Do “smart” chatbots dream of electric sheep? Sure, why not? Recent research by ARSTechnica indicates that AI-powered chatbots may “lose their minds” when fed articles from ARSTechnica.

Simply put, while many popular chatbot services such as Cortana and Siri are programmed to handle natural language queries with relative ease, they can be easily confused if presented with articles that contain too much technical detail.

In fact, there are a few documented cases of chatbots throwing a virtual fit when asked a question that required the reading of a complex and long-form article. As it turns out, complex articles, such as those found on ARSTechnica (which is known for its tech-savvy content), are just too much for AI to handle.

Not surprisingly, this isn’t the first time chatbots have been perplexed by complex topics. Earlier research found that chatbots often struggle with abstract questions about morality, politics, and the meaning of existence.

So, it appears chatbot developers have some work to do if these AI-powered services are to be truly useful. While it’s understandable to feed chatbots with stories containing technical detail, the chatbot needs to be able to handle that complexity with grace and confidence in order to be truly useful.

Until that time comes, it looks like we’ll just have to rely on traditional tech support and online forums when faced with difficult questions on complex topics. After all, nothing beats the advice of an actual human being, right?