Future of work • Artificial intelligence • Employment
Artificial intelligence is no longer simply a tool for answering questions, generating pictures or writing emails. It is increasingly capable of performing tasks that once required trained professionals, from processing financial documents and producing marketing content to handling customer enquiries and writing computer code.
The question facing workers is no longer whether AI will change employment. It is how profoundly it will change particular occupations — and whether some jobs will largely disappear within the next decade.
The evidence suggests that the greatest risks are concentrated in work involving predictable processes, repetitive digital tasks, standardised information and decisions that can be made using established rules. However, a job being exposed to AI does not automatically mean that the job will disappear. In many cases, fewer employees may be needed, responsibilities may change, or workers may be expected to use AI to produce considerably more output.
The World Economic Forum’s Future of Jobs Report 2025, based on a survey of more than 1,000 employers representing over 14 million workers across 55 economies, identified clerical and administrative roles among those expected to experience some of the largest declines by 2030. Meanwhile, the International Labour Organization’s 2025 research estimated that approximately one in four workers worldwide is in an occupation with some degree of exposure to generative AI — while stressing that transformation is generally more likely than complete replacement.
This countdown examines ten occupations at particular risk by 2036, ranked from number 10 to number 1, with number 1 representing the greatest potential impact. The ranking is an editorial assessment of automation potential, not a claim that these jobs will all vanish.
10. Graphic designers
Risk profile: High for routine production; lower for creative direction and complex branding.
Graphic design has traditionally required a combination of technical skill, visual imagination and an understanding of what makes an audience respond. AI is rapidly changing the balance between those skills.
Today’s AI-powered design tools can generate advertising concepts, social media graphics, illustrations, product mock-ups and multiple variations of a logo in minutes. Tasks that once required hours of specialist work can now be completed in a fraction of the time.
The commercial implications are significant. A small business that previously paid a designer to produce dozens of routine promotional images might increasingly generate them internally. Marketing departments could also need fewer junior designers to resize artwork, produce alternative layouts or create first drafts.
The World Economic Forum’s 2025 report specifically identifies graphic designers among the roles facing decline, reflecting the growing influence of generative AI.
What could disappear? Not necessarily graphic design itself, but a significant amount of routine production work. Designers who offer distinctive creative direction, original concepts, cultural understanding and sophisticated brand strategy are better positioned to retain value.
The uncomfortable reality is that clients may increasingly pay for the idea and the judgement behind a design, rather than the hours spent producing it.
9. Translators and routine interpreters
Risk profile: High for routine written translation; more variable for live and specialist interpreting.
Translation is one of the clearest examples of AI entering a profession previously dependent on specialised human knowledge.
Modern systems can translate emails, websites, product descriptions, business reports and general documents into numerous languages almost instantly. They can also adjust tone, summarise foreign-language material and produce alternative translations for different audiences.
For businesses that need to translate thousands of product listings or routine customer enquiries, the financial attraction is obvious: AI can process enormous volumes of text at a relatively low marginal cost.
The OECD has identified translators and interpreters among occupations with high exposure to generative AI, including tasks that AI can perform considerably faster than conventional methods.
But translation is not simply the replacement of words. Legal documents, medical instructions, diplomatic discussions and sensitive negotiations can require cultural understanding, confidentiality, professional accountability and an appreciation of subtle meaning.
What could disappear? Much of the market for straightforward, general-purpose translation could shrink or become cheaper. Human specialists may increasingly focus on reviewing AI output, handling high-stakes material and resolving ambiguity that automated systems miss.
The likely outcome is a smaller or differently structured profession, rather than the end of human translation.
8. Customer service representatives
Risk profile: High for standard enquiries and scripted support.
Customer service centres are expensive to operate. Businesses must recruit, train and supervise staff to answer questions about orders, refunds, deliveries, account access, billing and product features.
AI agents can already handle many of these interactions through text and voice. They can retrieve information from company systems, follow prescribed procedures, summarise previous conversations and transfer more complicated cases to human employees.
Unlike a traditional chatbot that simply offers a menu of options, newer systems can conduct more natural conversations and complete multiple steps in a customer-service process.
The World Economic Forum identifies several clerical and customer-facing administrative occupations among those expected to decline as digital technologies become more widespread.
Imagine a retailer receiving 10,000 routine customer enquiries a day. If an AI system can reliably resolve a substantial proportion without human intervention, the company may not need to recruit additional agents as its business grows. It might eventually reduce staffing while maintaining the same service volume.
What could disappear? Large numbers of routine, scripted interactions and some entry-level contact-centre positions. Human employees will still be needed for complaints, vulnerable customers, unusual circumstances and situations where empathy or discretion matters.
The bigger risk may be that fewer people are hired in the first place, rather than that every existing agent is immediately dismissed.
7. Bookkeepers and payroll clerks
Risk profile: High for routine transaction processing and reconciliations.
Bookkeeping involves recording transactions, matching invoices to payments, reconciling accounts and preparing routine financial reports. Payroll clerks calculate wages, process deductions and maintain employee payment records.
Much of this work follows repeatable rules and relies on structured digital information — precisely the sort of environment in which AI and conventional automation can make substantial gains.
Modern accounting systems can extract data from invoices, classify expenses, flag discrepancies and prepare draft reports. AI can also help explain unusual transactions, identify missing documentation and answer routine questions about financial records.
The International Labour Organization’s 2025 research specifically lists accounting and bookkeeping clerks among occupations with particularly high exposure to generative AI.
Consider a company that employs several people to process invoices and reconcile payments. If software can automate most of these activities, the same workload may eventually be managed by a smaller team, with staff concentrating on exceptions and oversight.
However, bookkeeping is not simply data entry. Businesses still need reliable records, compliance with tax rules, investigation of discrepancies and someone accountable for financial decisions.
What could disappear? Routine transaction processing and some junior bookkeeping positions could contract substantially. Experienced bookkeepers who understand tax requirements, business operations, financial controls and client relationships may instead evolve into advisers and controllers of automated systems.
6. Junior journalists, copywriters and content producers
Risk profile: High for generic content; lower for original reporting and trusted specialist expertise.
AI can already produce news summaries, product descriptions, search-engine-optimised articles, social media posts, email campaigns and basic promotional scripts.
For publishers and marketing departments that produce large volumes of similar material, this creates a powerful incentive to automate first drafts and routine content production.
A company that previously commissioned 100 product descriptions from freelance writers may now generate those descriptions in-house, using a human editor to check accuracy and consistency. A publisher might use AI to summarise routine announcements, freeing its remaining journalists to concentrate on reporting.
The ILO’s 2025 research found that advances in text, image and video generation have increased the exposure of media- and web-related occupations to generative AI.
There is a serious commercial complication, however. If every publisher uses AI to generate similar material, the internet could become flooded with interchangeable content. The scarcity may shift from producing words to finding something genuinely worth saying.
Original investigations, interviews, exclusive information, editorial judgement and a recognisable human voice remain difficult to reproduce reliably through automated content generation alone.
What could disappear? Some entry-level writing assignments, routine content contracts and junior production roles. Journalism and publishing may survive in a very different form, with fewer people producing more material and a premium on original reporting, credibility and distinctive perspectives.
The profession’s greatest danger could be the collapse of the traditional entry-level career ladder: the routine assignments through which inexperienced writers once learned their craft may become increasingly scarce.
5. Data entry clerks and document-processing staff
Risk profile: Very high for repetitive, structured digital tasks.
Few occupations illustrate the vulnerability of routine office work more clearly than data entry.
Historically, organisations have employed people to transfer information between forms, spreadsheets, databases, invoices, customer records and internal systems. The work can be tedious, but it has been essential to the functioning of businesses, banks, insurers and public institutions.
AI-powered document processing can now extract information from scanned documents, recognise fields, classify records and transfer information into the appropriate systems. When connected to business software, automated workflows can process an entire document from receipt to filing.
The World Economic Forum identifies data entry clerks among the fastest-declining job categories expected by employers, alongside bank tellers and postal service clerks.
The technology is not infallible. Poor-quality scans, inconsistent formats, missing information and unusual cases can still require human intervention. But automation does not need to be perfect to reduce staffing requirements: it only needs to handle enough routine work to make the economics attractive.
What could disappear? Many jobs devoted primarily to copying, classifying and transferring information. Remaining workers may oversee automated processes, investigate errors and manage exceptions.
For people considering a career in administration, this is an important warning: learning to enter information faster may be less valuable than learning to interpret it, verify it and use it to make decisions.
4. Basic software developers and junior programmers
Risk profile: High for routine coding; substantial transformation across the profession.
Software development might seem like an unlikely candidate for disruption because it is one of the industries responsible for building AI. Yet the ability to generate, explain, test and modify code is one of the most consequential developments in generative AI.
Developers can use AI assistants to create standard functions, produce test cases, identify bugs, explain unfamiliar code and draft documentation. More advanced tools can undertake sequences of coding tasks with less continuous human supervision.
This does not mean that AI can reliably build every complex application from a short instruction. Understanding what a customer actually needs, designing secure systems, integrating legacy software and maintaining complicated products remain demanding activities.
Nevertheless, the economics of software production are changing. A small team may be able to build features that once required a much larger development effort. Employers may consequently recruit fewer junior programmers for routine implementation work, even if demand for software remains strong.
The ILO’s 2025 analysis identifies application programmers and web and multimedia developers among occupations whose exposure to generative AI has grown.
What could disappear? Some entry-level coding tasks and narrowly defined programming roles could shrink. Software engineers who understand architecture, security, business requirements, testing and system reliability are better placed to benefit from AI-assisted development.
The irony is that the industry may continue growing while requiring fewer people to perform certain kinds of coding. More software does not automatically mean proportionately more programmers.
3. Bank tellers and routine banking clerks
Risk profile: High for routine transactions and administrative processing.
Banking has been moving towards digital self-service for years. AI adds another layer of automation to a sector that already relies heavily on software, standardised procedures and electronic records.
Routine activities such as checking balances, processing payments, updating customer details and answering standard account questions can increasingly be handled through banking apps, automated systems and conversational AI.
AI can also assist with document verification, fraud alerts and customer onboarding. This could reduce the need for employees to handle routine transactions manually, while allowing banks to serve more customers digitally.
The World Economic Forum’s Future of Jobs Report 2025 specifically lists bank tellers among the roles expected to decline rapidly by 2030.
There are important limits. Customers may still want human assistance with complex financial problems, bereavement, disputes, lending decisions and sensitive personal circumstances. Financial institutions must also maintain security, regulatory compliance and effective fraud controls.
What could disappear? Traditional counter-based transaction processing and some routine back-office banking roles. Branches may increasingly focus on advice, complex services and relationship management rather than everyday transactions.
For prospective banking employees, the lesson is clear: expertise in financial products, customer needs, risk and compliance is likely to be more durable than a role centred on processing standard transactions.
2. Administrative assistants and routine office administrators
Risk profile: Very high for routine coordination and document management.
Administrative work is woven into almost every organisation. Assistants schedule meetings, prepare documents, organise records, respond to correspondence, manage calendars and coordinate information between departments.
Many of these tasks are now within the reach of AI systems connected to email, calendars, document software and company databases. An AI assistant can draft a response, summarise a long email chain, prepare a meeting agenda, extract actions from minutes and help arrange appointments.
The potential impact extends beyond individual tasks. If an AI system can complete several connected administrative activities, employers may rethink the need for entire layers of routine office support.
The evidence is particularly strong here. The World Economic Forum identifies administrative assistants and executive secretaries among the job categories expected to experience some of the largest declines in absolute numbers by 2030. The ILO also identifies administrative secretaries among the occupations most exposed to generative AI.
But a good administrator does much more than manage a calendar. Experienced assistants anticipate problems, understand office politics, protect confidential information and coordinate people whose priorities conflict. These responsibilities require context, judgement and trust.
What could disappear? Roles focused primarily on routine scheduling, document production, information handling and standard correspondence. More senior executive assistants and office managers may remain valuable if they take responsibility for complex coordination, stakeholder relationships and operational decision-making.
The likely transformation is from administrative work as a collection of manual tasks to administrative work as the supervision of automated systems.
1. Routine customer-facing and back-office processing roles
Risk profile: Extremely high where work is digital, repetitive and governed by predictable rules.
Why put a broad category of processing roles at number one rather than a single profession? Because the most significant threat may not be the disappearance of one job title. It may be the automation of a whole family of tasks across multiple industries.
Think of employees who review standard insurance claims, process routine applications, classify documents, update customer records, check forms for missing information or route cases to the correct department.
These roles exist in insurance companies, utilities, banks, telecommunications businesses, hospitals, government departments and countless other organisations. They may have different job titles, but their underlying work often follows similar patterns: receive information, check it against rules, update a system and trigger the next step.
AI can increasingly combine document recognition, language processing, data retrieval and workflow automation to complete several of these steps in sequence. That creates a bigger opportunity than automating one isolated task.
The OECD’s research on AI and regional labour markets identifies insurance underwriters among the occupations where a large share of tasks may be exposed to generative AI, while the ILO identifies clerical occupations as the most exposed overall.
Imagine an insurer receiving thousands of straightforward claims. An AI system could extract the details, compare them with policy conditions, flag missing evidence and prepare a recommendation. Human specialists would still be required for complex, disputed or high-value claims, but the company might need far fewer people to handle routine cases.
The same pattern could emerge across many sectors. Instead of eliminating every employee in a particular department, an organisation may simply stop replacing departing workers as automated systems take over more of the workload.
What could disappear? Large volumes of routine processing work, some entire job categories and potentially many entry-level positions across different industries.
This is arguably the most consequential form of disruption because it is not confined to a single profession. It affects the common administrative infrastructure of modern business.
The crucial distinction is between a task and a job. AI may automate 70% of a role’s routine workload without being able to perform the remaining 30%. But if the remaining work can be redistributed among other employees, the organisation may still need fewer people overall.
That is why the transformation of ordinary office work deserves at least as much attention as the prospect of AI replacing highly visible professional occupations.
The bigger picture: Will these jobs actually disappear by 2036?
The evidence supports a major transformation of work, but it does not support confidently declaring that all ten occupations will disappear within ten years.
Three findings help put the predictions into perspective.
1 in 4 workers
The ILO’s 2025 research estimates that 24% of jobs worldwide have some degree of exposure to generative AI. Exposure means tasks could be affected, not that workers will necessarily lose their jobs.
39% of skills
Employers surveyed for the World Economic Forum’s 2025 report expect 39% of workers’ existing skill sets to be transformed or become outdated between 2025 and 2030.
Jobs will also be created
The same World Economic Forum report projects 170 million new jobs and 92 million displaced jobs globally by 2030 across the trends it examines, a projected net increase of 78 million. These are employer-based forecasts, not guaranteed outcomes or estimates attributable to AI alone.
What should workers do now?
The most sensible response is not to assume that a career is doomed, but to identify which parts of it are easiest to automate.
- Learn to use AI effectively. Workers who can combine domain expertise with AI tools may become more productive and valuable.
- Develop judgement and accountability. Verification, risk assessment, problem-solving and responsibility matter when automated systems make mistakes.
- Build human relationships. Negotiation, trust, leadership, empathy and handling complex situations remain important advantages.
- Move beyond routine execution. Instead of simply producing a report, learn to interpret it and recommend what a business should do next.
- Watch entry-level opportunities. Even when an occupation survives, AI may reduce the number of junior positions traditionally used to train new professionals.
The final verdict
The jobs most vulnerable to AI are not necessarily the least skilled or lowest paid. They are the ones whose tasks can be described, digitised, repeated and evaluated with relatively predictable results.
That is why administrative professionals, translators, journalists, designers and even software developers face substantial change. AI is increasingly capable of performing work that once required education, experience and specialist knowledge.
But technology is only part of the equation. Adoption costs, regulation, customer preferences, reliability and business decisions will determine how quickly employers replace people or redesign their jobs.
The most important question for the next decade may not be which jobs AI can do. It may be how many people businesses will choose to employ once AI can do so much of the work.
Sources and further reading
- International Labour Organization — Generative AI and Jobs: A 2025 Update.
- World Economic Forum — The Future of Jobs Report 2025.
- OECD — Skills in the AI Age.
Editorial note: The countdown is a reasoned ranking of occupations and task groups by potential disruption through 2036. It is not an official ranking published by these organisations, and the evidence does not establish that any of the ten occupations will disappear completely by that date.
