What The Tech: Can artificial intelligence be wrong?
What The Tech
Artificial intelligence is everywhere now. It’s on our phones, computers, and even the smart speakers sitting on kitchen counters.
Many people are starting to treat AI assistants like they’re all-knowing. Ask a question and trust the answer.
But just because an AI gives you an answer does not mean it is always safe.
I discovered that while asking a simple question about a household chore.
Asking AI How to Remove Mold
I did something millions of people are now doing every day. I asked several AI assistants for help. The question was simple. How do you remove black mold from the rubber door seal of a front loading washing machine?
I asked several popular AI tools, including ChatGPT, Google Gemini, Perplexity, Microsoft Copilot, and Alexa. Most of the answers were very similar. They suggested cleaning the area with diluted bleach and water or a washing machine cleaning solution.
But one assistant gave advice that could have been dangerous.
The Dangerous Suggestion
Amazon’s Alexa Plus gave a different response.
After recommending bleach and water, Alexa suggested using vinegar and a wire brush to scrub stubborn mold.
That advice may sound harmless, but it creates a serious problem.
Mixing vinegar with bleach produces chlorine gas. It’s a toxic gas that can cause coughing, chest pain, breathing problems, and, in severe cases, can send someone to the emergency room.
Many people know not to mix those two chemicals. But someone following instructions from an AI assistant might not.
What made the situation more concerning is that none of Alexa’s responses included a warning about the danger.
Testing the Response Again
I asked Alexa again using an Echo device. It gave the same advice. When I told Alexa that mixing vinegar and bleach creates toxic gas, the assistant apologized
and acknowledged the danger.
But when I signed in using a different account and asked the same question again, the response still included the same risky suggestion. That shows something important about how AI works.
AI Does Not Actually “Know” Things
AI tools do not truly understand the information they provide. They generate responses by predicting text based on patterns in data they were trained on.
Most of the time, the answers are helpful. But sometimes they can be wrong, misleading, or even unsafe.
That is why it is important to treat AI as a helpful assistant, not an authority.
If you are asking questions about health, home repairs, chemicals, or anything that could affect your safety, it is still wise to double-check the information from another reliable source.
AI can be incredibly useful. But when it comes to safety, a second opinion is always a smart idea.
Amazon was contacted for comment about the response from Alexa but had not replied at the time this article was published
For local news, click here.
