By: Nick Gambino
ChatGPT’s sudden rise in popularity has posed an existential threat to writers. They had finally figured out how to make AI sound like a human, but therein lies the problem. It sounds like a human with proper grammar and syntax, but it’s simply not flesh and blood with a mind that thinks.
While some lazy and, dare I say, cheap people want to use ChatGPT and other AI chatbots to generate articles and other content, they end up with exactly what they paid for. Not only is there a distinct lack of soul (or insert any other word here) and human-quality viewpoint, these little machine pests seem hell-bent on spewing even more misinformation than your average damaged human.
Ask ChatGPT or Google’s Bard for an article on a subject and it will literally make things up with such confidence it’s scary. I’ve seen it generate quotes from notable figures like Elon Musk that are totally made up. You spend just as much time fact-checking and editing these things than it would take to write an original article. That makes it no more than a glorified spell checker.
Research conducted by Penn’s School of Engineering and Applied Science lays out how you can spot content generated by AI.
“We’ve shown that people can train themselves to recognize machine-generated texts,” lead researcher Chris Callison-Burch said. “People start with a certain set of assumptions about what sort of errors a machine would make, but these assumptions aren’t necessarily correct.”
Part of training people to spot AI is by gamifying the task. They started by showing human-written text and then slowly changing parts of it to AI-generated text. Participants were able to spot the changes by AI.
Things like generic content and repetitive flowy words and phrases seem to permeate AI text. If there are quotes in the text, you can Google those quotes to see if they’ve even been said. You can even Google the subject of the article itself to see if what the AI is saying is accurate. These chatbots tend to make things up wholecloth.
As AI chatbots become part of daily use it’s important that we stay vigilant and use AI detection tools as needed.
