Mark Makary Blog

Introduction

Along with the benefits of the AI revolution came fears and panic centered around layoffs and job loss. While this is affecting both young and old, it seems to hit newly CS graduates harder than others.

I’m sharing my thoughts based on my own experience — both past and present. I’ve been in the field for a long time and I have witnessed many rounds of evolution. My goal is to ease fears and offer a more positive, realistic perspective. Hopefully, it helps others one way or another.

To start, a significant change is necessary. However, there is no need to panic—bots haven’t taken over all our jobs yet. It’s clear that technology is shifting how we operate, but it’s not replacing us entirely. To learn more, please read further.

What fundamentally changed?

For decades, software engineering centered on a straightforward craft: translating ideas into working systems by writing code. We took ambiguous problems, decomposed them into smaller pieces, and expressed logic in programming languages—one decision at a time, Personally, I have experienced the transition from assembly to COBOL, C, C++ and Java, I have seen the Internet and the Cloud Native shifts and the how Python is taking over the world, so how that AI thing differs from the rest? .

The fundamental engineering workflow has dramatically shifted. Increasingly, engineers can describe intent—requirements, constraints, desired behavior—and use AI to generate large portions of the implementation if not all of it.

This is more than just automation of repeated tasks.

It’s a redefinition of what “engineering” means.

Engineers will spend less time on repetitive construction. Instead, they’ll focus on designing systems, validating behavior, and making judgment calls — especially where context and prompt preparation matter.

Although, the job market demand is temporarily shrinking, still the profession isn’t shrinking.

It’s becoming more interdisciplinary and outcome‑focused.

Now, does this mean software engineering is shifting entirely to prompt engineering?

Partially — but there’s more to it.

Prompt engineering is no longer simple. The quality and relevance of the AI output depends heavily on how well the prompt is crafted. And it doesn’t stop there.

We’re entering a multidisciplinary phase.

Engineers now need to validate outcomes, improve results, and apply domain knowledge creatively. Success depends on how well you guide the AI outcome — not just what you ask it to do.

By early 2026, prompt engineering for Claude has evolved from simple instructions into structured, conversational workflows that maximize output quality by providing context, constraints, and iterative feedback. 

More on what this shift means for you in the next section.

What This Shift Means for You?

For software practitioners

• Adopt AI as part of your workflow: use it to draft code, tests, and docs—but keep ownership of correctness, security, and performance.
• Move up the stack: spend more time on requirements, system boundaries, data contracts, and reliability than on boilerplate implementation.
• Strengthen evaluation skills: get faster at reading unfamiliar code, designing test cases, and spotting edge cases and failure modes.
• Improve how you communicate design decisions: Learn how to quickly create design diagrams or documentation that explain trade-offs, risks, and why a solution fits the business goals better than another, utilizing the available AI tools.

For new computer science graduates trying to land a first role

• Show you can utilize AI to create solutions: build 1–2 small, complete projects (deployed app, API, data pipeline) with tests, a short README, and clear scope.
• Use AI, but be interview-ready: be able to explain your code, alternatives you considered, and how you verified correctness (tests, logs, profiling).
• Target “adjacent” entry points: QA automation, support engineering, data ops, implementation engineer, internal tools—then grow into core product roles.
• Make your resume evidence-based: “Built X that achieved Y” (latency, cost, users, reliability). Avoid course lists; highlight outcomes and ownership.
• Practice structured problem solving: clarify requirements, state assumptions, propose a plan, then implement—this mirrors modern engineering with AI in the loop.

Coding Won’t Disappear—But It Won’t Be the Center of the Job

AI is already accelerating routine parts of engineering—boilerplate, first-pass documentation, debugging suggestions, and early architectural scaffolding. What changes is not the need for engineers, but the definition of where their highest-value work sits.

That shift can be liberating, it reduces time spent on mechanical execution and increases time available for decisions that shape quality, safety, and business impact.

Engineers are increasingly valued less for volume of code and more for how effectively they can supervise, integrate, and improve intelligent systems—turning generated components into reliable products.

Coding remains valuable, but it becomes just one instrument in a broader toolkit. Clear and creative thinking becomes the differentiator.

From Builders to Orchestrators: Where Engineers Add Value

As AI handles more implementation, engineers will spend more time designing, validating, and governing system behavior—especially in complex, safety-critical, or regulated environments.

In practice, that means increasing emphasis on:

• Problem framing: turning ambiguous needs into clear, testable requirements.
• Constraint and guardrail design: defining boundaries, assumptions, and success criteria.
• System orchestration: integrating AI-generated components into coherent architectures.
• Human-in-the-loop oversight: identifying subtle failures, mismatches, and edge cases.
• Ethical and safety judgment: ensuring systems behave responsibly and predictably in real contexts.

In other words, engineers evolve from code producers to stewards of system behavior and outcomes.

How engineering roles would change in the new era?

Now we see “AI” is embedded in many titles, however I believe this is gradually is going to eventually disappear.

Still, the lines separating the current roles such as system analysis, architecture and engineering are going to be more blurry, time is going to show how this would change, focusing more on domain knowledge and experience, I just ignore this as a hype side effect for now.

How Computer Science education is going to change?

Computer science education is about to change more in the next decade than it has in the last fifty years — and the shift is deep, structural, and long overdue. 

While coding is still valid, it should shift more towards creative thinking vs syntax.

Computer science programs are about to shift dramatically. Instead of focusing on memorizing syntax and writing code from scratch, education will move toward systems thinking, architecture, and problem‑solving at a higher level. AI becomes the new baseline, so students will learn how to evaluate models, design prompts, build data pipelines, and understand multi‑agent systems.

Project‑based learning will take center stage. Students will spend more time integrating AI components, validating outputs, and iterating on real systems — not grinding through boilerplate assignments.

Conclusion: Engineering Becomes More About Direction Than Execution

AI does not diminish the importance of software engineers—it changes where they create leverage. As automation expands, human responsibility concentrates around direction, interpretation, and innovation.

In this environment, the strongest engineers increasingly operate as:

• Strategists
• Behavior designers
• Domain experts
• Guardians of safety and ethics

The profession is not going away. It is evolving into one of the most consequential roles of the intelligence age—because building reliable, responsible systems still requires human judgment, and we must learn how to adapt.