The entire article is based on the flawed premise, that “AI” would improve the performance of developers. From my daily observation the only people increasing their throughput with “AI” are inexperienced and/or bad developers. So, create terrible code faster with “AI”. Suggestions by copilot are >95% garbage (even for trivial stuff) just slowing me down in writing proper code (obviously I disabled it precisely for that reason). And I spend more time on PRs to filter out the “AI” garbage inserted by juniors and idiots. “AI” is killing the productivity of the best developers even if they don’t use it themselves, decreases code quality leading to more bugs (more time wasted) and reducing maintainability (more time wasted). At this point I assume ignorance and incompetence of everybody talking about benefits of “AI” for software development. Oh, you have 15 years of experience in the field and “AI” has improved your workflow? You sucked at what you’ve been doing for 15 years and “AI” increases the damage you are doing which later has to be fixed by people who are more competent.
I agree Copilot is trash and everything it generates is garbage. I have no clue how anyone uses it with any effectiveness. But I’ve had luck with Claude 3.7 in some aspects of code.
Still seems to falter heavily with anything CSS though.
What’s nice is, say I have to create 10 data processing scripts that are somewhat similar but not exactly similar, I can create one and then hand the task to AI to iterate on and verify for the other 9 while I work on something else. It’s a major time saver in that regard.
I still don’t think it replaces a developer, but for senior-level and above, when deployed wisely, I think it can reduce tediousness.
No, that should be a parameterized script (/unit test/function/what ever, just picking up your example). If you have a repeating pattern with slight changes “AI” can generate more of that (to some degree), but it cannot fix the code duplication. Every line of code written is a line of code that has to be maintained.
It’s actually one of the things copilot gets advertised for: see how great copilot can generate more of these repetitive unit tests? Yah, great, write more garbage faster. People need to know about test theories (parameterized tests) and think about what they’re doing.
So you copy your script 10 times with minor changes (or let copilot & co do it) and notice there’s some flaw in the script you started with; now you have to change 11 scripts - great.
Well it’s intentional duplication because they get deployed to separate instances, different resources types, different regions, using different resources, different logic, and different source triggers/types, and different destinations.
Fair enough. There might be some niche use cases where the results might be acceptable. But with everything I’ve seen I don’t trust “AI” with anything.
I’m with you there, I guess I don’t fully disagree. I have coworkers who use AI for like 80% of their work and I don’t get it. Half the time I feel like they spend more time figuring out what it did and fixing it versus they could’ve done it themselves from scratch faster (…or maybe they couldn’t?)
I’d say my use is closer to 10% for raw code and maybe 25% for intelligent tab completions. The black box of outputs just kinda weirds me out, even if I can see a diff.
The entire article is based on the flawed premise, that “AI” would improve the performance of developers. From my daily observation the only people increasing their throughput with “AI” are inexperienced and/or bad developers. So, create terrible code faster with “AI”. Suggestions by copilot are >95% garbage (even for trivial stuff) just slowing me down in writing proper code (obviously I disabled it precisely for that reason). And I spend more time on PRs to filter out the “AI” garbage inserted by juniors and idiots. “AI” is killing the productivity of the best developers even if they don’t use it themselves, decreases code quality leading to more bugs (more time wasted) and reducing maintainability (more time wasted). At this point I assume ignorance and incompetence of everybody talking about benefits of “AI” for software development. Oh, you have 15 years of experience in the field and “AI” has improved your workflow? You sucked at what you’ve been doing for 15 years and “AI” increases the damage you are doing which later has to be fixed by people who are more competent.
I agree Copilot is trash and everything it generates is garbage. I have no clue how anyone uses it with any effectiveness. But I’ve had luck with Claude 3.7 in some aspects of code.
Still seems to falter heavily with anything CSS though.
What’s nice is, say I have to create 10 data processing scripts that are somewhat similar but not exactly similar, I can create one and then hand the task to AI to iterate on and verify for the other 9 while I work on something else. It’s a major time saver in that regard.
I still don’t think it replaces a developer, but for senior-level and above, when deployed wisely, I think it can reduce tediousness.
No, that should be a parameterized script (/unit test/function/what ever, just picking up your example). If you have a repeating pattern with slight changes “AI” can generate more of that (to some degree), but it cannot fix the code duplication. Every line of code written is a line of code that has to be maintained.
It’s actually one of the things copilot gets advertised for: see how great copilot can generate more of these repetitive unit tests? Yah, great, write more garbage faster. People need to know about test theories (parameterized tests) and think about what they’re doing.
So you copy your script 10 times with minor changes (or let copilot & co do it) and notice there’s some flaw in the script you started with; now you have to change 11 scripts - great.
Well it’s intentional duplication because they get deployed to separate instances, different resources types, different regions, using different resources, different logic, and different source triggers/types, and different destinations.
So I disagree. Still, fuck Copilot.
Fair enough. There might be some niche use cases where the results might be acceptable. But with everything I’ve seen I don’t trust “AI” with anything.
I’m with you there, I guess I don’t fully disagree. I have coworkers who use AI for like 80% of their work and I don’t get it. Half the time I feel like they spend more time figuring out what it did and fixing it versus they could’ve done it themselves from scratch faster (…or maybe they couldn’t?)
I’d say my use is closer to 10% for raw code and maybe 25% for intelligent tab completions. The black box of outputs just kinda weirds me out, even if I can see a diff.
deleted by creator