Ah okay, I didn't realize that.
aio
joined 2 years ago
It's not quite clear to me that
We rely on some standard LLM detectors to focus our attention on papers that need to be checked.
implies they are using LLMs themselves. The phrase "LLM detector" is a bit ambiguous and could mean "LLM being used as a detector" or just "classifier program designed to detect LLM output".
im smarter than everyone else in the whole world, especially those whiny feminists. why hasn't society granted me a female to be my mate yet?
As I explained elsewhere, my comment was just about the inapplicability of mathematics to this question. But also, is that really what morality always says? What if polls predict 1% will vote blue? What if they predict only one other person will vote blue? Are you always obligated to martyr yourself?
You're the one who mentioned "game theory" in the first place, I was just directly quoting you. My sentence was of the form "game theory doesn't say X", not "game theory does say Y". I added quotation marks to clarify.
My point here is that you can make whatever philosophical and ethical arguments about the situation you want, but none of game theory, Arrow's theorem, nor the concept of a dictator have any bearing on it. It is an ethics question rather than a mathematical question, and it is an error to claim that your argument is a mathematical one.
If polls predict 40% blue you should not vote blue as a matter of game theory, because that is suicide.
I don't understand the relevance of Arrow's theorem. Why is your phrasing the correct way of analyzing the situation?
rationalism is when i pull five numbers out of my ass and multiply them together
... why 7/8?
the output is probabilistic not deterministic. By definition, that means it’s not entirely consistent or reproducible, just… maybe close enough.
That isn't a barrier to making guarantees regarding the behavior of a program. The entire field of randomized algorithms is devoted to doing so. The problem is people willfully writing and deploying programs which they neither understand nor can control.
computer, print awawa.
view more: next ›
In his very lukewarm defense, the rough CS equivalent to publishing a paper in a journal is presenting your paper at a conference. According to DBLP he has had three papers in AGI '06 '07 and '11 (which i would not call a serious conference) and one workshop paper (generally a tier below actual acceptance to the conference) at AAAI '15.