Artificial Intelligence and Marketing: Pitfalls and Opportunities
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.
Section snippets
Defining Artificial Intelligence
Whether one surveys psychologists, sociologists, biologists, neuroscientists, or philosophers, the term “intelligence” can take more than 70 different definitions (Legg & Hutter, 2007). It is therefore not surprising that the term “artificial intelligence” (AI), while so commonly used, remains so badly defined, and such a fuzzy concept (Kaplan & Haenlein, 2019). Rather than take a stand, we propose three definitions with varying degrees of inclusiveness.
Pitfalls and Dangers of Artificial Intelligence
The major strength of current AI algorithms lies in their ability to uncover hidden patterns in data and to autonomously create higher-degree constructs from raw data, with limited or no human intervention. For instance, a deep learning perceptron can autonomously identify previously unidentified interactions between predictors, a convolutional network can independently identify and recognize abstract concepts such as “logos” or “eyes” in pictures, and a recurrent neural network can discover
Knowledge Creation and Knowledge Transfer
In the first section of this article, we argue that what truly distinguishes recent AI applications (i.e., deep learning) from traditional statistical methods is their ability to generate higher-order learning, autonomously, and without relying on human expert knowledge. A deep neural network will discover complex relationships among hundreds of seemingly unrelated indicators to predict the likelihood an online visitor will click on an ad, or discover features in images to predict the
Conclusions and Research Agenda
The marketing literature teaches us that an organization can only achieve successful tacit knowledge transfer through shared experience and proximity. Consequently, marketing organizations should seek to facilitate and systematize interactions between AI and marketing stakeholders, and create an ecosystem to foster a form of “intimacy” between AI and experts (or consumers) through two-way observation, imitation, and practice. Tacit knowledge transfer should occur both ways, because as Michael
Declaration of Competing Interest
None declared.
References (103)
- et al.
Modelling and evaluating customer loyalty using neural networks: Evidence from startup insurance companies
Future Business Journal
(2016) Artificial intelligence, machine learning, deep learning, and cognitive computing: What do these terms mean and how will they impact health care?
The Journal of Arthroplasty
(2018)- et al.
Big data consumer analytics and the transformation of marketing
Journal of Business Research
(2016) - et al.
The role of emotions for the perceived usefulness in online customer reviews
Journal of Interactive Marketing
(2016) - et al.
Frontline cyborgs at your service: How human enhancement technologies affect customer experiences in retail, sales, and service settings
Journal of Interactive Marketing
(2020) - et al.
Transforming the customer experience through new technologies
Journal of Interactive Marketing
(2020) - et al.
Siri, Siri, in my hand: Who's the fairest in the land? On the interpretations, illustrations, and implications of artificial intelligence
Business Horizons
(2019) - et al.
Digital analytics: Modeling for insights and new methods
Journal of Interactive Marketing
(2020) - et al.
Brave new world? On AI and the management of customer relationships
Journal of Interactive Marketing
(2020) Explanation in artificial intelligence: Insights from the social sciences
Artificial Intelligence
(2019)
Endogeneity in survey research
International Journal of Research in Marketing.
Neural Networks and Deep Learning: A Textbook
Prediction Machines: The Simple Economics of Artificial Intelligence
A Taxonomy for Learning, Teaching, and Assessing: A Revision of Bloom's Taxonomy of Educational Objectives
When for-profit colleges prey on unsuspecting students
The Atlantic
Machine bias: There's software used across the country to predict future criminals. And it's biased against blacks
ProPublica
Thinking inside the box: Controlling and using an Oracle AI
Minds and Machines
Does selective sales force training work?
Journal of Marketing Research
Race and gender discrimination in bargaining for a new car
American Economic Review
Learning Optimal Online Advertising Portfolios with Periodic Budgets
Alliances speed knowledge transfer
Planning Review
Taxonomy of educational objectives: The classification of educational goals
Man is to computer programmer as woman is to homemaker? Debiasing word embeddings
Ethical issues in advanced artificial intelligence
Ironies of automation
Automatica
Intelligence without reason
What Uber did after the London attack is making people furious
Resistance Report
Learning from peers: Knowledge transfer and sales force productivity growth
Marketing Science
Neural networks: Introduction to artificial neurons, backpropagation algorithms and multilayer feedforward networks
Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
A dictionary of psychology
Artificial Neural Networks
Artificial neural networks architectures for stock price prediction: Comparisons and applications
International Journal of Circuits, Systems and Signal Processing
Artificial intelligence and marketing
Entrepreneurship
Towards A Rigorous Science of Interpretable Machine Learning
Learning how and learning what: Effects of tacit and codified knowledge on performance improvement following technology adoption
Decision Sciences
A new model selection strategy in artificial neural networks
Applied Mathematics and Computation
Optimizing Rhenania's direct marketing business through dynamic multilevel modeling (DMLM) in a multicatalog-brand environment
Marketing Science
Feedforward Neural Network Methodology
How Artificial Intelligence is Transforming Digital Marketing
Learning to Forget: Continual Prediction with LSTM
Artificial general intelligence: Concept, state of the art, and future prospects
Journal of Artificial General Intelligence
How to compute optimal catalog mailing decisions
Marketng Science
Deep learning, Adaptive Computation and Machine Learning Series
Race and Wrongful Convictions in the United States
Knowledge flows and the structure of control within multinational corporations
The Academy of Management Review
A brief history of artificial intelligence: On the past, present, and future of artificial intelligence
California Management Review
Criminal law – sentencing guidelines – Wisconsin Supreme Court requires warning before use of algorithmic risk assessments in sentencing
March 2017
Website morphing 2.0: Switching costs, partial exposure, random exit, and when to morph
Management Science
Website morphing
Marketing Science
Cited by (161)
Can AI fully replace human designers? Matching effects between declared creator types and advertising appeals on tourists’ visit intentions
2024, Journal of Destination Marketing and ManagementDelegation of purchasing tasks to AI: The role of perceived choice and decision autonomy
2024, Decision Support SystemsAI-powered marketing: What, where, and how?
2024, International Journal of Information ManagementIntegration of AI in CRM: Challenges and guidelines
2023, Journal of Open Innovation: Technology, Market, and ComplexityHow does AI drive branding? Towards an integrated theoretical framework for AI-driven branding
2023, International Journal of Information Management Data InsightsExamining the dark force consequences of AI as a new actor in B2B relationships
2023, Industrial Marketing Management