← Listen againIn Conversation
How to Build a Career Alongside AI October 6, 2026 · 4:49 · With Maya, Omar, and Leo
With me are Leo, who follows technology and digital life, and Omar, who follows the economy and work. Thanks for joining us. We’re looking at what research can tell workers and…
Transcript Maya You’re listening to Slow English FM. This is our round table on work and technology. I’m Maya. With me are Leo, who follows technology and digital life, and Omar, who follows the economy and work. Thanks for joining us. We’re looking at what research can tell workers and students about changing jobs and AI tools. Maya Omar, let’s begin with the labor market. What did the OECD find, and what does it mean for people choosing or building a career? Omar The OECD examined online job vacancies in ten countries. It found that occupations with high exposure to AI included software developers, financial analysts, managers, and administrative workers. Exposure means AI may affect tasks in that kind of work. It does not, by itself, tell us that a whole occupation will disappear. Maya So the finding is about where AI may affect work, not a list of jobs that are certain to go away. What else did the OECD see in those vacancies? Omar Vacancies in highly exposed occupations often asked for management, business, and digital skills. The OECD brief also reports rising demand in these occupations for emotional, digital, and social skills, along with cognitive skills such as reasoning and problem-solving. Leo That mix matters. Digital skills are part of it, but the findings also include social and reasoning skills. For someone thinking about a career, I would read that as a reason to look beyond a single tool. A tool can change. The ability to understand a task and communicate about it may matter across different kinds of work. That is an interpretation of the skills the OECD reports, not a promise about what any employer will require. Maya Omar, does that match the picture from the vacancy data? And what should listeners be careful not to conclude from it? Omar It matches the skills listed in the brief. But the OECD also cautions that wider digitalization may help explain some of the changes. In other words, a shift in the skills employers request cannot automatically be credited to AI alone. The report describes patterns in vacancies. It does not tell us exactly what a particular worker should study or what will happen to one person’s job. Maya Leo, let’s turn from job vacancies to people using AI. What does Anthropic’s research show about coding work? Leo Anthropic analyzed 500,000 coding-related interactions with its products. It classified 79 percent of conversations on its coding agent, Claude Code, as automation. On Claude.ai, it classified 49 percent that way. In this analysis, automation means the interaction was classed as the AI doing coding work, rather than only helping a person do it. Maya Those are large shares of the interactions in that analysis. Omar, what can that tell us about careers, and where does the evidence stop? Omar It tells us how Anthropic classified coding-related interactions with its own products. It does not tell us how often all software teams use AI, or whether those teams have stopped hiring. We should keep the measure close to what it actually counts: product use, not employment across the software industry. Maya Omar, the OECD names several skills employers request. Leo points to a gap between tool use and a whole job. How would you bring those ideas together without turning them into a guarantee? Omar I would treat them as two different kinds of evidence. The OECD looks at vacancies and the skills they request. Anthropic looks at interactions with its coding products. Together, they give people questions to explore, not a formula for job security. A student could ask what digital skills are used in a field, and also how reasoning or communication fits into its work. But these findings do not identify one course or skill that guarantees a job.
