Yeah I’d agree with this…
AI can be pretty good at writing code… (sometime it misses references or things like that, or misunderstands prompts, so you need to debug it yourself), but when it comes to the actual process/design of the code it can be pretty hit and miss, in my experience.
Often, although the code it gives you is syntactically correct, and sometimes works right away for what you need it to do (although, especially for more complex things, just as often it doesn’t), it can be way off the mark in terms of the actual design/architecture - until you point out a more efficient/effective way to do things, and have it re-written.
In my experience, using AI to write code works best when you tell it the actual code you want it to write (then it’s much more likely to output something that works), and just let it write it - rather than using it as a way design the code (in which case what it comes up with is sometimes inefficient, illogical, and sometime not even close to working).
The more specifically you tell it what to write, the better it is. It also makes it easier to debug and/or modify when it’s giving you the code you’ve specifically asked for, rather than something that’s totally different from what you intended to use.
To add… it’s also really useful for checking, and finding syntactical/reference/logical errors etc., in code you’ve written yourself, when something’s not working as expected…
2 Likes
Forum
Academy
Marketplace
Showcase
Pricing
Features
