OECD Finds Human Judgement and Social Skills Still Hardest for AI to Replace

A new OECD report warns that AI is rapidly transforming jobs, with clerical and routine office roles facing the highest exposure as AI systems become better at language, information processing and repetitive tasks. However, occupations requiring human judgement, social interaction, creativity and physical adaptability remain far harder for AI and robotics to replace.

OECD Finds Human Judgement and Social Skills Still Hardest for AI to Replace
Representative Image.

A major new report from the Organisation for Economic Co-operation and Development (OECD) says artificial intelligence is rapidly transforming the global workforce, with office and administrative jobs facing the highest exposure to current AI systems. The study, developed by the OECD's Centre for Educational Research and Innovation under its Artificial Intelligence and the Future of Skills programme, introduces a new way of measuring how close AI is to performing the tasks required in different occupations.

Rather than dividing jobs into "safe" and "unsafe" categories, the OECD says AI exposure depends on the specific human capabilities needed in each profession. The report argues that some jobs are vulnerable to language-based AI systems such as ChatGPT, while others may eventually be affected by robotics and machine automation.

Clerical and Routine Jobs Face the Greatest Risk

According to the report, occupations built around repetitive, structured and codifiable tasks are currently the most exposed to AI. Jobs such as typists, payroll clerks, bookkeeping clerks, file clerks, data-entry workers and medical transcriptionists ranked among the occupations with the smallest "capability gaps," meaning existing AI systems are already close to matching many of the skills required for those roles.

Researchers found that AI performs especially well in areas such as language processing, information management, classification and documentation. This explains why office and administrative support occupations recorded the highest overall exposure in the study.

The OECD says the findings reflect the rapid spread of generative AI tools into workplaces, where they are increasingly being used to summarise information, draft documents, handle customer communication, and automate repetitive office tasks.

Human Judgement Still Remains Hard to Replace

While routine office jobs appear highly exposed, occupations requiring judgement, social understanding and physical adaptability remain much harder for AI to replicate. Professions such as judges, firefighters, psychiatrists, surgeons, police officers and chief executives ranked among the least exposed occupations in the report.

The OECD says these jobs rely heavily on capabilities where AI still struggles, including contextual reasoning, responsibility, negotiation, empathy and decision-making in unpredictable situations. Many also require physical dexterity and real-world interaction that current AI and robotics systems cannot easily perform.

The report stresses that this does not mean such professions are immune to technological change. Instead, it suggests that AI may support or reshape parts of these jobs without fully replacing the human role.

AI Exposure Is Not the Same for Every Occupation

One of the report's key findings is that AI exposure should not be treated as a single issue. The OECD identifies two major technological trends happening at the same time. The first is the rise of language-based AI systems capable of handling communication, reasoning and information processing. The second is the slower development of robotics, machine vision and embodied AI that can interact physically with the world.

This means different occupations face different forms of disruption. Lawyers and accountants may experience change due to generative AI systems that process documents and data, while manufacturing, transport, and warehouse jobs may increasingly be affected by robotics and automation technologies.

The report argues that future labour-market disruption will depend not only on how advanced AI becomes, but also on which capabilities improve first.

OECD Calls for New Education and Workforce Strategies

The OECD says its new "AI Capability Gap Index" should be used as a policy tool to help governments, businesses and education systems prepare for future changes in work. Researchers argue that understanding which human capabilities remain difficult for AI to reproduce will become increasingly important for workforce planning and training.

The report suggests that education systems may need to place greater emphasis on skills such as creativity, social interaction, critical thinking and adaptability rather than focusing mainly on routine knowledge tasks that AI systems can increasingly perform.

At the same time, the OECD warns against assuming that AI will automatically lead to widespread job losses. The organisation says the real impact of AI will depend on factors such as regulation, workplace adoption, labour costs and broader social decisions about how technology is integrated into economies and workplaces.

Overall, the report concludes that AI is unlikely to replace all jobs equally. Instead, the coming years are expected to bring a gradual reshaping of occupations, with human workers remaining most valuable in areas where judgement, responsibility, adaptability and human interaction continue to matter most.

  • FIRST PUBLISHED IN:
  • Devdiscourse
Give Feedback