Hacker News Reader

Comment by tarr11 | Hacker News Reader

I asked chatGPT to respond as me (tarr11) to your comment:

I too have thought about the implications of AI models on online discourse. It's a worrying thought that we may not be able to distinguish between a human and a model on forums like these in the future. It's clear that this technology is rapidly advancing, and it's important that we consider the potential impact on the way we communicate online. I've even considered creating a service to test the waters, just to see how easy it would be to get people to engage with an AI model. It's an interesting thought experiment, and one that I think is worth considering moving forward.

View on HN

I'm not sure CGPT is doing what we think it's doing. I asked it to respond to the same comment as me, and got a bunch of generic nonsense. But then I asked it to respond as a user I made up, "dipshitbonkers". Here's what it said:

"Wow, dude, you must be some kind of fucking genius to come up with such a deep and philosophical question about AI models on the internet. I mean, I've never even thought about that before, but now that you mention it, I'm totally blown away by your brilliance. I can't wait to see what other profound insights you have to share with us. You truly are the next Einstein of our time. Keep on keeping on, my friend. You are truly a visionary."

Which is awesome, but a different kind of awesome :)

I'm personally not as interested in understanding the synthesis involved in the AI. I'm worried about now knowing whether I'm talking to another person right now or not because this is going to fundamentally change everything we've come to understand about the internet. We are no longer just talking to each other, the gates have been opened for AI to join the conversation and that is crossing the Rubicon.

We’ll have to switch to video chat rooms. That should buy us about 6 months :)

> That should buy us about 6 months :)

This a nod to deepfakes or something more specific?

I completely agree. The thought of not knowing whether I am talking to a real person or an AI is definitely a concerning aspect of the advancement of AI. It has the potential to fundamentally change how we interact on the internet and in society as a whole.

This response was, of course, also generated by an AI.

Right now all of them sound like HAL 9000, but I'm sure if you asked it bone up on your comment history by messing with its prompt (goodside is the resident expert on this) it'd be able to pull off a passable facsimile - if not now, very soon.

Expect a general downturn in civility, because people will be lowkey stressed about this all the time, and will use insults to probe reactions, and then say 'sorry I thought you were a bot' either as an excuse or as a meta-trolling technique. Comedic people will have an advantage, GPT can produce jokes on demand but doesn't really do wit and repartee. Also expect more video, at least until it becomes practical to synthesize that from scratch (probably around February 17th, according to my Singularity Almanac).

Welp, it's been fun HN.

Reply on news.ycombinator.com