Comment by raw_anon_1111 4 hours ago
If you look at the current hiring trends and how much longer it is taking developers to get jobs these days, a mid level ticket taker is definitely screwed between a flooded market, layoffs and AI.
By definition, this is the worse AI coding will ever be and it’s pretty good now.
I am really not convinced yet.
From all the data I have seen, the software industry is poised for a lot more growth in the foreseeable future.
I wonder if we are experiencing a local minima, on a longer upward trend.
Those that do find a job in a few days aren't online to write about it, so based on what is online we are lead to believe that it's all doom and gloom.
We also come out of a silly growth period where anyone who could sort a list and build a button in React would get hired.
My point is not that AI-coding is to be avoided at all costs, it's more about taming the fear-mongering of "you must use AI or will fall behind". I believe it's unfounded - use it as much or as little as you feel the need to.
P.S.: I do think that for juniors it's currently harder and require intentional efforts to land that first job - but that is the case in many other industries. It's not impossible, but it won't come on a silver plate like it did 5-7 years ago.
I mean it is online that major tech companies are have laid off a couple of hundred thousand people. What companies are going to absorb all of these people?
Anyone who hires can tell you one open req gets hundreds of applicants within 24 hours. LinkedIn easy apply backs that up.
I have two anecdotes from both sides. I applied for 200 jobs for a bog standard “C#/Python/Typescript” enterprise developer who had AWS experience. I heard crickets and every application had hundreds of applicants - LinkedIn shows you.
Did I mention according to my resume (I only went back 10 years) I had 10 years of experience as a developer including 2.5 leading AWS architecture at a startup and 3.5 actually working at AWS (ProServe)?
I had 8 jobs since 1996 and I’ve always been able to throw my resume up in the air and by the time it landed I would have three offers. LinkedIn showed that my application had hardly been viewed and my resume only downloaded twice.
Well everything I said above is true. But it was really just an experiment while I was waiting for my plan A outreach to work - targeting companies in a niche in AWS where at the time I could reasonably one of the industry experts with major open source contributions to a popular official “AWS Solution” and leaning on my network of directors, CTOs etc that I had established over the years.
None of them were looking for “human LLM code monkeys” that are a dime a dozen.
On the other hand, I’m in the hiring loop at my company. Last year we had over 6000 applicants and a 4% offer rate.
Who is going to absorb or need a bunch of mid level ticket takers in the future with AI improving? Or at least enough to absorb all of the ones who are currently being laid off and the ones coming in?
> By definition, this is the worse AI coding will ever be
This may be true, but it's not necessarily true, and certainly not by definition. For example, formal verification by deductive methods has improved over the past four decades, and yet, by the most important measures, it's got worse. That's because the size of software it can be used to verify affordably has grown, but significantly slower than the growth in the size of the average software project. I.e. it can be used on a smaller portion of software than it could be used on decades ago.
Perhaps ironically, some people believe that the solution to this problem is AI coding agents that will write correctness proofs, but that is based on the hope that their fate will be different, i.e. that their improvement will outpace the growth in software size.
Indeed, it's possible that AI coding will make some kinds of software so cheap that their value will drop to close to zero, and the primary software creation activity by professionals will shift precisely to those programs that agents can't (yet) write.