Why AI hasn't replaced every 'automatable' job — yet

In an era where artificial intelligence (AI) was expected to revolutionize the job market, many roles remain resilient against complete automation. Despite predictions that AI would replace jobs like radiologists, software engineers, and pilots, these professions continue to thrive. Benjamin Todd, president of the nonprofit organization 80,000 Hours, explains that AI typically automates only specific tasks within a job rather than the entire role. Todd emphasizes a crucial point: while some experts foresee a future where AI dominates various sectors, the current landscape reveals a more nuanced reality. For instance, the complex nature of many white-collar jobs means that the straightforward tasks often represent just a fraction of the work. In the case of radiologists, a study published in the Journal of the American College of Radiology indicated that they spend only 36.4% of their time interpreting images, with the rest dedicated to consultations, patient care, and other responsibilities that machines cannot replicate. Software engineering has also experienced significant disruption from AI tools, yet overall employment in the field has increased since the launch of ChatGPT. Data from TrueUp, a tech hiring analytics firm, shows that software engineering job openings in tech companies reached over 67,000 in 2026, the highest in over three years. Todd notes that many software engineers are now producing code more efficiently than ever before, highlighting a shift in productivity rather than outright job loss. Moreover, the translator profession has seen a slight uptick in employment even after the rise of automated translation tools like Google Translate. This trend suggests that while certain tasks may be automated, the demand for human expertise remains intact. Todd warns, however, that the real threat to job security arises when AI systems advance to a point where they can manage substantial portions of work independently. He advises individuals to prioritize developing skills that are hard for AI to replicate, complement AI systems, and meet societal demands. His recent book outlines four key characteristics of valuable skills in the age of AI: they should be difficult for AI to perform, enhance AI capabilities, fulfill societal needs, and be challenging for others to master. Jobs that are likely to endure involve complex, unstructured tasks requiring human judgment, such as long-term strategy, social coordination, and relationship management. Todd cautions that while the transition to a more AI-driven workforce will take time, it is essential to remain vigilant as AI capabilities continue to evolve.