AI productivity gains should be shared with employees

WHEN a company uses artificial intelligence to help employees accomplish more, how should those employees share in the gains? The question deserves attention alongside the usual calculations of investment costs and financial returns.

Consider a hypothetical customer-service team using AI to prepare responses and summarise inquiries. Employees spend less time on routine paperwork and more time resolving difficult cases.

If the business handles more customers with the same workforce, should the benefit appear only in its bottom line – or also in employees’ earnings and career prospects?

This is not an argument for turning every efficiency improvement into a permanent wage increase. Companies must balance employee remuneration with recovering investment costs and financing future growth.

But a company implementing AI with the aim of boosting productivity should explain both how the business will benefit and how employees can participate.

Two Singapore examples offer useful lessons. Neither is an AI case study, but both illustrate approaches that companies adopting AI can adapt: sharing measured gains and connecting redesigned jobs with better pay.

Share measurable gains

In August 2010, SIA Engineering Company ( SIAEC ) announced a S$600,000 payment to staff in three operating divisions under its productivity gain-sharing programme.

The payment covered gains achieved between February and June that year, making the link between operational improvements and employee rewards tangible.

For a company introducing AI, a similar approach could involve agreeing on a baseline measure of productivity before implementation, assessing improvements and sharing a portion of verified gains through a team payment.

Employees would know what the company was trying to achieve and how their contribution could be recognised.

Measurement needs care. Faster processing is not necessarily better performance if errors increase or customers become dissatisfied. Evaluate gains after factoring in relevant implementation costs, rather than equating them with gross time savings.

SIAEC’s announcement demonstrates an actual payment, but does not establish the scheme’s long-term effects or prove that promised rewards caused the improvements.

The transferable lesson is the arrangement, not an assurance that it will work everywhere. Variable payments may suit situations where the productivity gains can fluctuate; permanent wage increases require confidence that improvements can be sustained.

Redesign jobs, not just tasks

Dr Bags, operated by Leathersolution, illustrates another approach: linking digital transformation and redesigned work with a wage commitment.

In 2023, the bag-restoration specialist introduced a customised enterprise resource planning system to improve processes, including bag tracking, order management and accounting.

A case study by the National Trades Union Congress (NTUC) reported that searching for bags took up to 80 per cent less time and customer waiting times fell by 30 per cent. These were improvements in specific processes, not a company-wide productivity figure.

The company also committed to a 5 per cent wage increase for almost half its 40-person workforce who acquired new skills or took on redesigned jobs. But it should be noted that the announcement records a pledge, not independent confirmation of subsequent payment.

The lesson for AI adoption is to start with the work. Which routine tasks should be reduced? What more valuable responsibilities can employees assume? What training and supervision will they need?

In the hypothetical customer-service team, the aim should not simply be to demand more responses per hour. Employees could take greater responsibility for complex inquiries, service recovery and checking AI-generated answers.

Wage progression should reflect genuine improvements in capability and responsibility, rather than merely the ability to operate a new tool.

Connect investment with worker outcomes

Frameworks to channel business productivity growth back to employees already exist in Singapore.

NTUC’s Company Training Committee Grant, for example, supports up to 70 per cent of qualifying transformation or training costs.

Projects must improve business capabilities or productivity while delivering worker outcomes such as wage increases, skills allowances or implemented career development plans. Committed worker outcomes must be met before claims disbursement.

But the broader lesson applies even without grant support: Plan the employee outcome alongside the business investment.

Public funding can help with transformation costs, but cannot substitute for a viable business case. Management must still assess whether the project makes commercial sense and whether recurring wage commitments remain affordable after funding ends.

Three decisions deserve attention before implementation.

First, define success using measures that capture output, quality and costs.

Distinguish freed-up time from actual financial savings or additional revenue: An hour saved has limited commercial value unless the business can put it to productive use.

Second, match the reward to the result. A one-off gain may justify a one-off payment, while sustained improvements and more demanding roles may support longer-term wage progression. Keep the arrangement affordable without making the employee benefit so vague that it loses credibility.

Third, involve employees in redesigning the work. Explain how results will be assessed, when rewards will be reviewed and what happens if expected gains do not materialise. Employees should understand both the opportunity and the conditions attached to it.

Boards need not design individual payment schemes, but should ask whether management’s AI investment plan connects commercial returns with credible employee outcomes.

The objective is not to divide a hypothetical windfall before it exists. The goal is to give employees a clear stake in making AI useful and ensure that higher productivity means better work, not simply more work.

The writer is a retired investment banker with over 40 years of experience in corporate management, board directorships and advisory work

The commentary is based on the writer’s own experiences, observations and argument. AI tools were used for research and proofreading. The writer remains fully accountable for the commentary’s accuracy, originality and final form.