Comment by riverlong | Hacker News Reader

Comment by riverlong | Hacker News Reader

I'm worried that this situation is going to get a great deal worse. A flurry of disinformation and online noise is somewhat manageable so long as it is coming from humans, who may be paid by a state or an enterprise to drive engagement with controversial topics.

It's going to get a million times worse when this gets automated.

Right now the quantity of disinformation is practically limited by the human capacity to post stuff online, but GPT-3 style automation will remove this limitation, and we will live in a world of all noise, no signal. Our knee-jerk adherence to "free speech" is opening up a dangerous scenario where we are more vulnerable than anyone else to what is effectively a DDOS on our minds.

I explored this topic in more detail in an essay of mine, where I lay out that a world of Anonymity + GPT-3 can undermine free speech by drowning out real human views and opinions with machine-generated, human-looking propaganda. https://jayriverlong.github.io/2020/07/24/gpt3.html

Interesting essay Jay. The top comment in this thread is by throwaway13337. You see this when people have a hot take they want to post but aren't ready to own it in front of their peers. I'd bet on average comments from accounts with identifiable personal info are much more tame and civil. But if everyone is too scared to talk about something for fear of getting cancelled then anonymity is the natural answer.

---

Maybe in the future if something is written by a bot it will need to be disclosed that a real human did not %100 write it. I mean I just looked at your blog and your twitter and I feel like a devious enough actor could use AI to help them create just as genuine of an online presence as you have, even down to the profile pic.

I actually wanted to raise this discussion on HN and created a thread [1] but it didn't get any traction. Maybe it could be a future topic for you.

If things continue the way they are we'll all be paranoid about everything being fake, like an internet Truman show. The sad part is you know it's fake and there's nothing you can do about it. So basically we have to go back to meatspace for genuine human interaction which is not the absolute worst I guess as long as there's not a deadly pandemic going on.

1. https://news.ycombinator.com/item?id=24807090

I think there's a solution in pseudonymity: it has the liberating effects of anonymity, while providing some accountability that there's a real human behind the digital face.

For example, suppose there's some repository of hashes of Personal IDs. If you want to sign up somewhere, you have to submit a Personal ID that corresponds to such a hash -- the Personal ID doesn't get stored anywhere, but it authenticates that you're a real human. (Obviously this is a naive solution and something more clever would be necessary, but it's just an example.)

Reply on news.ycombinator.com