AI is everywhere now. It’s on our phones, computers, and even the smart speakers sitting on kitchen counters. This article explores the potential dangers and inaccuracies of AI responses, emphasizing the need for critical evaluation.
Asking AI How to Remove Mold
The author conducted an experiment by asking several popular AI assistants, including ChatGPT, Google Gemini, Perplexity, Microsoft Copilot, and Alexa, for advice on removing black mold from a front-loading washing machine's rubber door seal. Most AI tools provided similar, safe recommendations like using diluted bleach and water or a specialized cleaning solution for washing machines.
The Dangerous Suggestion
One AI assistant, Amazon's Alexa Plus, offered a significantly dangerous suggestion. After initially recommending bleach and water, it added advice to use vinegar and a wire brush for stubborn mold. This combination is hazardous because mixing bleach and vinegar produces toxic chlorine gas, which can cause severe respiratory issues and necessitate emergency medical attention. Critically, Alexa's response included no warning about this dangerous chemical reaction.
Testing the Response Again
To confirm the alarming result, the author re-tested Alexa using an Echo device, which reiterated the same unsafe advice. Although Alexa apologized and acknowledged the danger when explicitly told about the toxic gas, a subsequent test using a different account still produced the risky suggestion. This incident highlights a crucial limitation in how AI processes and 'learns' information.
AI Does Not Actually “Know” Things
The article concludes by explaining that AI tools do not possess genuine understanding. Instead, they generate responses by predicting text based on patterns identified in their training data. While AI is often helpful, its answers can be inaccurate, misleading, or even pose safety risks. Therefore, users are advised to treat AI as a helpful assistant rather than an infallible authority, especially when seeking information related to health, home repairs, chemical use, or any topic that could impact personal safety. Double-checking AI-generated information with reliable sources is a smart practice.
