Jobs & Future of Work

No question about AI generates more anxiety than its impact on jobs. Will AI take over work? Which jobs are most at risk? What should I do about my own career? This module addresses these questions honestly — looking at what the evidence actually says, which is more nuanced than either the panic or the dismissal suggests.

What the evidence says so far

We now have several years of data on AI's actual employment effects. The picture is mixed:

• Productivity gains are real — studies consistently show AI assistance improves individual productivity by 20–50% on many knowledge work tasks. GitHub Copilot studies showed developers completing tasks 55% faster. Customer service agents with AI assistance resolved 14% more issues per hour in one major study.
• Job displacement is happening but slower than feared — sectors like customer service, data entry, and basic content creation are seeing headcount reductions in some organisations. But mass unemployment has not materialised, partly because demand for services expands as costs fall, and partly because transition takes longer than technology timelines suggest.
• Entry-level white collar work is disproportionately affected — the jobs most disrupted so far are not the manual jobs economists long predicted automation would take. It is the entry-level professional jobs — junior analysts, junior lawyers, basic programmers, content writers — that are most immediately affected by generative AI.
• New jobs are emerging — prompt engineer, AI trainer, AI auditor, LLM operations, trust and safety — roles that did not exist five years ago now employ hundreds of thousands of people globally.

Which tasks are most exposed

The key unit of analysis is tasks, not jobs. Most jobs contain both tasks that AI can automate and tasks that require distinctly human capabilities. A job's overall exposure depends on what proportion of its tasks AI can handle.

The inequality dimension

AI's labour market effects are not evenly distributed — and this is one of the most serious concerns. Several patterns are emerging:

• Geographic concentration — the benefits of AI productivity flow disproportionately to workers and companies in wealthy countries with strong AI access. Workers in countries that depend on outsourced knowledge work — including much of India's IT services sector — face more direct disruption.
• Education and income polarisation — AI tends to augment high-skill, high-income work (making lawyers and developers more productive) while replacing lower-skill, lower-income tasks. This can worsen inequality rather than reduce it.
• The missing middle — the traditional career ladder relied on entry-level work as a training ground for mid-level and senior roles. If AI eliminates entry-level work, how do the next generation of senior professionals develop? This pipeline problem is already visible in some professional services firms.
• Creative work — artists, writers, musicians, and illustrators face a particularly acute challenge because AI can now produce commercially competitive creative work at near-zero marginal cost. The economic model for creative professionals is under genuine pressure.

Sectors with the most change ahead

Some sectors are in the early stages of significant disruption:

• Legal services — junior lawyer work (document review, research, first drafts) is already being automated. Senior legal work requiring judgement, client relationships, and court appearances is more protected for now.
• Financial services — analysts, report writers, and back-office processing roles are under pressure. Client-facing advisory roles with genuine human trust components are more durable.
• Software development — AI coding tools are dramatically changing what individual developers can accomplish, but the need for software continues to grow faster than the supply of developers, so employment remains strong — though the mix of skills demanded is shifting.
• Healthcare — AI is augmenting clinicians rather than replacing them in most applications. Administrative and documentation tasks are being automated, freeing clinical time.
• Education — teaching requires human relationships and mentorship that AI cannot replicate. But administrative work, content creation, and basic tutoring face significant AI competition.

What this means for your career

The most honest career advice for an AI world is not to avoid AI but to work with it effectively, and to cultivate the capabilities that AI currently cannot replicate well:

• Develop AI fluency — understand what AI tools can do and use them. Workers who use AI effectively are not replaced by AI — they become more productive than those who don't.
• Invest in human skills — complex judgement, leadership, deep domain expertise, ethical reasoning, relationship building, and creative synthesis are more valuable as AI handles routine cognitive tasks.
• Be a domain expert first — AI tools are most powerful in the hands of someone who deeply understands the domain. A doctor who uses AI diagnostic tools well is more capable than before. A non-doctor using the same tool without expertise is not.
• Stay curious and adapt — the specific skills in demand will continue to shift. The ability to learn new tools and approaches is itself a durable skill.
• Think about the whole job, not just tasks — if AI can do 40% of your tasks, that frees 40% of your time for the work only you can do. The question is how to use that time most valuably.

The policy dimension

How AI's labour market effects play out is partly a policy choice, not just a technological outcome. Active debates include: whether AI productivity gains should be shared more broadly (through shorter working weeks, universal basic income, or stronger labour protections), how to fund retraining and transition support for displaced workers, whether AI-generated work should be taxed differently, and how to maintain the training pipelines for future professionals. These are live political questions in every major economy.