Hacker News Reader

Comment by probably_wrong | Hacker News Reader

I know this is going to sound mean, but I hope it drives the point across: have the EA people actually done something so far? Besides talking, I mean.

I have yet to see a single AI's malicious behavior stopped by a proponent of AI risk, and that's considering that today's AI is still pretty dumb. I want to believe that they have released code, a specific policy, or something that I can use right now to stop the dumbest of botnets, but I'm not aware of anything.

Anyone better informed that me could chime in?

View on HN

Why is that the metric that you would judge EA on? Huge portions of EA are entirely uninterested in AI risk. And AI risk isn't interested in botnets. Honestly, have you read any of their material? If not, that's fine you don't have to be interested in it (it's getting plenty of attention at the moment and really is no longer a neglected cause) but why bother criticizing something you haven't attempted to understand?

Either way, MIRI has published some papers that they think are working towards their goal. https://intelligence.org/research/ I'm not well informed enough on this subject to try to judge that claim, though, and overall I'm conflicted as to how important I should rate AI risk. But I think having some people work on it is good, in much the same way that I value basic science research.

> have the EA people actually done something so far?

In terms of stuff done: OpenPhil, an EA org, has given north of $30million to groups researching strong AI and its risks[0]. Some effective altruists also choose to donate to AI research groups like MIRI[1].

Why would stopping a botnet be evidence of effectiveness at tackling strong AI? Botnets don't exhibit any kind of intelligence as far as I'm aware.

0. http://www.openphilanthropy.org/giving/grants

1. http://effective-altruism.com/ea/14c/why_im_donating_to_miri...

> Why would stopping a botnet be evidence of effectiveness at tackling strong AI?

Because it's much, much easier than their ultimate goal, and you could generalize from there, like biologists do.

But also: because it would show that they have more claws that (say) Asimov, who gave clear lines about what a robot should not do, and yet is entirely powerless to stop anyone from making such a robot.

If they want to stop strong AI, shouldn't they be able to actually stop first regular, weak AI? And if they want to stop weak intelligence, shouldn't they be able to stop a barely intelligent botnet first?

There's also real research progress being made. Here are some of the issues people are working on:

https://arxiv.org/abs/1606.06565

> have the EA people actually done something so far? Besides talking, I mean.

Yes: Donated money: https://www.givingwhatwecan.org/

I primarily give to the Against Malaria Fund and GiveDirectly. Some others donate to AI-relate research.

I personally haven't spent any time as an AI researcher OR as a malaria-net seamster. Does that mean I haven't done anything? Or does donating tens of thousands of dollars for bed nets count?

If that counts, why wouldn't donating to causes like AI risk count?

They're so worried about AGI, which doesn't exist, that they won't spend any time on the non-existential risks of non-general AI.

Listen to Cathy O'Neil for a more reasonable and actionable perspective on how we need to fix AI.

Anything specific that I should listen to? She seems to be a prolific writer, and I'm not sure where to start.

I suggest her book "Weapons of Math Destruction" in particular.

Reply on news.ycombinator.com