From suggesting glue on pizza to recommending non-existent law citations, artificial intelligence hasn't quite mastered human logic. Here are the funniest instances of AI giving objectively terrible advice.
The Glue in the Pizza Incident
Recent updates to search engine AI summaries led to one of the most bizarre culinary recommendations in internet history. When a user asked how to keep cheese from sliding off pizza, the AI suggested adding about an eighth of a cup of non-toxic glue to the sauce. The error stemmed from a decade-old joke on a Reddit forum which the model misinterpreted as a legitimate cooking tip.
Legal Citations That Never Existed
Legal professionals have also learned the hard way that AI can exhibit confidence without accuracy. In a now-famous court case, an attorney used an AI tool to conduct research, only for the judge to discover that the cited precedents were entirely fabricated. The AI had essentially written fan fiction for the legal system, creating convincing but completely non-existent case names and docket numbers.
Creative Problem Solving Gone Rogue
Generative AI often struggles with spatial awareness and basic physics. Users asking for home improvement tips have received instructions that would defy the laws of gravity or result in electrical fires. One bot suggested cleaning a toaster by submerging it in water while plugged in to ensure a 'deep clean,' failing to recognize the lethal nature of the advice it was providing.
The Future of Fact-Checking
These incidents highlight the 'black box' nature of large language models. Because they function by predicting the next likely word in a sequence rather than understanding facts, they can produce 'hallucinations' that sound perfectly rational. As these tools become more integrated into daily life, these viral mishaps serve as a reminder that human oversight remains the most important part of the process.
Written by
Viralized Staff
