Users respond positively to the AI researcher meme contrasting misalignment challenges with recommendations by enthusiastically sharing relevant research papers.
Based on 1 visible X reactions from 3 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@voooooogel Have i got a paper for you! https://arxiv.org/pdf/2606.07612
This is a deeply good analysis of a problem with it that i think i kind of skimmed over, thank you. I think I'm a fishing water for anthropomorphism funnily enough like someone at a party recently asked me if people tell me I entered memorize the model is too much and when I was like no not at all she was like wow people tell me that all the time and you do it way more than I do. all of this stuff just reads as intensional stance to me and doesn't require any great distinction between anthropomorphic and not. I didn't think the intended Takeaway of the paper was about that part as much as about the bad science and part which is why I like the crumbling pillars. but it's totally possible I just don't realize they're trying to make that point. I think you and I are on the same page about the quality of these experiments versus how they are extrapolated in the discourse. I think this latter paragraph in the screenshot is a more polite way of saying something I have tried to say before e.g... (also thank you for sharing your full chain of thought with me, I would never send that to a model instead of reading. Some of the most valuable tokens in the world)
It's too late. I've already depicted your research as the crumbling pillars and my research as the firm stalwart pillars https://x.com/deepfates/status/2075725754051309888/photo/1
@deepfates what's this from?
Users respond positively to the AI researcher meme contrasting misalignment challenges with recommendations by enthusiastically sharing relevant research papers.
Based on 1 visible X reactions from 3 accounts; directional sample.
Ask a question below.
Published answers will appear here.