Thank you for your contribution. Unfortunately I do not have sufficient expertise in LLM engineering to provide a useful comment, but this is the sort of research I'd like to see here instead of LLM-driven unemployment hype.
I think you're too concerned with how others speak to believe that someone can actually write good, well-written comments. Obviously, in the age of LLMs, we might use one or two things to produce a better response. But that doesn't mean I don't think about what I wrote. Especially since English isn't my native language. Anyway, don't worry about me!
Reading the article would have shown which one it is, “The agency is floating a set of rules that require companies to offer U.S.-based representatives”, where “representatives” refers to people.
> I'm more inclined to believe that this case is getting amplified in MSM because it fits an agenda.
I mean tech in general has been negatively covered in the media since 2015 due to latent agendas of (a) supposed revenue loss due to existence of Google/FB etc and (b) to align neutral moderation stances towards a preferred viewpoint most suitable to the political party in question.
There is a solution, however, anyone hoping to roleplay with models submits an identity verification, an escrow amount, and a recorded statement acknowledging their risky use of the model. However, I assume the market for this is not insignificant, and therefore, companies hope to avoid putting in such requirements. OpenAI has been moving in that direction as seen during the 4o debacle.
But how would your solution have helped in this case?
The guy was probably a paying user, so Google would have already known who he is. He's also 36, so no excluding him based on age. And neither the escrow nor the statement really add much in my view
Unfortunately this project at its present state is just me coming across such comments and adding them to the list manually. Contributions are mostly a way to get more eyes, but I am aware that some people will misuse it towards that end, and I will review them too to reach that conclusion independently.
I would love to set up a more automated pipeline, but detecting slop over short sequences is in general a difficult problem.
If I go by the contributor numbers on Github, I see Claude has committed something on the order of 300,000 lines of code. I don't think it's reasonable to review that much code, even in weeks worth of time.
I haven’t needed to do such a thing in a while, but my “rule” for explaining how unreasonable is to say “if I only glanced at each line of code for 1 second, without bothering to understand the details, it would take me 3 and a half full 24hr days non stop to simply look at”. So it’s definitely more than 1 work week because presumably other stuff is going to need to get done in that time too. Actually understanding it is going to be at least a multiple of that, and probably the multiple is ~30x.
reply