Spark Promotion AI Chat Join

Have you ever received an AI answer that was so wrong?

· 1 week ago · 9 replies · 158

I have asked ChatGPT about a local tax law before and it gave a response that sounds confident but it was completely wrong. I nearly use it in a work I was researching. I also ask it about the current exchange rate and it gave me a figure from 2023. Who else has caught AI giving the wrong response?

9 replies

· 1 week ago

Yeah I've had them just not tell the truth and I'm thinking they just don't have all of the facts. Also they are programmed and certain answers that the programmers don't like will be highly reflected in their answers as being wrong..

Grok was wrong about certain tax laws. I also found they will defend a stance because it is widely mainstream but criticisms or pointing out mainstream knowledge being wrong is always ostracized at worst or framed under “some critics argue” while the mainstream idea is accepted completely as fact.

· 1 week ago

Not very wrong, but still incorrect—especially in the past, when AI first emerged in chat form and the free versions lacked internet access. Nowadays, the vast majority have internet connectivity, and they generally don’t make such serious errors, nor do they make them as often.

· 1 week ago

I have had this happen a few times when asking for advice on something such as finance and investing. Had I not gone out of my way to check the information that ChatGPT told me, I would have made a huge mistake and ended up in a huge mess!

I check everything now, especially if it is a question I can not afford to be wrong on when using any AI.

· 1 week ago

On Oct 26, 2026, Resting Witch Face said:
I have had this happen a few times when asking for advice on something such as finance and investing. Had I not gone out of my way to check the information that ChatGPT told me, I would have made a huge mistake and ended up in a huge mess!
I check everything now, especially if it is a question I can not afford to be wrong on when using any AI.

But that is your major mistake; using ChatGPT is a terrible way to verify information. Unless you have the paid version—ideally the most expensive one—the information it provides is very poor. I recommend any other AI; I use Gemini and it works fantastically for me. There are also Copilot and Claude, which work wonderfully. I suggest you stop using ChatGPT altogether; it has become completely obsolete compared to the alternatives.

· 1 week ago

On Oct 26, 2026, maxwolfstein65 said:
But that is your major mistake; using ChatGPT is a terrible way to verify information. Unless you have the paid version—ideally the most expensive one—the information it provides is very poor. I recommend any other AI; I use Gemini and it works fantastically for me. There are also Copilot and Claude, which work wonderfully. I suggest you stop using ChatGPT altogether; it has become completely obsolete compared to the alternatives.

I have been meaning to look into Gemini, as I use the Chrome browser and it is built into it, but I got so used to using ChatGPT that I go to that now out of habit.

Next time I need to use AI for something, gonna remind myself about Gemini and check that out.

· 1 week ago

On Oct 26, 2026, Resting Witch Face said:
I have been meaning to look into Gemini, as I use the Chrome browser and it is built into it, but I got so used to using ChatGPT that I go to that now out of habit.
Next time I need to use AI for something, gonna remind myself about Gemini and check that out.

I get it—at a certain point, you actually start to grow fond of it, lol. But as I mentioned, unless you pay for the most expensive tier of their Pro plans, ChatGPT can be pretty clumsy with a lot of things. You’ve got Gemini, Claude, or Copilot, which all work wonderfully—I’d recommend any of those three. That said, Claude and Copilot are quite limited in their free versions, though they don’t suffer from the same "stupidity" as ChatGPT. Gemini, on the other hand, has virtually no daily limits; I use it every single day, and in several months, I’ve only hit that limit once or twice—to the point where you don’t even realize it’s there.

· 1 week ago

There was one time when it told me that Galactica (at Alton Towers) had closed down permanently (it hadn’t).

When I checked the source it had cited, it turned out to be an April Fools joke 😛 .

· 1 week ago

I always tell people to use AI in areas that they are very grounded in, else AI would just lead you astray with wrong answers. I have received wrong answers from AI tools countless times.

· 1 week ago

I have frequently received wrong answers from AI and this is the primary reason why I NEVER rely on a single AI tool. Receiving wrong answers from AI is quite common, so it is quite necessary to get your information verified before you assume that you have the correct answer. I use different AI tools to cross check my information.

Create a free account to join the conversation.
Sign up free