Is the ChatGPT fervour premature?
ChatGPT’s success, or at least its ability to generate public interest, has led some writers, including Will Knight, to cast doubt on the credibility of ChatGPT and question what it is capable of. Will Knight, writing in Wired’s January 13, 2023 issue, is the latest to do so. \”ChatGPT has Investors Drooling, but Can it Bring Home the Bacon?\”
In this article, he makes a couple of observations that deserve closer scrutiny. One of them I find to be valid and the other I believe harks to a Dreyfus like What Computers Can’t Do mentality. Both can be seen to be examples of Schadenfreude.
Wright concludes the article with a valid point he learned from Phil Libin, who was CEO of Evernote between 2007 and 2015. Wright summarizes some of Libin’s concerns, saying that ChatGPT, and other generative AI, are created by scraping human-made content from the internet, but they are now contributing more to text and images on the Internet. He [Libin] states that all of these models will soon f*** up their own training data. We’re about be inundated with a flood of bullshit.