> Would you stop visiting hacker news even though ChatGPT is using the content from its pages? NO.

Would I stop visiting hacker news once most of the articles and comments come from ChatGPT-driven bots? Yes.

And how are we to avoid that happening?

I think via regulation.

Two hundred years ago it would be hard to imagine that everyone would be registered in a central database and will need an ID to do many things.

I think eventually the internet will have to give up on the idea of anonymity and places like HN will require it's users to be proven humans.

Hour do you handle with humans using gpt to post? The same way you would handle botters in online chess.

Of course there are drawbacks to that central registration - anonymity as we know it gets left behind. Yes the central registration could offer also "anonymous" handles for social media daily use, but to the police and any able hackers this would be nothing.

I disagree vehemently with the idea of abandoning anonymity. I don't think it will fly.

I prefer closed communities with invite chains, where user have a limited number of invites.

Sure feels like we've made a wrong turn when we find ourselves in the middle of creating a problem and already have to reach for regulation as the cure.

I fully expect GPT and similar to ruin the internet in many of the ways mentioned in this thread. What's really confusing to me though is why we are going down this path at all when it is so clearly a bad idea. Where does the net positive come from that outweighs the massive risk of such massive AI systems?

Even the leaders at MS know this is a bad idea but fell into the trap of "we're the good guys, if we don't do this a bad actor will"

“Your scientists were so preoccupied with whether they could, they didn't stop to think if they should.”

And how do you make sure the registered people don't use an assistant to write their comments and post?

I've already encountered a number of posts that seem like they're written by bots here, and they indeed reduce my time spent on the site.

How do you recognize that a comment has been written by a bot?

> And how are we to avoid that happening?

Once AI are capable of writing comments that pass as human, why even worry?

https://xkcd.com/810

How will new people find HN without the source of information being referenced?

we wouldn't stop but new users would simply get the content via ChatGPT and never even get into the mental model of going to a different webpage. They will start seeing Google/Bing with ChatGPT as their "app" and they use the app to get everything they need. The proof is in how the internet has already evolved in the past: People dont really go much to most forums anymore, they rather go to big apps and get all their info there. Think about it this way: Why would I open up and register at csharp-fans.org when I can just get all my answers via direct query on stackbot3. It will even be able to respond to questions infinitely faster than any forum user!

Existing users wouldn't stop visiting. Flow of new users would slow down and eventually stop, depending on how ChatGPT provided results are surfaced on the search results UI.

If it's presented as the first search result, if the information presented by ChatGPT is good enough and if there are no links to the where ChatGPT got the results from, people would not visit the source for the information.

Remember what happened to email replies when popular email clients started defaulting placing the cursor before the quoted text instead of how it used to be?

I don't think the real problem is proliferation of AI itself. The real problem is control of AI. We need to commoditize and replicate FOSS AI models as soon as possible, and make progress with optimization so they can run without megacorporation levels of resources, so that they aren't able to retain exclusive control of AI, which lets them grow even larger.

Seems to me that every time someone spends millions training an AI, a cheaper version that does the same thing comes along in a few years that everyone can use.

It’s going to be very difficult for anyone to retain ‘control’ of AI when they’re fighting a running battle against Moore’s Law & The Bitter Lesson.

So you think the problem of high-tech spambots is that there aren't enough people using them? Is your goal to make forums and such unusable as fast as possible?

This is happening regardless of our desires. The best we can do is to be prepared and have the tools not available to only the most powerful few, but to ordinary people as well.