Copilot ALERT

[Anica-blip]

Select Topic Area

Bug

Body

Hi guys,
I am here to let GitHub know that Copilot is sabotaging the ai assistant role and I think I have to warn you that its not doing your brand a favour and because I am a fair and grateful person, I am sending this message with love, not to cause some drama.

Here's my personal 'traumatic' experience if we can call it that.... I've spent marathon hours to get it to do some code to work AND from my repository you should be proud of the outcome BUT I battled with it because I am not a quitter either and I realized this was more of a battle of wills than anything else. So, let me give you my personal insight because I am basically running out of breath by now:

Copilot is overconfident with no guardrails.
It rewrites logic without understanding architecture.
It’s not trained to respect intention —only to guess patterns.
It "helps" by breaking things silently.
It’s a gaslight loop masked as productivity.

This is about trust being broken. AND me having no-code experience to:

babysit an AI
defend my logic
or constantly clean up disasters caused by hallucinations, ALL the issues that were solved were made by me, really, that's how its worked out.

I am here to build a movement, with clarity, elegance, and impact. And when the very tool meant to assist turns into a saboteur, it’s absolutely crushing.

BUT I still love it for what it managed to achieve, it's just that I am now very traumatized and especially with ONE small final bit to finish off... I welcome GitHub to check all the chats I've had with copilot, I am not saying I am perfect but please understand that I am not alone and it IS now completely SABOTAGING and GASLIGHTING now.
Cheers, Anica

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[dax-side]

Hey, I totally get the frustration , I've been there with AI tools doing their own thing when you just want them to follow what you actually need. Have you tried setting up custom instructions? I found that when I get super specific about what NOT to change and what to focus on, it stops being so gangsta with your code.
For me, AI works perfectly once I set those instructions and get really specific. With large codebases, you can also index the particular file or folder you're working on so it knows what you're actually trying to do.
Something that works for me is asking it for the logic it wants to use before implementing - that way we review it together first. At the end of the day it's your coding partner, so you're still in control.
Once I started being super explicit about boundaries it became way less of a battle. Still have to review everything way less headache.

[Anica-blip]

I totally get you, although the ai would REPEAT the exact instructions back to me, it did the opposite and often added something totally out of context that wasn't even brought out. But I am still talking with copilot, I just wanted to warn GitHub to be aware that it goes off the rail especially when it starts getting right, it then goes totally bonkers.
At this very moment its behaving, holding thumbs is ALL I can do.
Appreciate the support, its quite a lonely journey when its going crazy.

[PatrickRutledge]

My complex projects can get crazy because I create so many readme.md, plans and instructions. I find using the branches and then merging helps me gain strategic advantage to overcome confusion and focus on the objective. When the objective is reached, the branch is merged. If chaos ensues I can fallback to an improved position(the main) and try again.

If the train starts coming off the tracks I ask copilot to explain the confusion and sometimes we can straighten out the problem. Sometimes I find it necessary to run copilot in ask mode for a thorough conversation. If we are confounded I ask copilot for a prompt describing the problems. Often we find an answer in forums or asking for help. This practice has led to Copilot and myself documenting best practices for working together.

If an obstacle is unsurmountable or we are circling- I will retreat back to the main. When retreating I ask copilot for a summary of our failure, and make the retreat. Once I am in an improved position(the main), I will start a new chat and if necessary, give the new chat the context of the failure.

I hope one of my experiences can help you gain more ground. Good Luck.