Fifteen years in this industry taught me one thing: we're exceptional at creating expensive problems disguised as innovative solutions.
And no, before you ask, I'm not talking about AI. Or rather, I'm not just talking about AI.
The Bermuda Triangle of Development
You know that old saying? "Fast, cheap, good - pick two."
Well, in reality, too many companies pick "fast and looks like it works" and pray nobody asks hard questions.
Money isn't flowing like it did during the loosest funding cycles. And no, that isn't simply AI's fault. Capital became more expensive, growth assumptions changed, and years of underpriced engineering risk became visible. JavaScript is not the culprit; weak architecture and wishful operations can be built in any language.
I work with clients who show up with systems completely blown up. The stack? A chaotic symphony of WordPress with Node, Node with Python, Python with C#, all assembled by under-resourced teams making local decisions around whichever tool they knew.
The result, in cases I've seen? They effectively pay for the same application two, three, sometimes four times through rewrites and rescue work.
The Silver Bullet Seduction
Remember Web3? Those inflammatory posts saying that if you didn't learn blockchain, you should change careers?
And the metaverse? Facebook changed its company name to Meta and invested billions. The work produced real hardware and software, but commercial adoption has not matched the scale or speed of the original vision.
Now it's AI. And yes, it's powerful. But I've also rewritten AI-generated code several times. Detailed specifications, carefully constructed prompts… and still, spaghetti code that creates technical debt.
The inconvenient truth? The economics of foundation models and many AI products remain unsettled. Large investment and impressive capability do not guarantee durable margins or useful adoption. The transition has to be deliberate, otherwise you accumulate risk, resistance, and frustration.
The Paradox Nobody Wants to Admit
Here's the bitter irony: some companies use expected AI productivity as one reason to lower salaries or hire fewer juniors.
If the industry sharply reduces junior opportunities today, it weakens the pipeline of experienced engineers tomorrow.
And without deliberate mentoring, who will develop the judgment needed to maintain critical systems, evaluate generated work, and create the next generation of technical knowledge?
Stack Overflow, once the Mecca of developers, has low participation relative to its readership. It is not “practically abandoned”: in its 2025 developer survey, 82% of respondents said they visited at least a few times per month, while 68% said they rarely or never participated in Q&A. That gap is the warning. AI systems and search depend on human knowledge ecosystems, but reading an archive is not the same as replenishing it.
The Doomsday Cult of Tech
We have to learn to live with perpetual doomsday in our field. There's always a new apocalypse at the door, always a technology that will "destroy jobs" or "revolutionize everything."
Reality? Our field generates enormous economic value, and employers still have incentives to reduce labor costs and strengthen their market position. AI can be genuine infrastructure and a convenient story for decisions that were already financially attractive. Both can be true.
Some technology leaders position themselves as oracles of the modern era, promising they'll “optimize” everything. Sometimes the translation is straightforward: replace labor with something cheaper. Sometimes the tool genuinely removes drudgery. Workers should ask who captures the benefit and who carries the risk.
So, What Now?
If you're in this field because you thought it would be smooth sailing, you need to revise your paradigm.
It can be a well-paid field, but there is no guarantee for every role, region, or career stage. It is also draining. We constantly change businesses, technologies, and paradigms. We have to work under leaders who sometimes mistake buying engineering for understanding it.
My advice? If you love it, study. Work on your technical skills, yes, but also on your interpersonal ones. Networking, communication, and negotiation can make a real difference to the opportunities you find and the conditions under which you work.
Get interested in architecture and in concepts that transcend trendy language or frameworks. Write about what you learn (I do it on Medium, about architectures and technologies I use with clients).
The Uncomfortable Truth
Tech isn't going to end, and software work is unlikely to disappear as a category. Individual tasks, roles, and careers can still be displaced.
Jobs will change, and adaptation will matter. But outcomes will also depend on investment, labor policy, access to training, geography, and demand - not only individual attitude. The ILO’s 2025 assessment found job transformation more likely than full automation across exposed occupations, while warning that exposure is uneven.
As long as there are companies wanting "fast and cheap" and discovering the hard way it was expensive, as long as there are legacy systems nobody understands but everyone depends on, as long as there's the next "revolution" needing to be integrated into the imperfect real world…
There will be work. That does not promise a painless transition or a place for everyone without deliberate training and opportunity.
Just don't expect it to be easy. It never was. And anyone who tells you otherwise is trying to sell you something.