It's actually quite a lot better than a Google search. Yesterday I gave it the output of SQL ddl and told it to write me a script in Go that reads that from a CSV, then I asked it to generate a sample CSV to verify the script.
How is that different from googling? Lots of articles won't bother to include import and the structs they use definitely won't match up to your use-case because their dataset if different. What if my use-case is a bit strange and I need to embed the file instead of reading from the file system, for example? I can ask Chat GPT and it will update the example program, using my exact file name and the variable names for the problem I've described and the program runs as written!
I don't write much Go (so I don't actually care to commit the hello_world.csv ritual to memory), but I know enough about it to verify that the program doesn't have any glaring issues and make my own tweaks as necessary. Saved so much time for me in this scenario.
It can't replace search, actually. It doesn't know everything a decent search engine can find, and might not even remember the specific php command line parameter you're looking for.
What it can do is reason some pretty tricky interdisciplinary answers. It's about as useful as having an intern that's finished literally every white collar college out there but has no idea what they're doing. It takes some prompting to get useful work out of it but it is possible and it is very effective when you get it right.
The $20 sub has saved me quite a bit of time, the main challenge is remembering to set it to GPT4 for every damn conversation, as the previous models are trash.
How is that different from googling? Lots of articles won't bother to include import and the structs they use definitely won't match up to your use-case because their dataset if different. What if my use-case is a bit strange and I need to embed the file instead of reading from the file system, for example? I can ask Chat GPT and it will update the example program, using my exact file name and the variable names for the problem I've described and the program runs as written!
I don't write much Go (so I don't actually care to commit the hello_world.csv ritual to memory), but I know enough about it to verify that the program doesn't have any glaring issues and make my own tweaks as necessary. Saved so much time for me in this scenario.