Re: Data Annotation (link)
Posted by Manstuprator on 2025-May-14 18:06:18, WednesdayIn reply to Data Annotation posted by Enochian on 2025-May-14 13:02:14, Wednesday
|
This 50-minute read explains a lot of what it's all about. I give the archive.org copy because at the site your article views may be limited. There are a couple of ways that this may be a scam--in some cases, that is. They DO need real experts for some things, and do pay them well. If they require "potential acceptees" to perform an unpaid "training task" then those applicants essentially work for free. If a company can get enough people to do that, then why pay anybody for anything? My personal opinion about LLM is that they can NEVER really be even close to 100% accurate, and attempts to improve them are like a dog chasing its own tail. Human brains can do things that LLM machines/computers will NEVER be able to do. That's a fact! Sure, the output can be improved. But that's a task that never can be completed. One thing to keep in mind--training LLM is a great way to get venture-capital companies to risk their money on it! Remember, too, that it's YOUR money being wasted. Every time a company pours money down a hole, it becomes a cost that has to be covered by increasing profits (charging more) for other products the company provides. Breaking windows doesn't make economic sense, no matter who claims that it does. Comments? Or did I just waste an hour reading the linked article, and another half an hour editing this post for style and accuracy? Gee, maybe I should sign up with one of those companies to do the tasks they offer? They say there's one born every minute... M. Computers are the stupidest things around -- the do EVERYTHING they do based on just two things--ones and zeros... http://web.archive.org/web/20250513033524/https://nymag.com/intelligencer/article/ai-artificial-intelligence-humans-technology-business-factory.html [@nonymouse] [Guardster] [Proxify] [Anonimisierungsdienst] |
Follow ups: