As the AI “gold rush” that stared with the release of ChatGPT continues, questions remain about the appropriate usage of the program.

Media outlets have been hesitant to adopt certain AI applications, leery of the potential high-visibility errors such programs can lead to. Outlets the Chicago Sun-Times and the Philadelphia Inquirer recently released a Summer reading list that comprised largely fake books generated by ChatGPT. A lack of proofreading on the part of both the author and editor resulted in an embarrassing, and completely avoidable, blunder that has added fuel to the fire when it comes to journalists’ reluctance to rely on software with a tendency to “hallucinate.”

Even AI optimist Zack Kass, the former Head of Go-To Market for OpenAI, warns that newsrooms need to treat such programs like they would an enthusiastic but less than reliable junior employee; not as a search engine or vetted source, but as something that absolutely requires scrutiny to use in order to avoid such public mistakes.