AI Won’t Erase Every Job, but It Will Redraw the Labour Market

AI will not determine the future of work on its own. Institutions will decide how quickly firms adopt it, how workers share its benefits and how vulnerable groups are protected. Australia’s experience indicates that the decisive contest is not between technology and employment, but between different ways of organising the transition.

Artificial intelligence (AI) is moving from experimental technology into the daily operations of banks, offices, public agencies and service businesses. Its arrival is creating new opportunities for productivity, but also raising difficult questions about who will benefit, who will be displaced and whether workers can move into new roles quickly enough.

The study "Exploring the Impact of AI on the Transformation of Labour Markets in Advanced Economies: Insights from Australia" by D. Cetindamar and Sancheeta Pugalia, published in Science, Technology & Society in 2026, examines these questions through the Australian labour market. The authors focus on the Finance and Administrative & Support Services sectors to show how the same technology can produce sharply different outcomes.

The research rejects both a mass-unemployment narrative and an uncomplicated productivity story. AI is more likely to transform tasks, alter recruitment and redistribute opportunities than eliminate employment across the economy. However, the transition will not be neutral. Workers in routine, entry-level and lower-skilled roles face greater exposure, while people with technical expertise, professional judgement and strong interpersonal capabilities may gain new opportunities.

AI's First Labour-Market Impact Will Be Uneven

Australia reflects a broader global pattern in which AI-exposed occupations continue to exist but grow more slowly than less-exposed roles. Between 2019 and 2024, employment in AI-exposed occupations increased by around 10%, compared with a 44% rise in less AI-exposed occupations, according to evidence examined in the study.

The figures point to a gradual restructuring rather than an immediate collapse. AI can automate a portion of an occupation without replacing the entire worker. A financial analyst may use AI to process data and prepare an initial assessment, while retaining responsibility for interpretation and decision-making. An administrative employee may use generative tools to draft correspondence, but lose much of the routine work that once provided an entry point into the profession.

Such changes can affect wages, promotion pathways and bargaining power even when headline employment remains stable. Workers may be expected to manage more tasks, supervise automated systems or acquire new skills without receiving equivalent training or compensation.

The pressure is particularly significant for younger workers. Entry-level positions frequently serve as training grounds where employees acquire institutional knowledge and practical judgement. If companies automate these tasks or reduce junior recruitment, future workers may struggle to gain the experience required for more advanced roles.

Finance Shows How AI Can Augment Professional Work

Finance is among the Australian industries most exposed to AI. Around 11.8% of job postings in financial and insurance activities required AI capabilities in 2024, making the sector a leading site of technological adoption.

The study describes a transition in which AI supports compliance reporting, fraud detection, risk modelling and customer service. Automated systems can identify unusual transactions, analyse large datasets and prepare preliminary reports. Human employees remain essential for regulatory interpretation, complex judgement, client trust and accountability.

The result is a shift in the composition of work. Demand may grow for professionals who understand algorithm auditing, cyber-risk, financial technology and AI governance. Traditional financial expertise will remain valuable, but it will increasingly need to be combined with the ability to evaluate and manage automated systems.

Routine roles face a less secure future. Bookkeeping, data entry, call-centre operations and some back-office insurance functions involve tasks that can be codified and performed by AI. The study cites an estimate of 7,800 job losses across Australia's four major banks in 2025. Those reductions cannot be attributed entirely to artificial intelligence, since restructuring, cost pressures and wider business decisions may also have played a role. The pattern nevertheless illustrates how automation risk can become concentrated in particular functions.

Finance illustrates both the potential and the limits of AI-led growth. The sector may generate new economic value while reducing demand for some forms of clerical labour. Whether workers benefit will depend on the availability of retraining, internal mobility and credible safeguards against exclusion.

Administrative Work Faces a Sharper Social Risk

Administrative and support services cover a wide range of activities, from office administration and reception to cleaning, gardening, travel services and employment placement. AI will not affect these occupations uniformly, but the sector contains a high concentration of work vulnerable to generative systems.

Office clerks, receptionists, bookkeepers and some sales, marketing and public relations roles are identified in the study as occupations likely to face employment losses by 2050. Generative AI can draft documents, transcribe meetings, answer routine queries and process standardised information. These capabilities directly affect the tasks that have historically supported large administrative workforces.

The sector's workforce composition adds an important equity dimension. Administrative and clerical occupations are heavily female-dominated, while many positions do not require tertiary education. A reduction in these jobs could therefore affect women and workers without university qualifications more severely than aggregate employment figures suggest.

Casual employment is also prevalent in the sector. Workers in insecure arrangements may have fewer opportunities to receive employer-funded training or negotiate the pace of technological change. They may also face greater difficulty absorbing periods of unemployment or moving into more secure occupations.

Other roles could experience sustained demand. Cleaning, nursing, hospitality and some management positions rely on physical presence, social interaction or situational judgement that current AI systems cannot easily reproduce. The sector may therefore experience internal divergence: office-based roles contract while interpersonal and physical service occupations expand.

The risk extends beyond job loss. Removing routine administrative work can weaken career ladders and make it harder for workers to build experience. The transition could produce a labour market in which employers expect advanced capabilities from new recruits while offering fewer opportunities to develop those capabilities on the job.

Policy Choices Will Determine Who Gains

Australia's AI governance has moved from voluntary ethical principles towards safety standards, public-sector risk assessments, technical monitoring and a national AI plan. The government established the Australian AI Safety Institute with AUD 29.9 million in funding and placed emphasis on skills, infrastructure and domestic capability.

The national approach stops short of mandatory guardrails for high-risk AI systems. It relies on existing legal frameworks, voluntary standards and technical oversight. The authors position this model closer to the market-oriented approaches of the United States and United Kingdom than to the European Union's more comprehensive regulatory framework.

The approach may help accelerate adoption and investment, but it also places considerable responsibility on employers and institutions. Voluntary safeguards may produce uneven results if businesses lack the capacity or incentive to identify employment risks. Technical monitoring can assess system performance, but it cannot by itself resolve questions about redundancy, wages, discrimination or worker voice.

Governments should therefore pair AI investment with sector-specific labour policies. Finance workers may need modular training in AI-assisted compliance, risk management and governance. Administrative workers may require digital literacy, career-transition support and pathways into growing occupations. Generic promises of reskilling will have limited value unless programmes reach the groups most exposed.

Businesses should involve workers in implementation decisions and disclose how AI is changing tasks, recruitment and performance assessment. Internal redeployment and paid training could preserve institutional knowledge while reducing the social costs of automation.

International organisations and development agencies can draw a broader lesson from the Australian case. Digital infrastructure is only one part of AI readiness. Countries also need labour-market data, education systems, social protection and regulatory institutions capable of managing unequal transitions. For developing economies, the danger lies in importing AI systems without building the capacity to assess their distributional consequences.

It should be noted that the study's single-country design restricts generalisation, and its reliance on existing reports and secondary data offers limited insight into workers' daily experiences. The evidence also captures an early stage of AI adoption, leaving uncertainty about long-term employment, wage and productivity effects.

Further research should compare countries with different governance models and follow workers over time. Surveys, workplace studies and longitudinal data could reveal whether AI improves job quality, erodes autonomy or widens wage gaps. Evidence is also needed on whether productivity gains translate into better pay or remain concentrated among firms and investors.