The text argues the real problem isn't AI itself, but our uncritical use of AI-generated content. Developers readily accept AI’s answers without verifying sources, leading to problems. AI models, by design, don't indicate source origin and readily fabricate citations. Examples of real-world issues, like government reports with fake sources, highlight the risks. The solution involves applying established information evaluation frameworks like CRAAP. This framework emphasizes checking currency, relevance, authority, accuracy, and purpose. The text offers a practical workflow for verifying AI responses, including checking citations, cross-referencing information, and assessing the AI's purpose. It also points out when AI is less reliable, such as with recent information or citations. The text advocates leveraging existing review skills, like code review, to evaluate AI-generated information. A tool called Sabia, designed to evaluate sources using librarian-grade criteria, is promoted as a resource. The concluding message encourages critical thinking and information literacy to prevent the spread of misinformation via AI.