[c1sc0]

But what about “assisted by” AI? Plenty of people use LLMs to enhance their writing abilities, like, say, ‘90s era grammar & spell check. Plenty of AI users are sophisticated enough to understand that dumping pure AI-Gen content is a bad idea. And what’s wrong with AI-enhanced speach?

[nullc]

Worse, OpenAI LLM pathologies are creeping into text written by actual humans because people are seeing so much garbage written by it that they're adopting its behavior.

Turns out that there is more than one kind of learning machine in play online and both can pick up the bad behavior of the other.

[maeil]

That's nothing new. Actual humans have been writing businessy LinkedIn posts this way long before GPT-3 came out. I'd even say such posts are even more awful than what GPT produces by default.