logo

How are banks building expertise when AI changes the training ground?

Add The Asian Banker on Google
Discover more trusted banking and financial services insights by adding The Asian Banker as a preferred source on Google.
How are banks building expertise when AI changes the training ground?
  • 61

As AI changes junior work, banks must reconsider how employees acquire experience and demonstrate readiness for greater responsibility. Discussions at Insights2040 explored how workplace learning, university reform and closer cooperation between employers, educators and governments could prepare people for changing roles.

Artificial intelligence (AI) can accelerate information gathering, analysis and customer service, while changing the assignments through which newcomers have traditionally acquired experience. Employers may expect more from junior staff even as the route to meeting those expectations becomes less clear.

The Insights2040 forum on 17 November examined these issues, connecting changes in banking work to the education and training needed to support them.

An immediate question for banks is whether faster performance is accompanied by deeper understanding. An employee may produce an assessment with AI assistance without being ready to explain its assumptions, identify missing information or recognise when its recommendations should be challenged.

Higher expectations at the entry level

Familiarity with AI does not necessarily prepare graduates for financial-sector work. Newcomers may be comfortable with the tools but have limited understanding of banking, its operating constraints or the consequences of a poor decision. Employers may nevertheless expect AI to help them contribute at a higher level.

Customer service illustrates how this changes the learning process. When automated systems handle straightforward queries, human agents can be left with dissatisfied customers, unusual requests and situations requiring discretion. The employee’s role becomes more demanding even if the volume of routine work declines.

Analytical work presents a similar challenge. AI can accelerate the preparation of an assessment, but judging whether it supports a sound decision requires knowledge of the business. In project finance, for example, understanding extends beyond the immediate transaction to how financing choices affect a business and its stakeholders over many years. Reviewing cases with experienced colleagues can help newcomers connect their analysis with those longer-term consequences.

Employees need to combine banking knowledge with AI skills, critical thinking, adaptability and collaboration. Technical proficiency extends beyond prompting to using, building with and improving applications. Familiarity with one model offers limited preparation for changing tools and responsibilities.

Junior assignments have often combined production with learning. When automation removes an assignment, banks need to identify which opportunities for practice, feedback and exposure to consequences it provided, and how employees will acquire them elsewhere.

Can AI create a better training ground?

Preserving routine work is not necessarily the answer. Scripted customer-service exchanges may give newcomers practice in following procedures without developing the discretion needed for more complex situations.

AI can support more purposeful preparation by analysing customer interactions to identify difficult cases, simulating unusual conversations and allowing newcomers to practise responses before handling them live. Employees can encounter uncommon problems repeatedly rather than wait for them to arise during routine work.

Coaching can then help employees understand why a response succeeded, what they overlooked and when escalation was appropriate. An exercise that supplies the preferred answer offers less opportunity to develop reasoning than one followed by explanation and feedback.

Temporarily restricting AI access is another approach being tested to build foundational skills. Early experiments address the concern that employees relying on assistance from the outset may struggle to assess their own output. Different activities may require different combinations of independent practice, AI assistance and supervision.

Simulation must also connect with supervised live work. Rehearsing a difficult case can build confidence, while actual assignments expose employees to incomplete information, competing priorities and consequences that an exercise may not fully reproduce.

Who provides the time and supervision?

These approaches depend on how banks allocate the capacity released by automation. Experienced employees could spend more time reviewing junior colleagues’ reasoning and discussing difficult cases. Additional mentoring will not follow automatically, however, if all available capacity is directed towards higher throughput.

Banks therefore need to account for supervision when calculating the productivity gains from AI. Expecting newcomers to handle more demanding assignments may require substantial senior input to prepare them, even when individual tasks take less time to complete.

As AI adoption expands, banks also need employees capable of governing its use and managing the associated risks. Staff must understand an application’s limits, recognise inappropriate outputs and maintain accountability for work undertaken with it.

Assessment needs to distinguish between completing an assignment and understanding it. Employees should be able to explain their reasoning, identify uncertainty and demonstrate when they would seek help or challenge a recommendation. These observations provide a stronger basis for increasing responsibility than speed or volume alone.

The same principle applies to experienced staff moving into redesigned roles. Course completion provides limited evidence that someone can apply AI appropriately within a particular business. Training needs to connect with practical responsibilities and feedback on performance.

From delivering content to developing inquiry

Universities face a related challenge. As AI makes information and ready-made answers more accessible, their value increasingly lies in helping students identify what they need to know and which questions to ask. Students need to learn how to define a problem, recognise missing information and judge whether an answer is relevant before acting on it.

Assessments must reflect that shift. A polished submission offers limited evidence of learning if students cannot explain their reasoning, question assumptions or defend their conclusions. Selective restrictions on AI can preserve opportunities for independent thought, while other assessments can test how effectively students use and evaluate its assistance.

Teaching environments also need to give students more opportunities for discussion, collaborative problem-solving and learning from peers. Educators themselves must become comfortable with AI and reconsider how they teach and assess, rather than treating access to the technology as the sole issue.

Interdisciplinary education can connect AI with finance, law and other fields. Industry practitioners can complement academic teaching by explaining how knowledge is applied under business and risk constraints. Mentored projects, internships and apprenticeships then give students opportunities to test that understanding.

Sustained engagement matters. Building a prototype demonstrates something different from working within an organisation long enough to see how a solution is evaluated, challenged and implemented. Programme effectiveness should therefore be assessed through the skills and employment opportunities participants gain, alongside their technical outputs.

Building judgment and interpersonal skills

Banking knowledge remains necessary for evaluating decisions, but the tasks through which employees apply it will continue to change. Education therefore needs to develop professional foundations alongside the ability to adapt as specialised AI systems become capable of handling more complex work.

A broader liberal arts foundation can complement professional and technical education by helping students examine ethical questions, interpret competing perspectives and communicate their reasoning. Entrepreneurial thinking adds the ability to identify problems, take initiative and work through uncertainty within an established organisation or a new business.

Human interaction also needs deliberate practice. Employees must listen, explain decisions and respond to customers’ circumstances. Within banks, they need to reconcile different perspectives and work with colleagues whose expertise differs from their own. Discussion, teamwork and direct customer engagement belong alongside technical training.

Employees who translate between banking and technical teams draw on these skills. They must define the business problem, explain what the tools can deliver and help colleagues interpret the results. Preparing people for such roles requires practical experience across disciplines.

Sharing responsibility for opportunity

Individuals cannot carry the entire burden of continually preparing themselves for changing work. Employers need to identify how roles are evolving, educators must adapt provision, and governments can support training and job conversion through funding and coordination.

These responsibilities extend beyond graduates. Mid-career employees may face decades of changing work, while ageing societies need ways to support continued participation by older workers. Accessible retraining and flexible working arrangements could help people apply their experience in different ways.

Access also affects whose needs future systems address. Increasing participation among women and underserved groups brings different experiences into the definition of problems and the design of solutions. Recruitment, programme support and access to sustained workplace experience influence who benefits, even when training is formally open to everyone.

AI could also lower barriers to skilled work. Natural-language interfaces may allow domain professionals to use analytical tools without first mastering a programming language.

Separately, lower acquisition, underwriting and collection costs could make underserved banking markets more viable. Serving those customers would create different demands for local knowledge, customer relationships and frontline judgment.

Banks must decide how much of the capacity released by AI to devote to developing their workforce. Higher output does not remove the need for supervised practice, feedback and evidence of sound judgment. Those requirements belong in the design and economics of the new roles.