Hacker News Reader

Comment by NamTaf | Hacker News Reader

No, in the same way that I am not tired of the restraints ethics boards put on medical experiments.

Tech is now pervasive and AI has the power to do some pretty powerful stuff. This nexus of circumstance means it’s high time similar questions get asked about whether we should.

In the same way that medical science isn’t one dude cutting apart things in his basement, bleeding-edge tech is a multi-person and very organised endeavour. It is now in the domain where it really should have some oversight.

View on HN

Is a system that can generate images of gore, or of sexual acts really on the same level of risk as medical experimentation?

Any human can draw genitals. We can draw boobs and butts. Should this ability be regulated as if it's a risk to human health?

> Is a system that can generate images of gore, or of sexual acts really on the same level of risk as medical experimentation?

Are you sure that's what AI ethics is really concerned about? I of course haven't read all the possible sources giving definitions of AI ethics, but I'm pretty sure you've got it wrong.

A widespread racist AI could indeed negatively impact the mental health of hundreds of millions of people.

If someone can use the tech to produce realistic images of someone they can use those images to harass and abuse that person. That’s a horrendous position for someone to be put it. I’d advise looking at some peoples stories of being the victim of revenge porn, upskirting etc. very hard to empathise after hearing those accounts.

Ok so you’re proposing … what?

That it be regulated and the people that create the tech act responsibly.

Regulated how?

How can regulation address what you mentioned?

Harassing someone is already illegal.

Should hammers be regulated because they can be used to bludgeon people?

For now, I’d argue they should be regulated by the people working on them. OP’s complaint should be ignored.

And yes, hammers are regulated by non-government adults to the degree that people cannot figure out how to handle the consequences of using them.

> Regulated how?

I don’t have the knowledge to decide this on a whim.

> How can regulation address what you mentioned?

By limiting the companies creating these models. Eg, you can’t create a model that produces nudity, gore etc.

> Harassing someone is already illegal. > Should hammers be regulated because they can be used to bludgeon people?

Stabbing someone was illegal but we in the UK we still banned carrying knives in public. These things are not black and white and trying to frame them as such will get you nowhere.

That's not what the actual risks of deepfakes are.

You don't think the Nazis would not have used AI deepfake technology to spread propaganda or try to refute the facts of the Holocaust?

We already have evidence that state actors have a lot to gain from misinformation. We have evidence they already use selective editing and image manipulation.

The risks are erosion in societal trust of data sources, reputational risk, and risk of misuse leading to violence or criminality, or misleading voters leading to things like misallocation of shared resources.

Reply on news.ycombinator.com