I'd do an awesome job for them if I'd agree to taking such a position. I'd also likely be forced to betray everything I believe in so there are probably better suitable candidates for that. Like anyone without a soul.

> If you want something as dumb as an LLM to help you

You should not become the spokesperson of OpenAI. You are way too honest in your assessment of current LLM iterations. And probably also of future ones, unless, maybe, something drastically changes in their design.

south_korea_ln

science

AI will soon be able to audit all published research...

What I find interesting is the phrasing

> This “science is broken” narrative undermines public trust.

I think that this is awesome. Question everything. Not just the X-tard but also the guy with the credentials to do much larger damage.

Perhaps, as the author suggests, AI will make this easier. However, thus far all I have really noticed was Qwen3 in its reasoning step finding a real error in some python code and then "deciding" to not tell me about it.

If you want something as dumb as an LLM to help you, you need to train it well. The good news is that you can train it fast. 