> the tools to detect will be exactly as good

[citation needed]

I was curious about this recently, so I built a very rudimentary neural net trained on GPT generated text messages, and human generated text messages. I was able to get a surprisingly good detection accuracy with just under 1k lines from each sample set. I'm not sure it's as apocalyptic as people think.

I guess time will really tell.

We might as well hit the problem Google Translate hit, where the training set started to contain more and more data created from GT itself. Similar thing may happen with your NN. At some point there may be so much AI-generated content (by different AIs) that it may be difficult to compose a trustworthy training set.

I suppose a solution to this would be something like pre-war iron. We would have to rely on archived sources, like Wikipedia past edits, that come from before ChatGPT existed.

I thought about this a bit, and am wondering about the results of LLM-fueled arms-race when it comes to figuring out whether an AI has created some piece of text.

I'm worried we may reach a point AI will get so good faking people that real peoples' output will be treated as fake due to only as many combinations possible that originates from humans. You will start depending on AI telling you what is true and what is fake for every single piece of information. This leads to a question how to tell which AI is right - you can't really verify an opinion, only facts.

And being able to manipulate opinions is a very strong perk.