Which Jobs Will Survive AI?
Companies still advertise for marketing managers, analysts, software developers and HR specialists using job descriptions that would have looked familiar several years ago. The degree requirements remain, the experience thresholds stay in place and AI often appears near the bottom as one more desirable skill.
The work has moved further than the recruitment language suggests. Research, drafting, coding, reporting and administration can already be completed with fewer hours and, in some cases, fewer people. Occupations rarely disappear as complete units, however. Their more predictable tasks are absorbed first, while the remaining position takes on a greater share of judgement, oversight and responsibility.
Lists of supposedly “AI-proof” careers offer limited guidance. A more useful assessment starts with the work attached to each position: which tasks can be automated, where a qualified professional must still make the decision and whether companies will continue paying for the combination of skills the role requires.
The labour-market figures show why the question has become urgent. Job advertisements on the German edition of Indeed have fallen by 37% since November 2022, while vacancies in conventional office-based fields such as software development, human resources and marketing have roughly halved. The figures come from Germany, but the underlying pattern is relevant across continental Europe, where employers face similar economic pressures, demographic constraints and expectations that AI will allow office-based teams to produce more with fewer people. National labour markets will not move at the same speed, although the direction of travel is unlikely to remain confined to Germany.
Some positions will survive because they depend on physical presence. Others will remain because a licensed professional must accept responsibility for the outcome. Many office roles will continue under familiar titles, although the people occupying them will be expected to contribute at a higher level than the job description currently suggests.
Doctors and healthcare professionals
Why the position will survive
Medical work combines specialist knowledge, physical examination, interpersonal trust and responsibility for decisions that can seriously affect another person. AI can support diagnosis, analyse medical images, prepare documentation and help clinicians compare treatment options. It cannot assume the professional, ethical and legal responsibility attached to patient care.
Demand is reinforced by ageing populations and existing shortages across European healthcare systems. Physiotherapists, occupational therapists, psychotherapists and other medical and social professionals have also experienced more resilient recruitment than many office-based occupations.
How the work will change
Doctors may spend less time compiling standard information or preparing routine records. More of their working day could move towards interpreting AI-supported findings, explaining uncertainty and deciding whether a recommendation is appropriate for the individual patient.
As automated systems produce increasingly persuasive answers, the ability to recognise an incomplete dataset, an inappropriate comparison or a clinically implausible conclusion becomes more valuable. AI may improve the speed of diagnosis without removing the need for medical judgement.
What employers will recruit for
Professional qualifications and accreditation will remain non-negotiable. Employers will also look for diagnostic reasoning, communication under pressure, ethical judgement and the ability to evaluate digital tools without becoming dependent on them.
Nurses and care professionals
Why the position will survive
Nursing and care take place in direct contact with patients. The work includes observation, physical assistance, reassurance and the recognition of subtle changes that may never appear in a formal dataset.
A frightened, confused or unwell patient does not simply need information. They need someone who can interpret behaviour, adapt their response and decide what should happen next. Those demands make care work difficult to separate from the person delivering it.
How the work will change
Technology can reduce documentation, improve scheduling and alert staff to changes in vital signs. Used well, it could remove part of the administrative burden currently placed on nurses and carers.
It may also make the role more demanding. When software generates more alerts and recommendations, professionals must distinguish between useful warnings and irrelevant noise. Less time may be spent recording information, while more attention goes to interpreting it.
What employers will recruit for
Clinical competence will remain central, alongside emotional steadiness, observation, prioritisation and decision-making under pressure. Hospitals and care providers will increasingly need professionals who can work confidently with digital monitoring systems while retaining close attention to the patient in front of them.
Engineers
Why the position will survive
Engineering connects abstract knowledge with physical systems. Buildings, factories, energy networks, vehicles and infrastructure must function safely under real-world conditions, where errors carry financial, regulatory and sometimes life-threatening consequences.
AI can assist with calculations, simulations, design alternatives and technical documentation. Companies still need qualified engineers who understand materials, standards, constraints and the interaction between individual components.
How the work will change
Engineers will be able to produce and test more alternatives in less time. Routine calculations and initial designs may become faster, leaving a greater share of the work focused on system integration, validation and risk assessment.
The shift creates a training challenge. Junior engineers have traditionally developed judgement through repetitive calculations and closely supervised design work. Employers will need more deliberate training methods when software completes much of that foundational work automatically.
What employers will recruit for
Engineering degrees will remain important because the profession depends on substantial technical knowledge. Employers will also seek systems thinking, the ability to challenge model outputs and an understanding of how technical decisions affect costs, regulation, safety and operational reliability.
Skilled tradespeople
Why the position will survive
Electricians, plumbers, heating technicians and other skilled trades work in physical environments that are inconsistent and difficult to predict. A repair in an older building can reveal conditions that were invisible before work began. The tradesperson must diagnose the situation, adjust the plan and complete the job safely around existing structures and equipment.
Robotics will automate parts of construction and maintenance, although deployment is likely to be slower than in digital office work. Every site introduces different physical constraints, making standardisation difficult.
How the work will change
Tradespeople will increasingly use diagnostic software, connected equipment and AI-supported planning. Faults may be identified remotely before a technician arrives, while digital guidance could support complex repairs on site.
Technology will shorten the time required to find the problem and prepare the intervention. The physical work, practical judgement and responsibility for a safe installation will remain.
What employers will recruit for
Vocational qualifications will continue to matter. Employers will place more weight on diagnostic ability, practical problem-solving and confidence with connected systems. The strongest candidates will understand both the physical installation and the data it produces.
Teachers
Why the position will survive
AI can explain a concept, generate exercises and adapt material to a pupil’s level. Teaching includes far more than the transfer of information. Teachers observe whether children are engaged, recognise when difficulties are academic or personal and create the conditions in which a group can learn.
Schools also fulfil a social and civic role. Children learn how to cooperate, argue, interpret authority and participate in a community. Software can support that process without replacing the adult responsible for it.
How the work will change
Lesson preparation, basic assessment and differentiated exercises can be accelerated. Teachers may also gain more detailed information about where individual pupils are struggling.
At the same time, assessment will become harder. When pupils can generate fluent answers instantly, teachers must determine whether they understand the material, can explain their reasoning and recognise unreliable information. Source evaluation and intellectual independence will occupy a larger part of education.
What employers will recruit for
Subject knowledge will remain essential, alongside classroom judgement, communication and the ability to diagnose why a learner is not progressing. Digital competence will be judged through teaching practice rather than a generic claim that the candidate is comfortable with AI.
Lawyers
Why the position will survive
Legal work contains large volumes of research, document review and standard drafting, all of which can be accelerated by AI. The profession also involves interpretation, negotiation, strategic judgement and responsibility for advice with serious consequences.
A system can identify clauses or produce a first draft. It cannot decide which risk a client should accept, how a regulator is likely to respond or which concession may resolve a dispute without weakening the client’s position elsewhere.
How the work will change
Junior lawyers may spend fewer hours on basic document review and standard research. They will be expected to move more quickly towards checking outputs, identifying unusual facts and understanding the client’s wider commercial position.
Law firms will also need to reconsider how associates learn. Much of their professional judgement has traditionally developed through routine work performed repeatedly under supervision. When AI removes part of that work, legal reasoning will have to be taught more deliberately.
What employers will recruit for
Legal qualifications will remain necessary. Firms will place greater emphasis on analytical discipline, commercial understanding, writing quality and the willingness to challenge an answer that appears authoritative. Familiarity with an AI tool will matter less than knowing when its conclusion cannot be trusted.
Accountants, auditors and tax advisers
Why the position will survive
Routine bookkeeping, reconciliation and standard reporting are highly exposed to automation. Professional accounting work also involves interpretation, control and responsibility for the information used by management, investors and tax authorities.
Complex corporate structures, cross-border transactions and changing regulation continue to require qualified professionals who can identify inconsistencies and explain their consequences.
How the work will change
Less time will be spent compiling information manually. More attention will go to investigating exceptions, reviewing assumptions and advising management on the implications of different accounting or tax treatments.
Audit teams will be able to examine larger quantities of data rather than relying as heavily on samples. That increases the opportunity to identify anomalies, but someone must still determine whether an anomaly represents fraud, error or a legitimate business event.
What employers will recruit for
Professional qualifications and technical knowledge will remain central. Employers will also seek scepticism, business understanding and the ability to explain complex financial issues clearly. Accuracy alone offers limited protection when systems can produce neat but misleading outputs at scale.
Senior software developers and technology architects
Why the position will survive
AI can already produce code, correct common errors and generate documentation. That places pressure on roles dominated by routine programming, but companies still need people who understand how systems fit together, how data moves through an organisation and where security or reliability can fail.
Someone must define the problem, select the architecture and accept responsibility when the software does not behave as intended.
How the work will change
Developers will spend less time writing every line manually. More of their work will involve reviewing generated code, designing systems and solving difficult integration problems.
The greatest pressure may fall on entry-level positions. Basic coding tasks that once allowed graduates to build experience can now be completed quickly by AI. Companies that reduce junior recruitment without creating another development route risk producing a shortage of senior engineers later.
What employers will recruit for
Computer-science degrees will continue to provide valuable foundations, although recruiters will rely less on qualifications alone. System design, security awareness, product understanding and evidence of continued learning may matter more than familiarity with a particular programming language.
Marketing strategists and senior communications professionals
Why the position will survive
Basic content production is among the activities most accessible to generative AI. Product descriptions, social posts, first campaign concepts and simple research summaries can already be produced quickly.
Companies still need people who understand the audience, define the proposition and judge whether communication supports the business. Regulated organisations also require professionals who can recognise legal, reputational and political risk before material is published.
A role built mainly around producing high volumes of conventional content faces considerable pressure. A position combining editorial judgement, business knowledge, data interpretation and stakeholder awareness is more likely to endure.
How the work will change
One professional may be able to produce considerably more material with a smaller internal or external team. Time previously spent preparing first drafts can move towards research, campaign direction, distribution and performance analysis.
Judgement will become easier to distinguish from production skill. When every competitor can generate competent-looking copy, advantage comes from knowing what is worth saying, what should be omitted and how communication connects to a credible commercial position.
What employers will recruit for
A marketing or communications degree may be useful, but portfolio quality, sector knowledge and evidence of sound decision-making will carry more weight. Candidates may be asked to assess an AI-generated campaign, identify its weaknesses and improve the underlying strategic argument rather than simply rewrite the copy.
Human-resources professionals
Why the position will survive
AI can draft vacancy notices, screen applications, answer routine employee questions and support workforce analysis. HR work also deals with conflict, organisational politics, employment law and decisions that affect individual careers.
Automated recommendations may appear objective while reproducing patterns contained in historical data. Companies need professionals who understand where efficiency ends and legal, ethical or cultural exposure begins.
How the work will change
Administrative HR positions are likely to face greater pressure than roles involving organisational design, employee relations and specialist advice. Recruitment teams may spend less time reviewing applications manually and more time auditing selection systems, advising managers and assessing candidates whose experience does not fit standard categories.
Workforce planning will also move away from counting employees under fixed job titles. HR teams will need to understand which tasks are changing, which capabilities the organisation lacks and whether existing employees can be retrained before new positions are created.
What employers will recruit for
Employment-law knowledge, analytical ability and interpersonal judgement will matter more than experience with a particular HR platform. Employers will look for people who can challenge managers, interpret workforce data and recognise when an automated process is producing an unreliable or discriminatory result.
Sales professionals and relationship managers
Why the position will survive
AI can identify leads, prepare account summaries and draft follow-up messages. Complex sales still depend on trust, negotiation and an understanding of how decisions are made inside the client organisation.
In financial services, professional services and high-value business markets, the buyer is rarely selecting a provider on specifications alone. They are assessing credibility, implementation risk and whether the seller understands the pressures surrounding the decision.
How the work will change
Sales professionals will have better information before meetings and may spend less time on routine administration. Generic outreach will become easier to produce and therefore easier to ignore.
The value of a strong relationship manager will lie in asking better questions, understanding informal decision structures and recognising when the client’s stated concern is not the real reason a transaction has stalled.
What employers will recruit for
Years in sales provide limited evidence unless the employer understands the complexity of the deals involved. Recruitment will concentrate more closely on listening, commercial judgement, negotiation and the ability to work across legal, technical and financial questions.
Managers
Why the position will survive
AI can summarise performance, prepare options and support resource allocation. Management still requires decisions about priorities, trade-offs and people. Someone must decide which objective matters most, who should be trusted with a difficult assignment and when an apparently efficient plan would damage the organisation.
Technology may expose managers whose main contribution consisted of passing information between different levels of the company. It will not remove the need for those capable of directing work and accepting responsibility for difficult decisions.
How the work will change
Reporting and coordination may consume less time. Managers will be expected to interpret a larger volume of information and make decisions more quickly.
Some organisations may remove layers of approval that no longer contribute enough value. At the same time, managers who can redesign work, develop people and resolve conflict will remain difficult to replace.
What employers will recruit for
Companies will look beyond years of supervisory experience and the number of direct reports listed on a résumé. Evidence of sound judgement, team development and responsibility for difficult decisions will become more persuasive.
Which jobs face the greatest risk?
The most exposed positions are not always entire professions. They are roles dominated by predictable digital tasks with limited responsibility for the final consequence.
Junior content producers face pressure because first drafts, variations and basic summaries can be generated quickly.
Administrative assistants are vulnerable where the work mainly consists of scheduling, transcription, document preparation and standard correspondence.
Basic data-entry and reporting positions are exposed because structured information can be processed automatically.
Entry-level programmers may encounter fewer vacancies when companies use AI for routine coding and retain a smaller number of experienced developers for review, security and architecture.
Junior research and analytical roles face pressure when employers treat the work as information collection rather than as the development of professional judgement.
These positions may not disappear completely. Their numbers may decline, several jobs may be combined into one broader role or companies may preserve them as structured training positions because the organisation still needs a route through which inexperienced employees become qualified professionals.
A degree, experience or adaptable skills?
Employers will not choose only one.
A degree remains valuable when the work requires deep knowledge, formal accreditation or the ability to reason within a complex discipline. Occupational experience provides context, judgement and an understanding of how decisions are made. Adaptable skill combinations indicate whether someone can apply that knowledge when the tools and boundaries of the role change.
What employers will recruit for depends on the position, the consequences of error and how much of the work can be delegated without weakening professional accountability.
For doctors, engineers, lawyers and accountants, formal qualifications will remain central because responsibility cannot be separated from professional knowledge.
For marketing, HR, sales and management positions, employers will place more weight on evidence of judgement, cross-functional understanding and the ability to learn.
For technology roles, academic foundations may matter alongside a portfolio showing how the candidate approaches systems, security and unfamiliar problems.
Adaptability will need to mean more than an undefined personality trait. Employers will look for observable evidence: learning a new tool, transferring knowledge between industries, working across functions or recognising when an established method no longer fits the problem.
Recruitment will begin with tasks, not titles
Before reposting an existing vacancy, employers will increasingly examine the work contained within it. They will ask which tasks remain necessary, which can be accelerated, where expert review is required and which errors would create financial, legal or reputational damage.
A company may discover that it still needs a marketing manager, but no longer one whose main value lies in producing copy. It may need a software developer with less emphasis on routine coding and more expertise in architecture. It may require an HR professional who can audit automated recruitment rather than process applications manually.
The title may remain, the selection criteria will not.
The positions most likely to survive AI are connected to responsibility, judgement, physical reality and human trust. They will change, and some will employ fewer people. Their value will move away from completing predictable tasks and towards deciding what should be done, evaluating whether it has been done properly and accepting responsibility for the outcome.