#Absolutely useless. "Searches" for answers, still wrong

1 messages · Page 1 of 1 (latest)

rugged oliveBOT
#

Reported by @light rapids

Bug Report: Absolutely useless. "Searches" for answers, still wrong
`Steps to Reproduce`
  • Ask a simple question. "Why does my screen get washed out when I take a snap shot?"
  • Wait for it to "search the web" (*doubt)
  • Get answer that is 100% incorrect, from premise, to explanation, to solution.
`Expected Result`

Well, I expect correct answers. Especially to issues that have been discussed online, which it made me wait while it "looked up" the answer to.

`Actual Result`

It spews utter bullshit. Not a single statement is correct.

`Environment`

Windows 11, Chrome browser.

#
Additional Information

Please provide relevant details to help resolve the issue, such as:

  • ChatGPT Shared Link (if applicable).
  • Screenshots or videos demonstrating the problem.

-# ➜ Need to contact support? Visit the OpenAI Help Center.

light rapids
#

Message link. From the very first response, everything was wrong.

That “washed out” effect is intentional. In Windows 11, when you use Snipping Tool or Win+Shift+S, the OS overlays a translucent white/gray filter across the screen. This indicates screenshot mode is active and helps you distinguish the area you’re about to capture.

Why it looks washed out: The overlay is a semi-transparent layer, not an actual change to brightness or color. It just dims or whitens everything behind it.

Why it’s inconsistent: Sometimes the overlay doesn’t appear (for example, if you’re using Print Screen or a third-party capture tool). Other times it does (Win+Shift+S or opening Snipping Tool).

Result: What you see is harder to read, but the final screenshot will look normal, with the original colors intact.

The behavior is inconsistent, so no, it's not "the tool" trying to "help me" understand the area I'm about to capture. It's the entire screen, and it appears before I make the capture area.

I don't care if it is a semi-transparent layer, the fact it dims and whitens everything behind it is the problem.

The explanation for why it's inconsistent is wrong. It does, in fact, appear when using Print Screen. And it does appear when using the Snipping tool.

The "result" statement is completely wrong. The final screen shot does not, actually, "look normal." If it did, I wouldn't give a crap, I wouldn't have asked, and I wouldn't be here typing furiously because this piece of flaming garbage can't even do a web search correctly.

https://chatgpt.com/c/68bb4a30-5400-8328-aec8-cb34c928a7ff

normal locust
#

And also it always It automatically starts searching if you use even a question mark at the end of your message. That's annoying.

normal locust
#

Again ist searches automaticly if certain words are used or question mark is used.

sudden shard
#

@light rapids You're raging on a bot. If the answer is wrong it's likely that it's not understanding what you're talking about.

Give it a screenshot (from your phone). It might go, "Oh THAT, sorry, I was talking about something else No, THAT is due to..."

This is the same approach one would hopefully use with about human being, before screaming "your answer is utter BS!"

Tell it to try again. Tell it the answer is probably not in it's training data. Tell it to look for a more in-depth and accurate answer. Tell it the exact version of Windows that you're using and ask if there's might be a recent update that caused this behavior to break. Tell it that you want evidence-based facts and not a response based on commentary in a forum.

Work with the technology.
Or don't.

normal locust
#

It just keeps doing the trash with searching. Please finally fix!

light rapids
# sudden shard <@346732865624735744> You're raging on a bot. If the answer is wrong it's likely...

Thank you for assuming the problem is me and that I haven't "tried again." Also, this isn't a human being. A human being would learn and understand. And if it didn't do what I wanted, I'd get someone else to do it. But this isn't a human, so I don't have that option, do I? Before you say "go use another LLM, hur dur," just stop. I paid for this service. I expect it to be useful. It used to be useful.

So, yes, I'm well aware that I'm barking at a bot. Do you genuinely think I don't understand that? The problem is that it is an insanely infuriating tool that makes promises that are not kept. And I post the links and screen caps here so hopefully someone that can do something about it sees how frustrating using this tool can be.

Why are you even commenting here?

"Hey it's a bot" Yes, I know. Not helpful.

"Oh, just try again and tell it it's wrong" Yeah, no kidding. Did you read anything I wrote? Telling it that it's wrong, doesn't work. That's a huge contributor to why I'm so frustrated with it.

"Work with the technology. Or don't" Wow, how insightful and pithy. Do you feel smart writing that?

You contributed nothing to the conversation. Either help, or don't. Personally, I'd rather you not "help," because you don't.

#

Also, I'm not ON my phone when I use it. Nice assumptions you're making there. Clearly we don't all use the product the same way, and you lack the insight to understand that. Once again, don't try to help, because you aren't helping.

#

Finally, just to be clear, this section is for bug reports. I'm not asking for help (as you've clearly established that you can't help). I'm posting here because there is a problem and I want the developers to know about it. Just because you don't have the problem, doesn't mean there isn't a problem.

sudden shard
#

I said use your phone to get a screenshot, as in, take a picture of the Windows screen, because it's obvious to both of us that you can't get a screenshot from the Windows side when the problem is with screenshots, rite?
C'mon man, be a little more balanced. Let's at least have a little chuckle with that one.

sudden shard
#

Notes...

The Me14 guy has never said anything good here about anything. He jumps to +1 everything negative, rejecting positive outcomes. Go ahead, confirm if you care. That should give you some indication of character, motivation, and relative value of his comments.

I've spent a lot of time to help people here (and elsewhere for decades). Most have found my brand of direct non-sugar-spiced help to be actually helpful because I get right to the problem, and it usually leads to positive results.

If you really care about solving the problem, you'd want to exchange notes with anyone qualified to help (yes, I'm qualified, but no longer willing).

So rather than just letting anger be your guide, consider that someone with this history, myself or someone else, might actually be on your side in seeking solutions.

I offered options beyond yelling at the toaster, (which you've made clear you're ok with that), and tried to suggest that swearing in a bug report isn't conducive to getting eyes on the problem. That's good advice no matter who it comes from.

If you don't like offers for help, a better response might be "thanks but I'd like to wait for official Support". It's polite and conforms to one of the forum rules: be respectful.

To be respectful, I will gladly acknowledge what you've already said, you won't hear from me again on your topics.

You might want to know though, OpenAI doesn't process reports here. This is a community server. We, community, all of us, ARE "Support" here. I guess like most people you haven't read the info available.
I'd offer to help you to get your issues to OpenAI, but it's clear you don't want my help. Me14 knows all of this, has been through it a couple times and has seen the process work, but he chose not to tell you, again preferring (your) chaos over (your) peace.

SMH, feels bad to watch this happen again.

Best of luck to you.