Re:OpenAI Trains Language Model, Mass Hysteria Ensues
by nl on Monday February 18, 2019 @06:21PM
(#19195397)
I've worked in the computational propaganda field, and I tend to agree that there is no real known defense yet.
I don't have a strong opinion about if they should have released this model or not.
I do know it would make a great commercial spam generator though. Want a million product reviews which seem legitimate quickly? This is the thing..
Yes, it can serve as great customized propaganda generator, and yes, people can be spin 'round and 'round with it. But they can be already with pretty much anything, from the simplest of phrases from "make X great again" to the elaborate scams of new age bullshit.
I simply disagree on the "virulence" or weaponization factor of this with others. (Especially when it comes to the possible "defenses", none can be "deployed" in 6 months. You can't teach critical thinking to billions of people overnight.)