The email landed at 11:47 PM: "Urgent—need to discuss AI's impact on our team."
The engineering manager had just read another article predicting that AI would eliminate 70% of coding jobs within five years. His LinkedIn feed was full of developers sharing AI-generated code and asking if they were training their replacements. A competitor had announced they were "pausing engineering hiring to evaluate AI capabilities."
"Should I be looking for a new career?" he asked when we spoke the next morning. Genuinely panicked.
I asked him what his day-to-day work actually looked like. Yes, he used AI coding assistants for writing boilerplate, generating tests, and drafting documentation. Those tasks that used to take hours now took minutes. But what did he spend his time on? Understanding what customers actually needed. Designing systems that would scale and evolve. Reviewing code and mentoring junior engineers. Navigating organizational complexity to get projects shipped. Debugging production issues that involved multiple systems and teams.
"So AI handles the typing," I said. "But the thinking—understanding problems, designing solutions, collaborating with humans—that's still you."
"Yes," he admitted. "Actually, I'm more productive than ever. The boring parts are faster, so I spend more time on the interesting parts."
This is the reality beneath the panic. AI is changing software engineering—significantly—but it's not replacing software engineers. It's amplifying what engineers can do, shifting which skills matter most, and creating new demands while reducing others. The job market is adapting, and understanding how it's actually adapting helps both hiring managers and engineers navigate the transition.
At SmithSpektrum, I've watched the AI impact unfold across our client companies over the past three years[^1]. The pattern is more nuanced than either "AI will replace all engineers" or "nothing has changed." Understanding what's really happening helps you hire effectively in this evolving landscape.
What AI Actually Automates
AI coding assistants are genuinely excellent at certain tasks.
Boilerplate generation is largely solved. Writing basic CRUD operations, setting up project scaffolding, implementing standard patterns—these used to consume hours and now take minutes. An engineer who would spend half their time writing routine code now spends that time on something else.
Documentation generation has improved dramatically. AI can draft function documentation, README files, and code comments that would have been tedious to write by hand. The documentation still needs human review and refinement, but the first draft is essentially free.
Test generation is increasingly capable. Given a function, AI can generate reasonable test cases—not perfect coverage, but a starting point that's often 70-80% of the way there. Engineers review and expand, but the initial scaffolding is automated.
Code translation between languages works for straightforward cases. Porting code from Python to TypeScript, from Java to Kotlin, from one framework to another—AI handles the mechanical translation well enough that human work focuses on the edge cases and framework-specific patterns.
Simple bug fixes can often be suggested or even automated. Obvious errors, pattern mismatches, and common mistakes are caught and corrected with minimal human involvement.
These are real productivity gains. Engineers who adopt these tools effectively are substantially more productive—not by small margins, but by multiples for certain categories of work. The question is what happens to that productivity gain.
What AI Doesn't Automate
Equally important is what AI doesn't do well.
Understanding what to build remains fundamentally human. AI can write code to specification, but understanding what the specification should be requires understanding users, business context, and strategic priorities. The hard part of software engineering is rarely writing the code—it's knowing what code to write.
| Skill Category | AI Impact | Hiring Implication | Interview Adaptation |
|---|---|---|---|
| Boilerplate coding | Largely automated | Less important | Focus elsewhere |
| System design | AI assists, human decides | More important | Deeper questions |
| Debugging complex systems | AI helps, human leads | Still critical | Real-world scenarios |
| Code review | AI catches obvious issues | Judgment matters more | Evaluate reasoning |
| Requirements/scope | Purely human | Much more important | Behavioral assessment |
System design at scale requires judgment that AI lacks. How should this system evolve over time? What are the failure modes? How do we balance competing concerns? These questions require experience, intuition, and understanding of context that AI doesn't have.
Cross-functional collaboration stays human. Working with product managers, designers, customers, and other engineers requires communication skills, empathy, and the ability to navigate organizational dynamics. AI can't sit in a meeting and build consensus.
Novel problem-solving remains human domain. AI excels at pattern matching—if a problem has been solved before, AI can find and adapt the solution. But genuinely novel problems, problems that require creative approaches, problems that don't map to existing patterns—these still require human insight.
Legacy system understanding and migration involves judgment and context that AI struggles with. Why was this code written this way? What undocumented assumptions does it contain? What will break if we change it? Understanding legacy systems requires almost archaeological reasoning.
Debugging complex distributed systems involves intuition, hypothesis generation, and contextual reasoning that AI can support but not replace. When a production system is failing in ways that span multiple services, human judgment guides the investigation.
How Hiring Is Actually Changing
The skills that matter in engineering hiring are shifting, though not as dramatically as headlines suggest.
Higher-level skills are becoming more important. If AI handles the routine coding, what differentiates engineers is what happens above the code: understanding problems, designing systems, making architectural decisions, collaborating effectively. Senior engineering skills matter more; junior coding skills matter relatively less.
AI proficiency is becoming table stakes. Engineers who effectively use AI tools are more productive than engineers who don't. The ability to prompt effectively, to know when to use AI and when not to, to review AI-generated code critically—these are becoming expected skills rather than differentiators.
Judgment and taste are harder to find and more valuable. Anyone can generate code with AI; the question is whether the code is right. Engineers who can evaluate whether a solution is appropriate, who have taste about code quality and system design, who know when something is good enough versus when it needs more work—these skills are more valuable when raw production is easier.
Domain expertise is more important, not less. AI can code, but it can't understand your specific business domain, your customers, your technical context. Engineers who deeply understand the domain—whether that's healthcare, finance, e-commerce, or anything else—add value that AI amplification can't replace.
Communication skills are increasingly valuable. If AI handles more of the coding, engineers spend more time on the human parts of the job: understanding requirements, explaining technical decisions, mentoring others, building consensus. Communication skills that were "nice to have" become essential.
What Companies Are Actually Doing
Across our client companies, I see several patterns in how AI is affecting hiring.
Most companies are not reducing engineering headcount. They're getting more done with the same number of engineers—or the same done with slightly fewer engineers in some cases—but outright headcount reduction has been rare. The backlog of desired work expands to absorb productivity gains.
Some companies are raising the bar on technical interviews. If AI can pass basic coding interviews, the interview has to test something else—judgment, system design, collaboration, problem-solving. The interviews that test rote coding ability are less useful.
More companies are hiring for AI integration skills. This isn't "prompt engineering" as a separate role—it's the expectation that all engineers can effectively leverage AI tools as part of their workflow. Candidates who demonstrate AI fluency have an advantage.
Senior-level hiring remains strong. The leverage that senior engineers provide—making others more productive, making architectural decisions that avoid problems, mentoring junior engineers—is as valuable as ever or more so.
Junior-level hiring has become more competitive. Entry-level engineers face more competition for fewer positions in some markets. The tasks that used to train juniors—writing basic code, fixing simple bugs—are increasingly handled by AI, which raises questions about how junior engineers develop skills.
The Junior Engineer Question
The question of junior engineering is genuinely difficult.
Traditionally, junior engineers learned by doing: writing code, making mistakes, getting feedback, improving. The volume of code they wrote—even if much of it was routine—built skills and intuition. If AI handles much of that volume, how do juniors develop?
Some companies are addressing this by being more deliberate about junior development. They assign juniors to work that specifically requires human judgment—understanding user needs, debugging complex issues, working on code that involves historical context. They use AI assistance but ensure juniors are doing the thinking, not just copying AI output.
Other companies are hiring fewer juniors and investing more in each one. Rather than hiring five juniors and hoping three work out, they hire two and provide more intensive mentorship.
Some companies are creating new pathways for junior contribution. Product understanding, customer interaction, documentation and communication, tooling and developer experience—these areas require skills that AI doesn't replace and can be entry points for junior engineers.
The path from junior to senior engineering is genuinely changing. The specific skills that juniors need to develop are different than they were five years ago. Companies that figure out how to develop junior talent in the AI era will have an advantage; companies that simply stop hiring juniors will have expertise gaps in a few years.
What Engineers Should Do
For individual engineers, the implications are relatively clear.
Embrace AI tools. Fighting the tools or ignoring them will make you less productive than peers who use them. Learn to prompt effectively. Develop intuition for what AI does well and what it doesn't. Integrate AI into your workflow as a multiplier, not a replacement.
Invest in higher-level skills. System design, architecture, understanding users and business context—these skills are more valuable as coding becomes easier. The engineer who can figure out what to build is more valuable than the engineer who can only build what they're told.
Develop judgment and taste. Being able to evaluate whether code is good, whether a design is appropriate, whether a solution is elegant—these qualities are harder to teach and harder to automate. Cultivate them deliberately.
Build domain expertise. Deep knowledge of a specific domain—healthcare, finance, e-commerce, whatever you work on—compounds with AI tools. You can use AI to produce more in your domain, but the domain knowledge is yours.
Don't neglect the fundamentals. Understanding how computers actually work, how systems behave at scale, what makes code maintainable—these fundamentals matter more, not less, when you're evaluating AI-generated code. AI can produce code without understanding it; you need to understand it.
What Hiring Managers Should Do
For hiring managers, the AI transition requires some adjustment.
Adapt interview processes. If AI can pass your coding screen, your coding screen isn't testing what matters. Evaluate judgment, design skills, and collaboration—skills that differentiate engineers in the AI era.
Don't assume AI proficiency. Test it. Ask candidates how they use AI tools. Have them demonstrate their workflow. The difference between engineers who use AI effectively and those who don't is substantial.
Value judgment and taste. In interviews and on the job, look for engineers who can evaluate whether something is right, not just produce output. The ability to say "this AI-generated code is wrong because..." is valuable.
Think about junior development. If you're hiring juniors, have a plan for how they'll develop skills. Don't just assume they'll learn by doing the same things juniors always did—because those things are changing.
Don't overreact to hype. AI is changing software engineering, but it's not replacing software engineers. Make thoughtful adaptations rather than panic-driven changes.
The engineering manager who was worried about his career? I checked in with him six months later.
He'd doubled down on the human parts of his job—understanding customers, designing systems, developing his team. He'd become highly proficient with AI tools, using them to amplify his effectiveness. He'd focused on judgment and architectural thinking rather than code production.
"I'm more valuable than I was before AI tools," he said. "Not because I write more code—I probably write less code by hand—but because I make better decisions faster. The AI handles the typing. I handle the thinking."
He'd also changed how he hired. His interviews now focused more on design discussions, code review exercises, and problem-solving conversations. Less time on whiteboard coding; more time on evaluating judgment.
"The engineers I'm hiring now are different from five years ago," he observed. "They need to be better thinkers. They need stronger communication skills. They need to know when the AI is wrong. The bar is different, but the role is still essential."
References
[^1]: SmithSpektrum engineering hiring trends, AI impact analysis, 2023-2026. [^2]: GitHub, "Developer Productivity Survey: AI Tools," 2025. [^3]: Stack Overflow Developer Survey, "AI Tools Adoption," 2025. [^4]: McKinsey, "The Impact of AI on Software Development," 2025.
Adapting your hiring for the AI era? Contact SmithSpektrum for technical assessment design and hiring strategy.
Author: Irvan Smith, Founder & Managing Director at SmithSpektrum