I think that AI tools have lifted the floor of what's possible, but we haven't seen the roof of what's possible yet either. We need to redefine what's low-effort, rather than going after just the tools, since a) that's incidental and b) will probably be something different yet again in a couple of months anyway.
And yet, book publishers have offered --in one instance--2.3 million USD for an AI written text.
Those aren't em dahses but the mis-spelling was deliberate to show this coment was not generated by AI.
Off-topic but I caught Gemma 4 32B making a typo on my Mac a few days ago-- it said mononpsy instead of monopsony. It felt like I should win a prize or something, I'd never caught an LLM making that kind of mistake before. It was the quantized Ollama version.
I think that AI tools have lifted the floor of what's possible, but we haven't seen the roof of what's possible yet either. We need to redefine what's low-effort, rather than going after just the tools, since a) that's incidental and b) will probably be something different yet again in a couple of months anyway.
And yet, book publishers have offered --in one instance--2.3 million USD for an AI written text. Those aren't em dahses but the mis-spelling was deliberate to show this coment was not generated by AI.
Off-topic but I caught Gemma 4 32B making a typo on my Mac a few days ago-- it said mononpsy instead of monopsony. It felt like I should win a prize or something, I'd never caught an LLM making that kind of mistake before. It was the quantized Ollama version.