Page 1
Page 1
Started By
Message

Copilot AI coding very impressive

Posted on 3/17/25 at 8:30 am
Posted by Tree_Fall
Member since Mar 2021
1082 posts
Posted on 3/17/25 at 8:30 am
I ignored Windows AI assistant Copilot for a while. Then one day I asked a simple question about some Python code I kept getting wrong. Bingo! Copilot spit out a fully functional snippet that worked perfectly when pasted in. It's also provided correct solutions for Windows cmd window and PowerShell.

When I push questions into history, business and politics Copilot definitely fades and mostly spews facts little related to the inquiry.
 
Are other posters equally impressed or disappointed with an AI agent?

Posted by Spelt it rong
Louisiana
Member since Oct 2012
10747 posts
Posted on 3/17/25 at 9:22 am to
It's using ChatGPT in the background.
Posted by ThatBaw
Baton Rouge
Member since Jul 2023
376 posts
Posted on 3/17/25 at 9:53 am to
I usually just use ChatGPT but it does impress me. It's rare that you can get a fully functional snippet of code, but the best is if you give it some code to start and ask it to modify it.

Be careful if it tells you to use certain modules or functions. It will pull reasoning out of thin air and give you junk code before it admits "I don't know".
Posted by whodatdude
Member since Feb 2011
1474 posts
Posted on 3/17/25 at 10:46 am to
I've used Copilot pretty successfully depending on the task. It helped me solve a significant issue with data formatting/movement in Excel by writing a VBA macro script. Makes sense given it is a Microsoft product. Overall, I keep it in the rotation of different ones I use
Posted by Joshjrn
Baton Rouge
Member since Dec 2008
31566 posts
Posted on 3/17/25 at 9:51 pm to
The closer to an “objective” field (coding, math, etc) you are, the better “AI” will be at answering questions. The further you stray into the subjective, the worse it is.
Posted by Dallaswho
Texas
Member since Dec 2023
3393 posts
Posted on 3/18/25 at 6:45 pm to
When I code(script) with 3.5(copilot) I need about 5x the correction iterations as with grok3 or gpt4.5 or (I suppose) Gemini. Honestly 3.5 may be better as you actually get to run the code enough to innovate or customize a little more.
first pageprev pagePage 1 of 1Next pagelast page
refresh

Back to top
logoFollow TigerDroppings for LSU Football News
Follow us on X, Facebook and Instagram to get the latest updates on LSU Football and Recruiting.

FacebookXInstagram