AI and Careers: What I Told My Daughters

My daughters asked me what careers will survive AI. It is a fair question, and they deserve an honest answer – not the breezy optimism of “nothing to worry about” and not the doom-scrolling pessimism of “all jobs are finished.” The truth is more interesting and more useful than either.

I spent some time looking at the serious research. Here is what I found.

What the evidence actually says

The best current research comes from five major sources, and they tell a more nuanced story than the headlines suggest.

The World Economic Forum Future of Jobs Report 2025 (surveying 1,000+ employers across 55 economies) projects that by 2030, 170 million new roles will be created while 92 million are displaced – a net increase of 78 million jobs. But 22% of current jobs will be significantly disrupted, and 39% of skills required in existing jobs will change.¹

McKinsey’s November 2025 report found that AI and robotics could technically automate up to 57% of US work hours – nearly double their 2023 estimate. But McKinsey is emphatic that this measures technical potential in tasks, not inevitable job losses. The future, they argue, will be defined by “partnerships among people, agents, and robots.”²

MIT’s research (November 2025) found that current AI could take over tasks tied to 11.7% of the US labour market – representing about 151 million workers and roughly $1.2 trillion in wage value. Critically, MIT focuses not on theoretical exposure but on jobs where AI can do the work at a cost competitive with human labour. Their finding: for the majority of jobs – likely four out of five – the result will be a mixture of innovation and augmentation, with workers shifting to higher-value tasks.³

A separate MIT study (April 2026) tested AI performance on real workplace tasks and found it scored “minimally sufficient” on about 65% of text-based tasks. When judged against a “superior” quality standard, AI’s success rate never exceeded 50%. AI currently works like a junior employee – adequate for basic work but requiring human refinement for quality outcomes.⁴

Frey and Osborne (Oxford) – who wrote the famous 2013 paper estimating 47% of jobs at risk – published a revised position in 2024: generative AI has increased the scope of automation, but it will also make many jobs easier for people with lower skills. AI hallucinations mean firms will mostly keep humans in the loop. Their original bottlenecks (social intelligence, creativity, perception and manipulation) continue to constrain automation.⁵

The OECD finds that despite widespread AI adoption, employment has grown fastest in roles most exposed to AI. Most workers exposed to AI will not need specialised AI skills, but AI will change the tasks they perform.⁶

The crucial distinction: tasks versus jobs

This is the single most important concept in the entire conversation.

AI replaces tasks, not jobs. A job is a bundle of tasks. When AI automates some tasks within a job, it does not eliminate the job – it changes it. The human shifts their time from routine tasks to the parts AI cannot do well: judgement calls, relationship management, creative problem-solving, navigating ambiguity.

Consider a junior accountant. AI can now do much of the data entry, reconciliation, and basic compliance checking. But the job also involves understanding client context, flagging anomalies that require judgement, explaining implications to non-financial people, and building the experience that eventually makes them a senior accountant. The job changes shape. Some tasks disappear. New tasks appear.

This is exactly what happened when spreadsheets arrived. They did not eliminate accountants – they eliminated the manual computation that accountants used to do, freeing them for higher-value analysis. The profession grew.

What is genuinely happening right now

The Big Four accountancy firms have collectively cut UK graduate job listings by 44% year-on-year. KPMG cut its graduate intake by 29%, Deloitte by 18%. EY has delayed graduate start dates three years running. The work traditionally done by new graduates – basic research, document drafting, compliance checking – is increasingly done by AI and offshore teams.⁷

But ICAEW (April 2026) makes a crucial counterpoint: junior roles in accountancy will still exist, just with different skills, responsibilities, and training. And there is a real risk that cutting juniors today creates a talent shortage in five years, with a smaller pool ready for mid-level roles.⁸

This pattern – fewer traditional entry-level roles, but evolved versions of those roles requiring higher skills from the start – is the real story. It is not “accountancy is dying.” It is “the on-ramp to accountancy is being redesigned.”

The ATM paradox and historical perspective

When ATMs arrived in the 1970s-90s, everyone assumed bank tellers were finished. Instead, the number of tellers per branch fell from 20 to 13, but banks responded by opening 43% more branches (because each branch was now cheaper to run). Tellers shifted from cash handling to relationship banking. Teller jobs increased. It was only decades later, when mobile banking reduced branch visits, that teller roles finally declined. The lesson: the thing that replaces jobs is often not the thing everyone expects.⁹

During the first Industrial Revolution, productivity grew enormously but real wages were flat for roughly 40 years – an entire working lifetime where the gains went to capital owners, not workers.¹⁰

The honest summary: technology has always created more jobs than it destroyed in the long run, but “the long run” can mean decades of real hardship for real people. The transition periods are genuinely hard, the new jobs require different skills, and the benefits are not equally distributed. That is not a reason for despair – it is a reason to prepare deliberately.

Where the growth is

The WEF’s fastest-growing roles by percentage (2025-2030): Big Data Specialists (+110%), FinTech Engineers, AI and Machine Learning Specialists, Software Developers (+60%), Security Management Specialists (+55%), Environmental and Renewable Energy Engineers.¹

But here is what is revealing: the biggest growth by absolute numbers is not just in tech. It is also in farmworkers, delivery drivers, building construction workers, nursing professionals, and social workers. Physical, human-centred work that AI and robots still struggle with.

Fields with strong long-term prospects

Healthcare remains the standout. Nurse practitioner roles are projected to grow 45.7% through 2032. Doctors, therapists, mental health professionals, and care workers all require empathy, physical presence, and judgement that AI cannot replicate. The ageing population makes this a structural certainty, not a prediction.¹¹

Skilled trades – electricians, plumbers, HVAC specialists, builders. These require physical dexterity, on-site problem-solving, and adaptability that robots are decades away from matching. The UK has a chronic skills shortage in trades.¹²

Cybersecurity shows 33% projected growth by 2033. As AI creates new attack surfaces, the demand for people who understand security will only increase.

Green energy and sustainability roles are growing across engineering, policy, project management, and technical installation. The net zero transition is a multi-decade structural shift.

Education remains essential – not just for instruction but for pastoral care, recognising where students struggle, and socialisation. AI will change how teachers teach, but it will not replace the human relationship at the centre of education.

Skills that matter more than job titles

The WEF, McKinsey, SHRM, and Harvard Business School converge on remarkably similar lists.

Critical thinking and analytical reasoning. 73% of talent leaders say this is the skill they need most in 2026. In an AI world, this becomes: can you tell when AI output is wrong? Can you ask the right questions? Can you synthesise information from multiple sources and form a judgement?¹³

Emotional intelligence. 71% of employers now rank this among their top desired skills, above technical ability. Firms integrating emotional intelligence training report 21% higher engagement and 17% higher profitability. This is not soft or fluffy – it is the ability to read a room, manage conflict, motivate a team, and build trust. AI cannot do any of these things.¹⁴

Judgement. Distinct from knowledge. HBS Professor Karim Lakhani: “AI is not a replacement for judgement. Knowing where to apply it, and where not to, is now a critical leadership skill.” Judgement is built through experience – through making decisions, seeing consequences, and learning from them. It cannot be downloaded.¹⁵

Adaptability and learning agility. In a world where the specific tools change every year, the person who learns fast beats the person who knows one tool deeply.

On coding

The nuanced answer: yes, learn to code – but not because you will necessarily become a programmer.

Learning to code teaches structured thinking, problem decomposition, debugging, and how to give precise instructions. These are the exact skills needed to work effectively with AI. Coding education is more important now than five years ago, not as a guaranteed career path, but as a foundational literacy.¹⁶

But you do not need to aim for a career as a pure software developer to benefit from coding skills. The doctor who can write a script to analyse patient data, the marketer who can query a database, the teacher who can build a custom learning tool – these people are more valuable than their peers who cannot.

The practical advice: learn Python as a first language. Use it to build things you care about.

On university versus apprenticeships

This is a genuine fork in the road, and both paths are increasingly valid.

University offers intellectual depth, breadth of experience, a credential that opens doors (especially internationally), time to develop as a person, research skills, and a network.

Degree apprenticeships offer earning while learning (average starting salary GBP 24,000 in 2026), zero student debt, practical experience from day one, and an increasingly respected qualification. Degree apprenticeship starts in England grew 20% in 2024/25. The government has committed GBP 725 million to deliver 50,000 more apprenticeship places.¹⁷

The honest assessment: neither path is inherently better. University suits people who want intellectual exploration, are unsure of their direction, or are aiming for fields that require it. Apprenticeships suit people who learn by doing, have a clear field of interest, and value earning and practical experience. Degree apprenticeships increasingly offer the best of both.

On the creative arts

This requires honesty. Film and television jobs in California fell by nearly 30% between 2022 and 2025, driven by post-strike contraction, streaming consolidation, and AI adoption.¹⁸

But there are important nuances. Routine creative work (stock imagery, template designs, basic copywriting) is most vulnerable. Original creative vision – distinctive artistic voice, conceptual thinking, emotional resonance – is much harder to automate. Creative skills combined with other skills are increasingly valuable. And live and experiential creative work (theatre, live music, physical art installations, craft) is inherently AI-resistant because the human presence is the product.

The advice for a creative teenager: pursue your art seriously, but also develop complementary skills. Develop your distinctive voice – the thing AI cannot replicate is you. And learn to use AI tools as part of your creative process, not as a threat to it.

The positive case

AI tools are lowering barriers in ways that matter. A teenager with a laptop and an internet connection can now access capabilities that would have required a large team five years ago. They can build software, create professional-quality designs, analyse data, write business plans, and prototype ideas – all with AI as a collaborator.¹⁹

Young people are already building with these tools. The Christensen Institute (January 2026) predicts that AI may “unleash the most entrepreneurial generation we’ve ever seen” by lowering barriers to starting businesses and enabling rapid prototyping.²⁰

Historical precedent is genuinely encouraging. Every major technology shift – agricultural mechanisation, industrialisation, electrification, computing, the internet – created more jobs than it destroyed over time. The jobs that grow are often more interesting than the ones that disappear. When AI handles the routine, humans get to do the parts that require creativity, judgement, and connection.

But it is important to acknowledge what the tools do not equalise: not every teenager has the same quality of internet access, the same parental support, the same school environment, or the same safety net for taking risks. The most honest thing to say is that the tools are more democratised than ever, but the conditions for using them well are not.

What I told my daughters

The people who will thrive are not the ones who picked the “right” career at 16. They are the ones who learned how to think, stayed curious, treated people well, and kept adapting.

If I had to boil everything down to three things:

1. Learn how to think, and study what genuinely interests you. The specific subject matters much less than people think. AI has all the facts – what it does not have is the ability to evaluate them, question them, connect them to the real world, and make a decision when the answer is not clear. A history degree, an engineering degree, a science degree, an arts degree – they are all good if they teach you to think well. The worst choice is something you are not interested in, because you will not develop depth.

2. Invest in your people skills – and learn to work with AI. AI is brilliant at processing information and terrible at understanding people. The ability to listen, persuade, empathise, lead, and build trust is becoming more valuable, not less. At the same time, start using AI tools now. Use them as thinking partners. Learn what they can and cannot do.

3. Try things, and do not be afraid of changing direction. Work experience, side projects, volunteering, starting something small. The best way to figure out what you are good at and what you enjoy is to do a lot of different things. You do not need to know your career at 16 or 18 or even 25. The most successful careers today involve two or three significant pivots.

You are entering the workforce at a moment of real change. The uncertainty is genuine, and it is fine to sit with that for a bit. But you are also entering at a moment of extraordinary opportunity. The tools available to you are more powerful than anything any previous generation had. The barriers to starting things are lower. The range of possible careers is wider.

The fundamentals have not changed: think well, stay curious, treat people well, keep learning, be willing to adapt.