3 Will AI take my job?

The chapter explores fears that AI will “take our jobs,” with a focus on software engineering, and concludes that the impact is uneven. It proposes three job qualities that tend to resist automation: a wide excellence gap between “just fine” and outstanding work; the need for stringent validation of outputs; and the need for tight control over what is produced. Rather than predicting wholesale replacement, it argues AI will mostly absorb repetitive, formulaic tasks while elevating the value of human judgment, collaboration, and problem framing.

Across creative, professional, and technical domains, work that prizes originality, stakeholder engagement, and long-term impact is harder to automate than fast, rehashed, “good enough” output. Where correctness truly matters—because of legal, safety, or commercial risk—humans remain in the loop to verify AI, often limiting productivity gains. In software, small mistakes can be catastrophic, so validation practices (tests, reviews, QA, deployment safeguards) are indispensable; AI can write code that compiles, but it still hallucinates, misassumes APIs, and requires debugging by experts. Generative systems shine when loose control is acceptable (e.g., disposable graphics), but when precise, exacting outcomes are required, prompting approaches begin to resemble step-by-step engineering, reinforcing the need for human-led specification and integration.

The chapter’s guidance is to move beyond “just coding” toward solving real business problems and delivering robust systems end to end: learn the language of the business, work in cross-functional teams, practice test-driven and trunk-based development, and build DevOps and cloud skills. Roles that tightly couple technology and business (operations research, full‑stack data science) and niches with scarce, up-to-date expertise are better insulated. Expect rising demand for AI product validation and benchmarking, and be cautious of “toy” work where clients accept loose control and shallow quality. Looking ahead, AI may also create opportunities to detect and elevate authentic, human content online. At the macro level, while transitions can be painful for displaced workers, productivity improvements from automation historically raise overall prosperity rather than cause prolonged mass unemployment.

• Jobs in which excellent work is noticeable and appreciated are likely to better resist AI. In software engineering, one way to achieve that is to work at the intersection between technology and businesses—you partner up with businesspeople to help them build successful products, as opposed to just coding. Some suggestions to accomplish that are the following:
• Learn about business, for example, by reading popular business books like Lean Startup.
• Be involved in tasks beyond technical ones, such as product management or business analysis.
• Specialize in a technical field designed to help businesses attain their goals, such as operations research and data science.
• Jobs which require a stringent and detailed validation of the output, either for safety, legal, or commercial reasons, will better resist AI. This tends to be the case with software, but some people may want to turn engineers into “validators” of AI-written code. Some ideas to remain relevant are the following:
• Become a person whose job is to deliver reliable software, not just code.
• Work in modern teams that constantly intertwine coding and testing. If you haven’t already, learn about modern engineering practices like test-driven development (TDD) and trunk-based development.
• Learn how to build and software holistically, including deployment, cloud, and DevOps.
• Learn good practices to evaluate AI-based products (see Chapter 5). Help your clients understand the performance of AI at their intended tasks and help them find opportunities for improvements.
• Jobs which require fine control over the output may better resist AI, as AI is most helpful when you’re ready to accept its outputs without having to refine them much. Some ideas to protect your job are the following:
• Avoid specializing in building rough prototypes or toy products, such as required by early-stage entrepreneurs for idea validation.
• Try to work with picky clients who want things done in a certain way. You might have the highest chances of finding this kind of client when building software for mission-critical applications.
• The proliferation of generative AI may create new software jobs to build algorithms that will help cope with the newly generated content, for example, by helping filter and retrieve content that’s written by humans.
• Those whose professions are affected by AI will suffer through the transition to a new profession. However, if AI were to truly automate jobs in an efficient way, rather than cause prolonged unemployment and misery, it will make our economies more efficient and prosperous.

FAQ