic bollards can stop a vehicle that has lost control, as well as stopping a deliberate ramming attack. There is no question that they are a safety feature but ethics doesn't really factor into it. People might debate cost-benefits but nobody really debates "shouldn't we allow trucks to ram their way into schools and hospital lobbies?"
True... I was trying to define them the way (I think) companies are defining them (like what their alignment teams are looking at) and the way it's reported. I think in these specific contexts they're used with overlap but yeah I do bounce back and forth a bit here.
For example, from Timnit Gebru:
> The fact that they call themselves "AI Safety" and call us "AI Ethics" is very interesting to me.
> What makes them "safety" and what makes us "ethics"?
> I have never taken an ethics course in my life. I am an electrical engineer and a computer scientist however. But the moment I started talking about racism, sexism, colonialism and other things that are threats to the safety of my communities, I became labeled "ethicist." I have never applied that label to myself.
> "Ethics" has a "dilemma" feel to it for me. Do you choose this or that? Well it all depends.
> Safety however is more definitive. This thing is safe or not. And the people using frameworks directly descended from eugenics decided to call themselves "AI Safety" and us "AI Ethics" when actually what I've been warning about ARE the actual safety issues, not your imaginary "superintelligent" machines.
https://www.linkedin.com/posts/timnit-gebru-7b3b407_the-fact...