- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
We’re about to learn a painful lesson about delayed gratification in software engineering.
New data from China, 26,811 students tracked January 2023 through June 2025. Students using AI for homework saw their scores jump 20 percent. Completion time dropped nearly half. They aced the assignments.
Then exam season came. Those same students scored 20 to 40 percent worse when they couldn’t use the tool.
The homework phase is over. The exam phase is coming.
We’re doing this in software right now. Vibe coding feels incredible. Features ship fast. Nobody’s asking what happens in Month 18 when the original dev has left and nobody understands the codebase.
Commercial pilots fly with autopilot for most of every flight. They’re required to maintain manual flying proficiency regardless. If the system fails mid-air and the pilot can’t take over, people die.
Most teams using AI right now have forgotten how to fly manually. They’ve become passengers in their own systems. The autopilot flies, nobody checks instruments, and the first sign of trouble will be a breach notice or outage.
Three rules:
- Command the mission. Define architecture before prompting. Ambiguity kills in code and in flight. Delegate selectively. Offload mechanical work. Keep design and security reviews human. Verify everything. Audit before production.
- Never trust the automation without checking instruments.
- Quick wins feel good. Sustainable engineering feels boring. Boring keeps systems standing.
Organisations surviving the next two years won’t ship the fastest. They’ll be the ones who remember how to fly without the aids.
people insisting that you actually be skilled, independently of your tools, doesn’t make them Luddites. Rather, being unable to do so makes you a phony.


It’s funny watching this type of bellyaching play out like a loop over history.
People went through this EXACT same bellyaching when compilers first started showing up. The machine coders considered themselves superior and their skills indispensable. Now those people are extinct and modern people consider their compiler skills to be elite, not realizing they’re using the baby mode that people railed against in the past.
It turns out that you, in fact, don’t need to know what you don’t need to know, for the most part.
Did anyone do studies on the homework and exam scores of people using compilers vs people not using compilers?
Did anyone establish that people who used compilers are measurably dumber than the people who didn’t?
No?
Well then obviously it’s not the same situation, now is it?
Dont expect the pro-AI crowd to have any reasoning capacity. The only argument they have for AI is “I’m lazy”.
But you could never write code like this
Create webserver that responds with HTTP 418 I'm a teapotand have it compile it to anything. You still had to think of the logic and design.
Yes they did. Why did you assume they didn’t?
https://link.springer.com/article/10.1007/s40692-018-0103-3
It is mostly between students who used compilers and those who did it with pen and paper.
It did not have a significant impact.
spoiler
Easily found this study with the help of Duck.ai, try it yourself next time. Or try to wade through Googles SEO riddled bullshit if you care so much.
Then this is obviously different, because we are seeing a significant impact on test scores. Students that use AI test poorly.
…
I’m going to acknowledge that you responded so that, when I don’t continue this conversation, you’ll know that it wasn’t because I decided to abandon my position.
You could say your response did not have a significant impact.
Understanding the layers of abstraction down to the metal is still a useful way of developing an instinct for how compilers, interpreters, languages, and algorithms etc. work and what is possible. If you ever truly need to optimize something you’ll end up descending those layers, it’s why C has inline assembly and with RAMageddon upon us optimization may well return to being a valued skill.
Which is to say you don’t need to know what you don’t need to know, until you do, and you’ll be better anyway if you can.