A reminder to NOT use AI as your primary source of information, certainly not Google Insights

We already know a case where a person asked and believed the first Google result (the first result is also a generated answer) about whether he needs a visa to Australia. (He needed what he would have learned from the government site about traveling to Australia, ie about the third result. However, AI confidently said that he did not need them).

Here we have a really bizarre case of nonsense that was hallucinating about a non-existent dental emergency, which is mediated by the local police (?), but enough people believed even such obviously incorrect information and acted according to it, that the Municipal Police made a post about it on FB.

Folks, as someone who uses AI on a daily basis for code completion (which is literally the easiest thing AI can do), I know that when I press "enter" and the correct paragraph in yamli will be completed for me, which I don’t have to write by hand, it will give me a dose of serotonin. And I feel so good about it that I almost forget how many paragraphs I had to delete before because the AI ​​has a hit rate of ~60%, which is a little better than a coin toss.

Every day I get angry when a guy tells me that he is all happy "you found out from chatGPT how to install a gas stove".

I know that in the last 20 years we have learned that it is necessary "vygooglit" you are the answer to our questions, unfortunately google doesn’t work anymore. If you have children in your neighborhood or work in education, do not rely on children "they know how to google"it is necessary to teach them that they must skip the first result, then skip all sponsored results, and up to the fifth and subsequent results can be relevant. They just read the ai summary, which is mostly wrong.

Have you noticed that AI works best if you ask it about things it doesn’t know? Because otherwise you will very quickly start to feel the 60% hit rate 🙂