The use of artificial intelligence (AI) in retail continues to generate discussion. This is largely due to the benefits that retailers have realized by using the technology across various parts of the business value chain.

AI can improve back-office efficiencies and optimize the customer experience. However, the incorrect implementation of AI functionality can result in unintended consequences for retailers.

Retail businesses must therefore carefully consider the intended use case of any AI tool and conduct a risk assessment before implementation. This helps manage the risks that may arise from incorrect or inappropriate use of such tools. The pre-adoption exercise should effectively serve as a due diligence process involving all relevant stakeholders within the business.

Below are insights into some use cases and potential legal and operational risks to consider when introducing AI tools into a retail business.

In a world where customers are spoilt for choice, the struggle to retain them has become more pronounced. Retailers are increasingly seeking innovative ways to capture and maintain consumer attention. The development of AI tools that enable retailers to personalize the shopping experience has created new opportunities to reduce customer churn.

An increasing number of tools are now able to learn customer preferences and tailor the shopping experience accordingly.

Tools that interact with customers directly are also helping retailers gain insights into customer behavior. Chatbots on online platforms are a prime example of how AI can help retailers engage directly with consumers. Customers can ask questions and receive instant responses. These interactions feed the AI behind the chatbot, helping it learn about each customer’s behavior and preferences. This data can then be used to refine how the website engages with customers in the future.

The value of AI tools that enhance the customer experience lies in their ability to draw insights directly from customer behavior. However, in the South African context, a key legal and commercial question arises: whose customer experience is being improved?

In a society where South Africans have unequal access to the internet and where segments of the population do not use online shopping exclusively or frequently, AI tools that learn only from online behavior run the risk of generating biased results. If a tool learns from a sample that is not representative of the broader South African population, the resulting insights may fail to reflect the actual behaviors and preferences of the wider customer base.

From a due diligence perspective, it is important that retailers understand their target market and assess whether the functionality of the AI tool under consideration aligns with the intended use case and the types of insights required.

Proper inventory management is critical to avoiding waste and improving profitability. AI tools that review customer purchasing data and supply chain analytics can assist retailers in predicting future buying trends and eliminating inefficiencies.

To realize the full benefit of such tools, they must be trained on large amounts of high-quality data. AI can only provide useful insights when trained on accurate, relevant, and structured data. In practice, many retailers do not have centralized or well-organized datasets that can be easily integrated into these tools.

If data used to train an AI tool is outdated, inconsistent or inaccurate, no level of technical sophistication can compensate for this. Retailers must therefore undertake a comprehensive data hygiene process before deploying AI tools. This includes removing incorrect, irrelevant or outlier data from the dataset to improve accuracy and reliability.

Another consideration is that many retailers continue to operate on legacy information technology (IT) systems that may not support the integration of AI tools. As part of the broader due diligence process, retailers should determine whether the AI tools they intend to implement are compatible with their existing systems and whether a broader infrastructure upgrade may be necessary.

Beyond large-scale AI implementation projects, there is a quieter and often overlooked area of AI use in many retail businesses. Employees increasingly rely on generative AI (GenAI) tools, such as AI chat interfaces and drafting tools, for quick answers, content generation, research and idea development.

While these tools offer clear benefits, they can also introduce risk if used improperly. For example, if employees input proprietary company information into a GenAI tool whose terms and conditions specify that the provider owns or may reuse uploaded content, this may result in unintended intellectual property exposure. Similarly, sharing confidential or commercially sensitive information via such tools may give rise to data privacy and contractual risk.

Retailers must ensure that employees understand when and how GenAI tools may be used. At a management level, retailers should implement internal policies and training to regulate the use of GenAI and other AI tools.

By Wendy Tembedza, Partner at Webber Wentzel