3 ms·
Is AI Em Dash Addiction Real? A Model Comparison
- lich_king 7mo agoAn article about the AI writing style, written using AI.
- don_r 7mo agoI hadn't heard the RLHF feedback loop argument before, but it seems like a solid explanation for all the bad ticks LLMs have when writing and why it's so hard to break them out of it. All of that stuff... "That's not ____; it's _________" and the like... you can imagine human evaluators going "Wow, that's profound! Great output!" and now we all get to suffer through reading that for the rest of our lives.