By: Ntsako Baloyi - Data & AI lead for Accenture, South Africa

Across South Africa, organisations are accelerating their investment in artificial intelligence (AI), deploying copilots, automation tools and advanced data platforms with the promise of unlocking productivity and reducing costs. Yet, for many, the expected gains remain uneven – and in some cases, elusive. The reason is simple, but often overlooked: employees are not fundamentally changing how they work.

In many workplaces, AI remains an optional add-on rather than a default way of operating. Even where tools are readily available, employees frequently revert to familiar processes – manual workflows, legacy systems and established habits. This is not resistance in the traditional sense; it is comfort. But in a constrained economic environment like South Africa’s, comfort comes at a cost. 

This creates a growing disconnect between AI investment and actual productivity outcomes. Organisations are spending, but not necessarily saving. They are enabling capability, but not consistently driving adoption. The result is a form of “shadow inefficiency,” where work gets done – but not in the most effective or cost-efficient way.

This highlights a critical shift required in how organisations approach AI transformation. The conversation must move beyond deployment to focus on behavioural change and measurable impact. It is no longer enough to train employees on AI tools; organisations must actively embed and incentivise their use. This raises an important – and sometimes uncomfortable – question: should the use of AI be linked to performance?

In high-performing organisations, productivity is not just about output; it is about how that output is achieved. If AI enables faster delivery, improved quality and lower cost, then choosing not to use it is not a neutral decision. It has direct implications for competitiveness. Leading organisations are beginning to recognise this, embedding AI usage into performance metrics, rewarding efficiency gains, and encouraging experimentation.

This shift is particularly relevant in a workforce that spans generations. Younger employees, even those coming from academic environments where AI use may be restricted, tend to adopt these tools instinctively in the workplace. For them, AI becomes a natural extension of how work gets done. In contrast, experienced professionals – while highly skilled – may take longer to integrate AI into their workflows, not due to lack of capability, but because of deeply ingrained ways of working.

The risk is not a skills gap, but an adoption gap. Closing this gap requires more than training programmes or access to new tools. It requires organisations to redefine what “good performance” looks like. Efficiency, adaptability and the ability to leverage technology effectively must become core indicators of success. Incentives – whether through recognition, progression or rewards – should reinforce behaviours that maximise the value of AI investments.

At the same time, organisations must ensure that AI adoption translates into real, measurable outcomes. Key questions need to be asked consistently: Are tasks being completed faster? Is the employee experience improving? Are customers seeing better outcomes? Is the cost to serve decreasing? Without clear answers, the issue is not whether AI works—it is whether it is being used effectively.

There is also a sequencing challenge that many organisations overlook. Before investing heavily in sophisticated AI solutions aimed at transforming customer experiences, there is an immediate and often more accessible opportunity: improving how employees work. The internal productivity dividend is typically the fastest and most tangible return on AI investment. When employees are empowered – and expected – to use AI effectively, the benefits cascade into better service delivery, improved customer experiences and stronger financial performance.

The success of AI in the workplace will not be defined by the sophistication of the tools, but by the willingness of people to use them differently. In South Africa’s current economic climate, where organisations are under pressure to do more with less, that distinction is critical. AI has the potential to transform productivity – but only if organisations move beyond access and focus on adoption, accountability and outcomes.