Artificial Intelligence and Marketing: Pitfalls and Opportunities

Highlights

• What differentiate AI from traditional statistical techniques is the autonomous discovery of higher-order constructs.

• A black box that remains impervious to knowledge transfer will continue to pose significant threats and challenges.

• Such challenges include specification of objective functions, biased or unexplainable AI, and the automation paradox.

• These challenges will plague marketing to a great extent because of the ubiquitous role of tacit knowledge in said domains.

Abstract

This article discusses the pitfalls and opportunities of AI in marketing through the lenses of knowledge creation and knowledge transfer. First, we discuss the notion of “higher-order learning” that distinguishes AI applications from traditional modeling approaches, and while focusing on recent advances in deep neural networks, we cover its underlying methodologies (multilayer perceptron, convolutional, and recurrent neural networks) and learning paradigms (supervised, unsupervised, and reinforcement learning). Second, we discuss the technological pitfalls and dangers marketing managers need to be aware of when implementing AI in their organizations, including the concepts of badly defined objective functions, unsafe or unrealistic learning environments, biased AI, explainable AI, and controllable AI. Third, AI will have a deep impact on predictive tasks that can be automated and require little explainability, we predict that AI will fall short of its promises in many marketing domains if we do not solve the challenges of tacit knowledge transfer between AI models and marketing organizations.

Introduction

Despite the profound impact AI is likely to have on a wide array of business functions, it is somehow disheartening to realize the extent to which some managers still have an insufficient understanding of what AI is, what it can do, and what it cannot.

In an executive education setting, a C-level manager ventured to say that “what is amazing with AI is that, with enough data, it can learn anything.” Given the relative newness and complexity of the subject matter, such misconceptions were to be expected in the short term but present two classes of challenges nonetheless. First, marketing managers who have been sold on the “magical powers” of AI-driven solutions may underestimate its dangers, limitations, and pitfalls. Second, business managers might misjudge the areas where AI is the most likely to bear fruit—and where it will most likely fail—if adopted by marketing organizations, and therefore misallocate their efforts and resources.

This paper aims at addressing these two challenges and is divided into three sections.

The first section is primarily educational and lays the groundwork for the remainder of the paper. We first define AI through the lense of “higher-order learning.” Focusing on the most recent advances in deep learning and artificial neural networks, we then introduce the most common neural network topologies (multilayer perceptrons, convolutional neural networks, and recurrent neural networks), the three learning paradigms in which these networks are trained (supervised learning, unsupervised learning, and reinforcement learning), and how these technologies apply to various marketing contexts.

In the second section, we build upon the concept of higher-order learning to discuss related pitfalls and dangers of AI solutions as they pertain to their adoption by marketing organizations. All machine learning enterprises face common challenges such as the need for abundant data, development of the firm's analytic capabilities (e.g., Kumar et al., 2020), organizational adoption and change management, etc. However, the autonomous generation of higher-order constructs by AI algorithms creates (or aggravates) specific challenges. We focus our attention on these challenges, such as the difficulties to specify valid objective functions, to simulate a safe and realistic learning environment, the risks of—unwillingly—develop and deploy biased, unexplainable, or uncontrollable AI, and the automation paradox.

In the third section, we discuss the impending challenges AI applications will face in marketing where tacit knowledge is crucial—and the facile transfer of tacit knowledge among marketing stakeholders a source of competitive advantage. Recent AI techniques have demonstrated their abilities to learn autonomously from big data and self-generated experience, impervious to human expertise. One could justifiably marvel at these achievements. Nevertheless, we argue that their impermeability to marketing stakeholders' tacit knowledge, while being the source of their early successes, may very well be the cause of their near-term limitations in domains where tacit knowledge is crucial, such as in sales, branding, or relationship marketing.

We conclude by a call for marketing organizations and academic researchers to focus on more efficient tacit knowledge transfer between marketing stakeholders and AI machines.