David Heinemeier Hansson

Programmers in the tech industry are facing layoffs and a more precarious job market, despite the industry's overall wealth and prosperity. The future of human programmers in light of AI and automation is uncertain.

Developers are on edge

Golden run or gold rush? Surely DHH remembers a time when programming was less accessible to the masses. I would not attribute the current downturn in developer jobs to unreleased tools like Devin. Finance drove the massive dot com layoffs and the end of ZIRP is driving these.
But the writing is on the wall, things are changing. Type and memory safe languages changed things, as did cloud based hosting and architectures. Neither are ubiquitous in the codebase today.
AI developed code might reach adoption at a faster rate, but at this point we still haven’t produced a fully AI developed app using the technology. Ultimately the industry needs to demonstrate a major win with an AI generated project before the next gold run can begin.
But back to accessibility, where we’ll start to see the erosion of jobs. The easier to learn tasks will be automated away first and the net effect will be fewer developers entering the field. The more difficult tasks will still require more experienced or more highly trained developers for quite some time.
Humans will desire to understand AI models and what they produce for as long as possible. And for now, this means more highly skilled and specialized jobs, not less.

I think it is way to soon to blame AI for the downturn. Most of the downturn was due to overhiring and financial worry. The economy has not been healthy for a while and there is still a scarcity of developers even though there have been layoffs. Good developers are always hard to find. While I do agree with the author with not worrying about Devin, I do so for a different reason.
<ol>
<li>
Devin was demoed but has not been seen in practice. This could be entirely like Google Gemeni (which was all smoke and mirrors to begin with, let alone all the issues it currently has).
</li>
<li>
I think if Devin does live up to what it hopes it will be more of a tool for engineers not a replacement. I don’t think Devin is going to move to a place where it will replace, or obsolete developers. If Devin attains sentience that might not be true but then we have bigger issues if that is the case.
</li>
<li>
These same worries have plagued every industry, so this isn’t new. AI has yet to live up to it’s dreams. Don’t believe me… look at the self-driving car. That hasn’t actually got to a place where it is working consistently, or affordably. Heck many AI models still have issues detecting a cat vs dog let alone replacing an engineer.
</li>
<li>
Lastly AI is as good as the data you feed it. While you expect it to “learn”… that learning is usually in terms of feeding it more data. Running those types of models also gets very expensive.
</li>
</ol>
Again AI is a tool, not a replacement. Even though there is hype on Devin, it is not cost effective nor efficient to say get rid of all your devs and let the AI overlords rule. There will always be a place for engineers, but the cheese is moving and we have to change with the times.