The AI’s ability to (usually) create grammatically correct syntax is certainly impressive, but there’s no real intelligence behind the text, meaning the way you and I would figure out the validity of some statements or come up with original thoughts in response to a detail or an event isn’t present in AI-generated works. At the moment, this is being consciously proven World of Warcraft Community that managed to create fake advertisements to trick AI bots on certain websites into posting fake information as news.
“I’m so excited they finally introduced Glorbo!!!” reads a Reddit thread
Well it worked. This incorrect information led to this an article generated by the AI that ends up on a website by the name The portal
The AI text documents the excitement surrounding Glorbo as if it were a real thing, with a caption that reads, “World of Warcraft (WoW) Players are delighted with Glorbo’s introduction.” The text liberally quotes from the Reddit thread, which is filled with misinformation about the game, even resorting to absurd terms like “the aspect of the ducks”.
Although there are no live links documenting this and the article in question has since been deleted, The portal appears to have responded to its own coverage of Glorbo by publishing another article documenting how gamers are “reacting to the use of AI to extract content from their subreddit for popular gaming sites,” giving the game away, so to speak The portal is really powered by AI doing just that.
The article “Players react to AI-generated content” is no longer available. Moreover, it seems so The portal changed the title of the original Glorbo article to declare it “satire” before removing it altogether.
Even though the article was removed, I can only imagine what can happen (and probably will) if 2,000 upvotes are enough to publish an AI article about fake information when people out there are eager to cause a stir about other fake news. For now, Glorbo is a silly example, but demonstrably more insidious, intentional outcomes are possible here.