ChatGPT is far superior to Github Chat at the moment. I can give the same prompt to both, with no other context, and GPT produces a far more correct answer generally. Iterating on what it produces is also a major shortcoming at the moment for got Github Chat. Also, regarding the 'non programming questions', I've had it erroneously flag programming questions as non-programming questions. Quite annoying.
They cheaped out on the underlying model and gave it no reasoning enhancements. These are day zero features that didnt need till a "beta" to be added. Incredible trolling by the devs...
Github Chat feels like a huge downgrade compared to ChatGPT. One of the things that bugs the most (and what I like about ChatGPT) is that it doesn't seem to understand context in follow up questions. When I ask it do something and then ask it to refine the output, it threats the follow up question as a new separate question and responds with unrelated nonsense
I treat it as a teenage boy that has the internet at his fingertips but close to no wisdom at all. I find i spend more time telling it the code it provided won't work than actually getting good code out of it.
Heh, in some cases I think this is very true. It's knowing where to use it. I guess that a lot of the time right now that probably is gruntwork. I've also used it to refactor some fairly gnarly typescript.
![CHILE vs. BRASIL [1-2] | RESUMEN | ELIMINATORIAS SUDAMERICANAS | FECHA 9](http://i.ytimg.com/vi/5ObJt_71AYc/mqdefault.jpg)
Most important for me is how good it is to generate unit tests. Also how about an integration test?
ChatGPT is far superior to Github Chat at the moment. I can give the same prompt to both, with no other context, and GPT produces a far more correct answer generally. Iterating on what it produces is also a major shortcoming at the moment for got Github Chat. Also, regarding the 'non programming questions', I've had it erroneously flag programming questions as non-programming questions. Quite annoying.
True
They cheaped out on the underlying model and gave it no reasoning enhancements. These are day zero features that didnt need till a "beta" to be added. Incredible trolling by the devs...
Github Chat feels like a huge downgrade compared to ChatGPT. One of the things that bugs the most (and what I like about ChatGPT) is that it doesn't seem to understand context in follow up questions. When I ask it do something and then ask it to refine the output, it threats the follow up question as a new separate question and responds with unrelated nonsense
ChatGPT will do the same with longer conversations too.
@@IanWootten Yes, but copilot chat ALWAYS does this, even if it's in the next question. There is no comparison.
Not a fan so far, it's still more efficient for me to copy and paste my code to gpt4
I treat it as a teenage boy that has the internet at his fingertips but close to no wisdom at all. I find i spend more time telling it the code it provided won't work than actually getting good code out of it.
Heh, in some cases I think this is very true. It's knowing where to use it. I guess that a lot of the time right now that probably is gruntwork. I've also used it to refactor some fairly gnarly typescript.