Claude code even added an auto mode so that you don’t get blocked by that pesky reviewing anymore. Since then, the usual mode of asking before running a command, for instance when the thing wants to read the entire codebase looking for information only available in an online doc, is now called manual mode; the non 10x developer mode.
But they needed those .env files to transplant an entire software stack from the developer’s laptop into production in a reproducible manner! How else were they ever going to get software into prod? By good documentation, broad version requirements, and following the Robustness Principle? Ha! How are you supposed to move fast and break things then?
It’s still so much faster. An experienced engineer spoon-feeding the AI the exact logic/algorithm to use is still faster than writing it by hand. I think a lot of people stopped early when AI was worse and/or before they started being good at using it.
When you spoon feed it the algorithm, it’s good, at least better than your average programmer. You just have to make sure it doesn’t have too much thinking to do. Honestly people in here really overestimate the talent of most dev.
There was a comment somewhere I really liked: people thought it would be one junior dev with AI replacing 10 senior ones. In reality, it’s one senior dev with AI replacing 10 junior vibe coders.
If you vibe code, sure you’ll get slop. Ask it to do something to get X result, you’re leaving it room to think so you get slop. Tell it exactly what logic you want it to use, step by step like it’s a child, then you get the same you’d do yourself but faster.
Look at the PRs for quality-critical projects like the linux kernel and I guarantee you coudn’t tell which PR has some AI and which does not.
The bad news for them is that AI is terrible at getting it ‘right’ the first time, so they have to do things over and over again guessing until it works. So the AI-heavy user pretty much has to just let the commands run to get any benefit since otherwise it’s spending 95% of its time waiting on user to approve commands. Further, it’s likely commands the user doesn’t understand, because they are asking AI to do it in the first place frequently because they don’t know themselves.
I’m with you, but as a result other people proclaim to be ‘better’ AI users largely because they trust the AI. Their stuff is crap and once in a while one of them blows themselves up, but in the short term they are getting praise from management.
Man I can’t wait for the bubble to pop and management no longer being hyped for the sake of hype over it.
letting your agent run commands without reviewing them first is peak stupid
Creating an environment that incentivizes not thinking is peak stupid.
Claude code even added an auto mode so that you don’t get blocked by that pesky reviewing anymore. Since then, the usual mode of asking before running a command, for instance when the thing wants to read the entire codebase looking for information only available in an online doc, is now called manual mode; the non 10x developer mode.
Cursor just added this too. Some auto mode which is “safe” because of some sandbox bs.
Took me 10 mins to find how to turn it off. And they keep trying to sneak it on again.
At least twice a week I have to cancel it when it asks to run something which will fuck things up.
deleted by creator
.envfiles are wild to me, environment variables have never been a good way to pass data to applications, let alone secret data.So the solution people came up with was to store them in plain text next to the binary, and then have a loader apply them before running the main app.
But they needed those .env files to transplant an entire software stack from the developer’s laptop into production in a reproducible manner! How else were they ever going to get software into prod? By good documentation, broad version requirements, and following the Robustness Principle? Ha! How are you supposed to move fast and break things then?
deleted by creator
What’s the point of AI if you need qualified review?
It’s still so much faster. An experienced engineer spoon-feeding the AI the exact logic/algorithm to use is still faster than writing it by hand. I think a lot of people stopped early when AI was worse and/or before they started being good at using it.
It’s faster, but is the end result good, or just “good enough, it works for now, and I get paid”?
When you spoon feed it the algorithm, it’s good, at least better than your average programmer. You just have to make sure it doesn’t have too much thinking to do. Honestly people in here really overestimate the talent of most dev.
There was a comment somewhere I really liked: people thought it would be one junior dev with AI replacing 10 senior ones. In reality, it’s one senior dev with AI replacing 10 junior vibe coders.
Ah yes, faster spaghetti. Wonderful.
If you’re getting spaghetti it’s a skill issue.
If you vibe code, sure you’ll get slop. Ask it to do something to get X result, you’re leaving it room to think so you get slop. Tell it exactly what logic you want it to use, step by step like it’s a child, then you get the same you’d do yourself but faster.
Look at the PRs for quality-critical projects like the linux kernel and I guarantee you coudn’t tell which PR has some AI and which does not.
Normally, speed. Especially if you aren’t a picky writer; which you should be; for scenarios like this one.
The bad news for them is that AI is terrible at getting it ‘right’ the first time, so they have to do things over and over again guessing until it works. So the AI-heavy user pretty much has to just let the commands run to get any benefit since otherwise it’s spending 95% of its time waiting on user to approve commands. Further, it’s likely commands the user doesn’t understand, because they are asking AI to do it in the first place frequently because they don’t know themselves.
I’m with you, but as a result other people proclaim to be ‘better’ AI users largely because they trust the AI. Their stuff is crap and once in a while one of them blows themselves up, but in the short term they are getting praise from management.
Man I can’t wait for the bubble to pop and management no longer being hyped for the sake of hype over it.