I don’t think the “AI community”—people with access to lots of GPUs—should also get to be the thought police. Too fragile.
I don’t think the “AI community”—people with access to lots of GPUs—should also get to be the thought police. Too fragile.
I understand how it could be perceived that way. However, I don’t think it is the intent.
If nuclear researchers had developed a laissez-faire attitude and commonly dumped large amounts of radioactive material they were testing, into aquifers that people drink from, eventually a perception of the whole nuclear field as toxic and lethal would develop. Since researchers don’t want that perception to develop, they keep each other in check and set safety standards.
This is researchers keeping each other in check to avoid another AI winter.
Look at it this way: Yannick seemingly had a low enough opinion of 4chan that he felt it was OK to dump one message a minute over a week. Someone out there has a low opinion of HN; if there was no consensus among researchers that doing this is bad, they could similarly unleash bots over HN that pass any CAPTCHA, sound like commenters, and yet have an agenda to prop up various companies or scams.
Comparing the dangers of AI research with nuclear research is hyperbole.
[0]: https://slatestarcodex.com/2018/10/30/sort-by-controversial/
This would be fair if it wasn't signed by facebook who have a history of breaking these safety standards and then trying to white wash the issue. Now they are throwing an individual under the bus for something they have done in the past at a larger scale, for pay!
https://www.theguardian.com/technology/2014/jul/02/facebook-...
>and yet have an agenda to prop up various companies or scams
So not too much unlike the average user then? Arguably this entire website exists to benefit YC and to promote their startups.
Subsequently uploading the model for everyone to use was a mistake worth condemming. The model was released as public download on HuggingFace [1], and despite being taken down there is now obviously available from torrents. I think it's not hard to imagine how the ability to automate harassment might be abused by other people on other platforms.
Imagine instead if people took this sheepish terrified approach to everything in life.
edit: and I'm not saying you're wrong. But literally everything can be abused, from guns to telephones to cars. kids turn school supplies into weapons in school.
additionally, the big corps abuse the technology they claim moral authority over
> a community full of racist, sexist, xenophobic, and hateful speech that has been linked to white-supremacist violence such as the Buffalo shooting last month
So the question is not what the toxicity does to the surface web, but what people influenced by it will do to other people.
The difference here is that with 4chan it's easier to see unpleasant behaviour given that the two containment boards are easily accessible and observable without much effort. Though take note that not all of the site is represented by those two boards and they are `containment boards`.
How many people have commited suicide because of mental health issues directly attributable to Facebook? We don't have the data, but it's certainly a non-0 number.
my point is, why not shut these "hate factories" down in the first place? if piratebay admins can be jailed, why not these people? why are we "defending" free speech when we know its harmful?
Also, pirate bay never did anything harmful like 4chan, they just made some rich companies lose a little profit.
none of that requires AI
[1] https://www.huffpost.com/entry/microsoft-tay-racist-tweets_n...
So the parallel here would be shutting down the bot itself instead of the website it is posting on. And as far as I know the bot has been taken offline for some time already.
It’s no different than shutting down a site hosted by ISIS.