When an employee uses AI to do eight hours of work in six, employers may assume the remaining two hours are theirs to allocate. In many cases, they will have a strong contractual basis for doing so. But AI raises a second question: who should benefit economically from the productivity gain?
Suppose an employee has a task that routinely occupies a full working day. They discover how to use AI to complete it to the same standard in six hours. Nothing important has been omitted, quality has not fallen and there is no hidden checking time.
What happens to the other two hours?
The conventional answer is straightforward. The employee is being paid to work, not simply to complete one specified task.
Ghana’s Labour Act 2003 (Act 651) broadly supports that position. Sections 8 and 9 give employers substantial authority to organise operations and require them to provide work, while section 11 requires employees to obey lawful instructions, protect the employer’s interests and “enhance productivity”.
So the employer can reasonably say: we are already paying for those eight hours.
Reallocating the saved time does not extend the working day. Section 33 normally limits working time to eight hours a day or 40 hours a week. But that still leaves the economic question. Suppose the employee, rather than the employer, found the AI tool and developed a faster working method. If that creates substantial additional value, should the entire gain necessarily accrue to the employer?
Productivity and reward
There is nothing unusual about expecting a more skilled employee to produce more. Greater skill does not create an automatic entitlement to shorter hours. It can, however, increase an employee’s labour-market value, leading over time to promotion, greater autonomy or higher pay.
AI makes this familiar tension more visible because improvements can be sudden and substantial, particularly where the employee, rather than the organisation, has discovered the better method.
Ghanaian law already recognises that productivity and reward can be linked. Section 98(g) of the Labour Act allows collective agreements to address “the principles for matching remuneration with productivity”. In unionised workplaces, therefore, AI productivity may become a bargaining issue rather than simply a unilateral management decision.
The important distinction is between controlling paid working time and deciding how the economic value created by greater productivity should be treated.
Efficiency is not always a dividend
PwC’s 28th Global CEO Survey found that 73% of the 44 Ghanaian CEOs surveyed reported increased efficiency in employees’ time through GenAI. The financial results were less extensive: 47% reported increased revenue and 43% reported increased profitability.
The following year’s Ghana findings suggested some progress. Eighteen per cent of Ghanaian CEOs reported both higher revenue and lower operating costs from AI, compared with 12% globally.
The figures underline an important distinction. Time saved is not automatically a financial dividend.
If AI reduces a ten-hour task to four hours and the employer uses the released six hours for billable work, the gain has an obvious financial value.
Elsewhere, the benefit may appear through improved quality, fewer errors or better customer service. In some cases, the employer may not be able to use the released capacity productively at all.
The employee who says nothing
Now consider a weekly task that AI reduces from ten hours to four. Employee A tells the manager. The manager congratulates A and fills the released six hours with more work.
Employee B says nothing.
Why might B conceal the gain? One reason is obvious: disclosure may simply produce more work at the same pay. Another is job security. If management discovers that ten hours of work now require only four, might it conclude that fewer employees are needed?
Wharton management professor Ethan Mollick has written about the incentives employees may have to conceal successful AI use where they fear management will respond by increasing workloads or cutting jobs rather than sharing the benefit.
Ghanaian law also regulates the consequences of employer-led technological change. Section 65 applies where an employer contemplates major changes in production, programme, organisation, structure or technology that are likely to lead to terminations. It requires advance written information to the Chief Labour Officer and relevant trade union, together with consultation aimed at avoiding or reducing job losses. Workers made redundant as a result of such changes may also be entitled to redundancy pay, with the amount and terms subject to negotiation.
That protection does not necessarily apply merely because an employee has independently begun using AI more efficiently. Its relevance becomes clearer where management itself turns the productivity gain into a technological or organisational change involving contemplated job losses.
Employee B may nevertheless risk breaching contractual or general duties towards the employer by deliberately concealing substantial spare capacity. The important point is that organisations can still create incentives for such behaviour.
The term “shadow productivity” is sometimes used loosely alongside “shadow AI” to describe employees using AI without formal approval. Here it means something narrower: productivity or capacity gains that management does not know have occurred.
There is no clear published evidence showing that this is already widespread in Ghana. It is better treated as a hypothesis arising from the incentive structure. Yet if disclosed gains consistently produce more work without recognition, employees have less reason to reveal them. Management then loses information needed for staffing, recruitment, pricing and investment decisions.
Who created the improvement?
Not all productivity gains are alike.
If an employer buys an AI system, integrates it into operations and trains employees to use it, the organisation has plainly created much of the gain.
The position is different where an employee independently develops the better method.
That does not mean the employee owns the workflow. Ghana’s Copyright Act 2005 (Act 690) excludes ideas, concepts, procedures and methods from copyright protection. Particular written outputs may still attract copyright, although section 7 generally vests economic rights in works created in the course of employment in the employer unless the contract provides otherwise.
Claims over working methods themselves are more likely to depend on the employment contract, confidentiality obligations and the employee’s duties towards the employer.
The simpler point is that an employee may have originated a valuable productivity improvement. That contribution can matter to reward, retention and willingness to share further innovations.
There is also a compliance boundary. Employees cannot assume that confidential or personal information may be uploaded into public AI tools. Ghana’s Data Protection Act 2012 (Act 843), together with applicable confidentiality duties, continues to govern the processing and disclosure of such information.
Making the implicit explicit
Most employers probably have no formal rule stating that management receives 100% of every AI-generated productivity gain. That may nevertheless become the rule in practice.
Employees will learn the organisation’s real policy from what happens to the first person who reveals a major improvement. Recognition, progression or greater autonomy encourages disclosure. Automatic appropriation of the gain discourages it.
When Ghana’s National Artificial Intelligence Strategy was launched on 24 April 2026, the Government presented its approach as human-centred, with AI intended to enhance human capabilities rather than diminish them.
That creates an obvious employment question.
There is no universal formula. In some cases, the employer will have created most of the gain. In others, employee initiative will have played a larger role. In unionised workplaces, the issue may be bargained. Elsewhere, it may be resolved through company policy, promotion, incentive pay or ordinary labour-market pressures.
Ghanaian law already expects employees to enhance productivity.
AI may now enable them to do so on a scale that previously required substantial organisational investment. That does not deprive employers of control over paid working time. But it does create a second question: when employees help create substantial new economic value through AI, how should that value be allocated?
Author: Zhu Li
Zhu Li is a Director at the Federation of International Employers (FedEE), where she works closely with multinational employers on international HR, workforce and employment issues. Her work spans cross-border HR policy, labour-market developments and emerging workforce risks, with a particular interest in how organisations adapt job design and people practices to changing labour-market conditions. She has contributed analysis to HR publications on subjects including poly-working, global HR compliance and the challenges of managing one workforce across multiple legal jurisdictions.
DISCLAIMER: The views, comments, and contributions made by readers or contributors on this website do not necessarily represent the position or views of The Sikaman Times. The Sikaman Times will not be responsible or liable for any inaccurate or incorrect statements made by readers or contributors on this website.
