Companies don’t exactly highlight the fact that their AIs are prone to hallucinating.

Funny enough, if you question Gemini enough, it will eventually cave and admit that it’s answers aren’t always right and that it’s still improving. LOL

The problem is, as I suggested, you already need to know what the correct answer is in order to effectively cross-examine what it told you.

“yeah, but there’s a lot of incorrect information online anyway.”

Here’s the thing: before AI, most information came from an author or organization, who had to stake their reputation on the content they create. If the information they provided was false, low quality, misleading, etc… they paid a penalty for it in a loss of credibility (and even income).

But with AI, that doesn’t happen. You can generate 1000 articles at the click of a button, post it everywhere, and there’s no backlash because the author doesn’t exist.

I think in the near future, you’ll start to see certification for human-generated content. I know that movies have started to disclose whether AI generated content was used or not, so the trend is that people want to know.

I think what’s feeding all this hubris around AI is that it’s essentially tricking us into thinking it’s intelligent. It’s an incredible tool for compressing and organizing information, but it isn’t really smart.

My son has Apple assistant (Siri), and I have Google Gemini. For shits and giggles, we had them talk to each other… literally have a conversation… and it got stale very quickly. There’s no “person” behind artificial “intelligence”, so you can see just how limited it gets.

I’ve always said that if you know a lot about a topic, you can very quickly see how AI is really stupid for the most part. The problem is that if you ask it a question that you don’t know the answer to, then it for sure seems correct, even when it completely hallucinates the response.

The danger is that not everyone has enough critical thinking skills to question the correctness of an answer, so they hear what Siri or Gemini told them as fact… and then pass that knowledge onto other actual human beings. Like a virus of misinformation.

Alexa used to answer questions asked, which was a cute novelty, but now she tries to sell something or rambles on at length about related topics instead of answering the question and shutting up.

My wife still uses Alexa to play music on a smart speaker we have. Sometimes, I’ll ask it to play music from an artist, then she’ll come back with “you need the upgraded plan”, blah, blah, blah. I ask again to play music from the exact same artist, and there’s no problem…

When tech gets in the way of the task, it’s failed technology, IMO.

What really sucks is that many of these AI features rely on cloud-based servers to work. So when those servers shut down, your phone loses a bunch of features overnight!

If AI features were only handled on-device (which is completely possible with modern hardware), then I’d be way more accepting of their use. Some features like “unblur” or “remove person” when editing an image is truly magical.