In my mind, I thought AI relies on the facts it was trained on. But yesterday I realized that assumption was wrong. Modern search engine AIs just skim the surface of the Internet and feed you the most viral content. Doesn't matter how many books they have trained it on, AI favors the virality of the content over its factuality.

Here's the story. If I'm not sure about a topic I'm writing, I sometimes use AI to fact-check things before I write it in an article. But turned out I can't rely on AI for that either. Yesterday, I was writing a story about a topic I learned from AI a month or so ago, and I wanted to fact-check some historical records. Google AI clearly told me there were no such records. I'm like, what? A month ago, you told me about that.

Then I built up the conversation from other related facts, and finally AI admitted it was wrong. Then it said, "I only relied on the top search results for your search term and completely ignored other records." So that's it, my guys. Our supposed-to-be genius robot buddy turned out to be the same guy who had Google open up in the class quizzes.