Will AI Create Jobs or Replace Them? India’s Banking, IT and Media Careers Face a Turning Point

Artificial intelligence can raise productivity and open new careers. It can also shrink routine work and make the first job harder to secure. Who benefits will depend on decisions made now.

Navyug News | Employment and technology analysis | Updated 3 October 2026

Four workers, one question

A banker reviews a customer’s documents. A programmer fixes an error. A journalist turns an interview into a story. A graduate prepares for a first job.

Increasingly, software can assist with each activity. It can extract information, suggest code, transcribe recordings and draft responses. The excitement is understandable. So is the anxiety.

“Will AI create jobs or replace them?” has no single answer. Both outcomes are possible, sometimes within the same organisation. A company may hire specialists to build AI systems while recruiting fewer people for routine processing.

For India, the challenge is therefore bigger than learning a new tool. It is ensuring that productivity gains translate into opportunity rather than a narrower doorway into professional employment.

What the research actually says

The International Labour Organization’s 2025 assessment finds that one in four workers globally is in an occupation with some exposure to generative AI. Its central conclusion is that transformation is more likely than complete replacement because most occupations still involve tasks requiring people. Exposure measures technological potential; it is not a count of jobs already lost.

The World Economic Forum’s Future of Jobs Report 2025 projects 170 million jobs created and 92 million displaced by 2030, producing a net increase of 78 million.

Those figures cover multiple forces—including technology, demographic change and the green transition. They must not be advertised as jobs created or destroyed by AI alone. They are employer-informed forecasts, not guaranteed outcomes.

Even a positive global balance can conceal painful transitions. A displaced employee cannot automatically move into a new occupation requiring different qualifications, location or experience.

Tasks disappear before job titles do

A job is a bundle of activities. Automating one component can free time for others; automating several can reduce the number of employees needed.

Consider a hypothetical customer-service team. AI handles straightforward queries, while employees investigate disputes and exceptions. If the organisation expands service, staff may become more productive without job cuts. If it prioritises cost reduction, recruitment may slow.

This illustrates the economic mechanism, not a prediction for every employer. The outcome depends on demand, system reliability, implementation costs and management choices.

The same technology can support workers or substitute for them. Productivity is a capability; sharing its benefits is a decision.

Banking: faster processing, greater responsibility

Banking offers opportunities for document extraction, customer-query assistance, transaction-pattern analysis and internal knowledge search.

Routine checking and standard communication may require less time. But identifying suspicious activity is not the same as proving fraud, and generating a loan assessment is not the same as making a defensible lending decision.

Employees must understand exceptions, explain outcomes and recognise when a system is wrong. An automated recommendation can reproduce biased data or overlook circumstances that matter to the customer.

For example, a small business’s seasonal income may need contextual assessment. A model’s confident output should not displace that inquiry.

RBI established a committee on the Framework for Responsible and Ethical Enablement of Artificial Intelligence, or FREE-AI, and released its report in August 2025. This is evidence of institutional attention to responsible financial-sector adoption; a committee report should not be confused with every recommendation already becoming a binding direction.

Potential career opportunities include model evaluation, data governance, cybersecurity and customer redress. Their growth, however, will not necessarily match reductions in routine roles one-for-one.

IT: generating code is only part of building software

AI coding tools can help produce drafts, suggest fixes, explain unfamiliar code and create test cases. This puts pressure on work sold mainly as repetitive execution.

Yet functioning software requires more than plausible code. Someone must establish requirements, integrate systems, protect data, evaluate performance and investigate failures.

A suggested fix may work in a demonstration while introducing a security weakness. A generated test may repeat the same mistaken assumption as the code it tests.

The valuable developer increasingly combines tool fluency with engineering judgment. Being able to explain why a solution works matters more than presenting an impressive output.

India’s services businesses face a commercial question: if clients expect faster delivery, will they buy more projects or pay less for existing work? Either outcome can affect hiring.

Opportunities may emerge in AI integration, evaluation, security and industry-specific applications. These are plausible directions, not a guarantee that every displaced programmer will secure an AI role.

Media: faster production cannot replace reporting

AI can assist with transcription, translation, archive search and draft headlines. These uses may give journalists more time for interviews, document examination and field reporting.

But a convincing paragraph can contain a fabricated quotation, incorrect number or unsupported allegation. Producing text is different from establishing facts.

The Reuters Institute’s 2025 research across six countries found greater comfort with supporting tasks such as spelling correction and translation than with artificial presenters or realistic generated news imagery. Only 33% believed journalists always or often checked AI outputs before publication. These are audience perceptions, not measured newsroom compliance rates.

For news organisations, the commercial risk is clear: cutting verification capacity may reduce costs while damaging trust.

Potential work includes verification, data journalism and multilingual publishing. However, those opportunities require investment. A newsroom cannot simply eliminate junior reporting posts and assume experienced investigators will appear later.

Generated illustrations should be labelled. Readers should be able to distinguish a conceptual visual from photographic evidence.

Entry-level careers: the missing first rung

The most difficult question concerns beginners. Routine assignments have traditionally provided a route into professional judgment.

Junior developers learn through small fixes. New bankers learn through supervised cases. Trainee journalists develop accuracy through editing and reporting.

If those assignments are automated, employers may expect recruits to arrive with experience they have had fewer opportunities to acquire.

Stanford Digital Economy Lab’s August 2026 update reports that US workers aged 22–25 in AI-exposed occupations were employed at levels approximately 19% below a comparison path based on less-exposed peers. That is a relative gap, not a 19% fall in all youth employment. The research found no comparable broad pattern for experienced workers.

These US payroll findings cannot establish equivalent losses in India or attribute every hiring decision to AI. Nevertheless, they provide a serious warning about uneven effects on younger workers.

New jobs may arrive without helping everyone equally

Building and operating AI systems requires people who can prepare data, evaluate outputs, redesign workflows and maintain safeguards.

There may also be opportunities for domain specialists: bankers who understand credit, journalists who understand evidence and developers who understand complex systems.

But new occupations can demand qualifications that displaced workers lack. Some may be concentrated in larger cities or available through insecure contracts.

The distribution matters as much as the total. A technology transition can generate income overall while leaving particular workers worse off.

Government and industry should therefore track wages, hiring, working conditions and access to retraining—not merely count AI-related job advertisements.

India needs a strategy built around its workforce

The Economic Survey 2025–26 frames AI as an economic strategy and calls for sensitivity to India’s labour-market realities. It emphasises human capital, experiential learning and aligning adoption with employment and social priorities. These are policy proposals and analysis, not assurances that displacement will be avoided.

India’s opportunity includes applying AI to practical problems: multilingual services, small-business administration, healthcare support and improved public access to information.

However, deployment must account for uneven connectivity, skills and resources. A training programme that assumes advanced English, reliable equipment and spare study time can exclude precisely the workers who need support.

Success should be measured by useful applications and accessible careers, alongside technological performance.

What workers can do—and employers must do

Workers should combine AI literacy with competence in their field. That means recognising limitations, protecting confidential information and checking outputs.

A banker can practise assessing an AI-generated case summary against source documents. A developer can explain and test suggested code. A journalist can trace every factual claim to evidence.

For graduates, a portfolio should demonstrate reasoning: the task, the method, the checks and the correction of errors.

Employers also have responsibilities. They should provide paid learning time, supervised projects and routes for redeployment. Training should prepare employees for real vacancies, rather than merely issue certificates.

Apprenticeships can be redesigned so recruits work with AI while still learning the underlying process. Automation should not erase the institution’s capacity to develop future experts.

Who gets the productivity dividend?

If AI saves time, the gains can support better service, additional output, higher wages or shorter workloads. They can also become higher profits with fewer employees.

That allocation deserves open discussion. Workers should understand how technology will affect their roles and how performance will be judged.

Organisations should measure error rates and customer outcomes alongside speed. A faster decision that repeatedly needs correction may be less valuable than it appears.

The future is being chosen now

AI will create opportunities and replace some work. The balance will vary by occupation, employer and country.

The larger question is whether India will preserve pathways into skilled employment while improving productivity.

A successful transition needs competent workers, accountable employers and training that reaches beyond those already advantaged. Technology can expand human capability. Turning that possibility into secure careers will require deliberate effort.