Isn't AI great? It uses extreme amounts of resources in form of GPUs, RAM, and energy. It also needs a lot of cooling, so that is another energy expense. But in return we have... chat bots that hallucinate between 30% to 79% of the time. No, really. They just make up stuff around 1/3 to 2/3 of the time. And that isn't even the worst of it. The worst part is that these mistakes compound. Most AIs don't correct themselves when they hallucinate. They don't fact-check their own output. And when you have a conversation with AI, every time you add to that conversation, you are sending the entire conversation as a prompt. Meaning that with each answers, the errors compound.

So, in the end, you get a string of errors built on top of errors.

And what I hate the most about it is that when you correct the AI it acts all nonchalant and innocent. "Oh, I'm sorry, you are right! You are not supposed to pour bleach into your eyes, actually. Sorry for the mistake."

AI needs to be accountable! And since it is not an independent agent, the companies that own the AIs need to be responsible for what the AI says and does. There are real consequences to this. Most people don't fact-check every single thing when asking AI for assistance. And they shouldn't be forced to do that, because at this point why even use the AI at all?

And by the way, we should stop calling these chat-bots "AI". Chat GPT, Google Gemini, DeepSeek, these are not actual AIs. Artificial intelligence (AI) is the capability of computer systems to perform tasks that typically require human intelligence, such as learning, reasoning, problem-solving, perception, and decision-making. Which, the current LLM models can't do. They can't learn, perceive, or reason, they just regurgitate already existing data. It's like talking to a book that knows only its own contents.