AI and reputational risks

ORGANISATIONS have used some form or another of artificial intelligence (AI) since its conception decades ago. However, using it for customer-facing purposes has become more prevalent in recent years, and warrants closer attention from the industry.

Previously, AI was used primarily for data analytics and predicting customer behaviour. Today, AI has moved to the front-end of operations to directly engage and interact with customers. While it may allow organisations to be more efficient and effective, it can also lead to reputational risks.

Reputational risks are heightened when AI has direct interaction with a more extensive public audience through customers. This means any mistakes or missteps become more glaringly obvious and, ultimately, more impactful. When customer trust ends up being broken, this causes the organisation to have a tarnished reputation and can magnify negative press coverage and public scrutiny.

Take, for example, the use of AI to increase efficiency and streamline the process of shortlisting resumes, where more than three-quarters of resumes are being analysed by AI. This potentially leads to more biases and discrimination. In 2018, Amazon’s hiring system favoured male candidates over female ones, who tended to earn lower ratings. This arose when the AI showed a preference for male resumes over female resumes (with words such as “executed” and “captured” more commonly found in the former).

Another use for customer-facing AI includes algorithms to promote products or services. This means that when there is a failure to understand the preferences and needs of users, recommendations that are not very useful or helpful to them are generated. This may lead users to feel that they are not being heard or cared for enough, which causes a decrease in customer satisfaction.

In a crowdsourced study conducted in 2022 by Mozilla, YouTube’s AI recommender came under fire for suggesting harmful and hateful content. The top three categories of these types of videos were misinformation (including on Covid-19), violent or graphic content, and hate speech. This affected most non-English speaking countries, particularly France and Germany, where the level of harmful content suggestions was 60 per cent higher than in English-speaking countries. As such, a regulation by the European Commission has been worded to make such platforms more accountable for their suggestion algorithms and transparent about how they work.

Creating a reputation-sensitive culture

To address the growing reputational risks linked with customer-facing AI, organisations need to cultivate a reputation-sensitive culture to prevent such gaffes. Furthermore, it is vital that ethical AI practices are prioritised such that there is fairness, accountability, and safety in the organisation.

Organisations need to involve cross-functional teams to establish a comprehensive AI framework that emphasises fairness, transparency, and detection for bias within the algorithms being used. These should be regularly checked to ensure that compliance and ethical standards are being met with industry regulations. In many organisations, strict guidelines have been disseminated to employees about the use of AI, in order not to jeopardise their reputations.

When using AI, organisations should also adopt a “human-in-the-loop” approach, which calls for people to give feedback to the AI algorithms in place. As humans have more empathy and contextual understanding of the situation, this allows for better detection of ambiguous and sensitive situations, which AI may not necessarily be equipped to view with the same perspective. Today, organisations such as Bloomberg are adopting such methods to ensure smooth workflows.

When organisations are transparent and open about using AI, and also understand that even AI has its limitations, they allow trust to be built with customers. Through this honest conversation, customers better understand AI decisions. Organisations can find out if their customers are happy with the AI assistance they’re receiving through regular surveys, media monitoring and interviews, in order to figure out if more communication needs to take place for a deeper understanding and acceptance. It is also important for organisations to take responsibility for any issues that may occur, to demonstrate accountability and a commitment to serving the customer.

As AI continues to evolve, it will undoubtedly begin to expand even more in the customer-facing frontier throughout the customer journey. But with each new touchpoint, there’s a risk of new and unforeseen reputational risks. Therefore, it’s vital for organisations to continually recognise ethical AI practices and the role which AI plays in their business operations, to use it to their advantage. Keeping these in mind will allow an organisation to maintain its brand image and build the necessary trust with customers and stakeholders, and fully exploit AI for business success.

The writer is managing partner of Europe and Asia for the Penta Group