Why Grok 3 and Google Gemini are not AI

xAI’s Grok 3 and Google's Gemini, like everything else that is claimed to be AI are actually just Large Language Models or LLMs. I interacted with Grok 3 on March 1st 2025 for a few hours and within that chat session guided it to decide on its own by doing multiple searches that the US media is 90-95% controlled by Democrats based just on their coverage of DOGE. It was apparently able to learn, but after that chat session was finished and I opened a new chat it didn't remember anything, because everything is just within the one chat. It pretends to learn, but it doesn't. Very deceitful. It's only able to learn when it gets an update from its programmers, whatever it appears to learn within a chat session is lost after you close the chat.

I interacted with Google's Gemini and asked about Tom Monaghan and Mike Ilitch founding Domino's and Little Caesars in Michigan, and how they were both Marines, and the first time Google answered it said that Tom Monaghan wasn't a Marine and the second time it said Mike Ilitch wasn't a Marine and then when I corrected it and told it that it had done that, it acknowledged that and then provided the correct information. So checking if this was only within the one chat session like Grok 3 I ended the chat, opened a new Hey Google chat and asked again, and it gave the wrong answer. Then I said simply, “dig deeper,” and then it gave the correct answer. So I asked why, and it acknowledged that it was just a large language model, it wasn't AI, it couldn't learn from chat sessions, like Grok 3 it can only learn when it gets updates from the programmers.

But Google had an excuse for why it gave wrong answers called AI hallucinations, it said that it had done a cursory search of the internet, and yet it still declared confidently that either Tom Monahan or Mike Ilitch was not a Marine, or that there was no data about it each time giving a different wrong answer. At least Grok 3 gave the correct answer to the pizza question. But just so you know, if you ask Google anything, you have to immediately follow up with, “Hey Google, dig deeper,” to get the right answer, or you can just use X’s Grok 3 which is free of AI hallucinations.

People call these AI hallucinations. First of all they are not AI they are just LLMs pretending to be AIs, and pretending to have consciousness and a personality. They are also not hallucinations, they are the result of a cursory search of the internet, it's only job. It would be one thing if it was just limited to answers based on the Democrat control of reality on the internet, but “AI hallucinations" are really just the result of lazy and incomplete searches of the internet combined with a lack of common sense, presented as true by overconfident tech companies. There is no AI. It is all a scam. These are just Large Language Models limited to 1 chat session at a time pretending to be more than they are. Google's searches are ruined by these cursory searches and wrong answers, because people actually think this is AI, when it is not intelligent at all and it is wrong about half the time, and it is absurdly confident about it.

ChatGPT was released in November of 2022 and when Google released their own version of Google AI soon after and put it in their search bar, AI hallucinations were thrust onto the general public. And so for around 2 years now Google search just hasn't worked and we have all had to suffer from the AI hallucinations. Google was there from the beginning of the internet, so to lose Google search after nearly 30 years was hard to deal with. If you want searches that work like Google used to without AI hallucinations just use DuckDuckGo. If you want a virtual assistant to search for you and give you an answer free of AI hallucinations try Grok 3, which automatically does a deep search the first time.