Lines of code is a poor metric for productivity, revenue, and profit. The question is: is AI making existing developer populations more efficient to where the demand trajectory for software engineers declines over time. I think we're too early to tell and that early indicators are the unsophisticated in power going "Magic robot makes you work better, use magic robot or we will replace you, and we're not going to hire materially more people until we approach what appears to be a failure mode." Bit of a collective industry death march based on vibes and cargo culting.
Neither. Hiring and firing is the primary game of software employment. It’s why so much corporate software sucks compared to open source alternatives under strong leadership. It’s the game most developers play instead of actually writing software.
Instead I really wish there were was more emphasis, like at my current employment, all focused on ownership, training, and retention.
The last time I asked an LLM AI to review my code, it added extra lines that weren't originally there. Then, it claimed that there were many bugs in those lines. When I pointed this out, it apologized. But I don't know how much time I wasted on this thing. I'm a developer with over 30 years of experience. Should I ease into vibe coding?
There’s almost nothing to be gained by writing about AI (particularly anything remotely positive). The anti-AI people are absolutely foaming at the mouth about how bad it is. Best to let them have their fun while the rest of us are quietly getting value out of it.
The problem isn't the overall demand for developers. The issue is that AI will replace entry-level developers, thus eventually reducing the ranks of experienced developers. A senior developer brings much more to bear on her work than simply coding skills. Where is she going to get that experience?
BTW, I also posted a longer version of this realisation here for what's worth: https://go.cbk.ai/more-work
The demand will be for developers to be phenomenal debuggers of massive ai code bases.
Time to retire. Unless one enjoys debugging instead of coding.
AI software developers are similar to self driving cars. Both are great for increasing stock prices, but not much more than that.
I can tell you anecdotally as someone who only does green field implementations which are proof of concepts/MVPs to train clients and to hopefully get a larger implementation down the road where we will need to bring in a team, before LLMs got “good enough”, I would have had to staff 1-3 junior engineers to do the grunt work while I did a lot of the higher customer facing, architecture, team lead type work, now I can do it all myself in the same amount of time.
The code that even ChatGPT generates is usually just as well structured using good software practices as I would have done - and I’ve been in the field professionally since 1996.
The only work I do manually is moving functions out into their own libraries and telling it to use my libraries to reduce duplicate code.
It’s going to hurt junior devs the most. I still have to do the higher level design work and tease out requirements.
No one is going to hire more developers, that's the entire point of AI. You'll do more work and provide more value to shareholders for the same if not less pay until you're rich enough to escape the hell-cycle of capitalism or you die.
There's a divide between Engineering a system, and implementing it in most other fields. We're now facing the situation where LLMs have artificially accellerated the trend we were already on, towards a professionalization of the art of Software Engineering. The title inflation of most of the programing jobs being incorrectly labeled as "software engineer" has obscured this trend.
The demand for actual Engineering skills won't go away anytime soon. The thing about Engineers is accountability and responsibility. If you're responsible for outcomes, you must exercise authority over the process. This gives you bargaining power for wages, which FAANG and others don't want you to have.