Advancements in artificial intelligence (AI) are enhancing the capabilities of a growing number of firms to collect, store, analyse, and utilise a vast variety of customer information. While the potential of AI in marketing is widely known, there are also questions about how customer information is utilised and optimising technology’s potential. This is especially true in the era of Big Tech’s dominance, where a consumer provides copious data to search engines, social media platforms, e-commerce websites, ride-hailing services, and hardware providers. The information thus collected is stored, processed, analysed, and used for targeted marketing. It is vital to examine the role of AI technologies in marketing from multiple viewpoints to address extant and emerging concerns and define the use case of AI in marketing more clearly.   

Defining the multi-level model and its relevance

1. Consumer – ethics & privacy concerns: AI technology feeds on a virtuous growth cycle wherein enhanced data collection efforts result in more effective analysis making the collection of consumer data even more effective and efficient, and so on. An outcome of this growth is the issue of data privacy and ethics. However, the degree of awareness of and appreciation for these concerns, and measures to address them, vary considerably across different cultures. For example, European consumers enjoy a high degree of privacy and data protection through the General Data Protection Regulation, or GDPR, a gold standard in data privacy, where Chinese consumers have little to no data privacy, given the country’s prioritisation of national security considerations. Such disparities highlight the issue of ethics and privacy in the use of AI technologies and the degree of heterogeneity among countries in terms of AI regulation.
   
   
2. Company–glocalisation needs: A select few advanced countries have made rapid strides in the development and use of AI technologies, whereas their potential and deployment are not country or region-specific, making local context a relevant factor. Given the wide array of companies that can and do use AI technologies, glocalisation becomes relevant as both AI firms and users of AI technologies seek to manage the tension between global technologies and local adaptation.
   
   
3. Country – economic inequalities: Tremendous economic inequality between developed and developing nations is an acknowledged reality, and AI has the potential to either shrink or widen this gap. Hence, examining if and how AI can be used effectively to ensure such gaps are not exacerbated but plugged is essential. For example, Singapore has a per capita GDP of over $80,000 and is a world leader in the development and use of AI technologies, while Angola, with a per capita GDP of less than $3000, is a laggard. However, both countries can benefit equally from such technology; perhaps the latter can benefit even more than the former, given the lower penetration and absence of compensatory alternatives. Consider the value of an AI application that facilitates customised and self-paced remote learning in Angola (with underdeveloped education services that the entire population cannot access) vs its value in Singapore, which enjoys world-class education provision.    

This three-part (i.e., country, company, consumer) perspective of AI technologies in marketing can help marketing scholars and managers preserve its benefits while highlighting potential concerns for both marketing theory and practice. In turn, this understanding can help allay fears and maximise benefits, making technology a unifier by providing economic opportunities to impoverished countries while allaying consumer concerns about data privacy and misuse.

How does interaction work in the 3-level model (country, company, consumer)? Consider, for example, the implications of a deep neural net algorithm that analysed over 35,000 facial images and correctly distinguished between homosexual versus heterosexual individuals 50% more successfully than human judges in a country that criminalises homosexuality. Consumers providing personal data, including their images, to a company operating in such a country with minimal data privacy laws expose themselves to prosecution, with the company being an inadvertent enabler.

Two key dimensions of AI technologies in marketing

The above three facets (i.e., economic inequality at the country level, glocalisation at the company level, and ethics and privacy at the consumer level) can be employed to examine two key dimensions of AI technologies in marketing:

• Human-machine interaction; and
• Automated analysis of text, audio, images, and video.

From a marketing perspective, these two dimensions are vital because AI is a technology that supplants humans’ role in obtaining data and conducting automated analysis. These two dimensions’ outcomes are expected to revolutionise how firms engage in various sales-related tasks, including targeting, demand estimation, lead generation, and closing sales. 

Let us now look at these two key dimensions at each of the three levels in our multi-level model to understand how these global concerns will likely affect the ability of firms to engage in automated analysis and will also alter the nature of human–machine interactions. Real-life examples illustrate how firms are working towards the goals of reducing inequality, glocalisation, and assuaging privacy concerns:

1) Human-machine interaction (HMI)

1.1: Country-level role of HMI: Economic inequality

Disparities in access to technology and the internet across and within countries is an exacerbated reality in an age of technology dependence. In contrast to promoting this digital divide, HMI can act as a unifier by reducing economic inequality and bringing the world closer together. Numerous instances of HMI-based telehealth services and remote learning opportunities seem to validate this. A good example of how country-specific HMI implementation helps to serve economically disadvantaged customers and reduce inequality is how Amazon employs AI-based digital technologies to assist the marketing efforts of small, uneducated Indian retailers. As a result, retailers who otherwise do not have access to marketing expertise now have access to custom-crafted marketing solutions due to the power of AI-fuelled algorithms.

1.2: Company-level role of HMI: Glocalisation

Organisations, especially MNCs, struggle with the issue of tensions in terms of economically standardising HMI technology vs expensive localisation for the markets they operate in. If HMI is not localised, AI technologies run the risk of market failure. Hence, it is important to train HMI technologies using data drawn from consumers who live and engage across various local markets. For example, to adapt its marketing to local needs, Netflix has designed its machine learning-powered algorithms to develop programming adapted to local consumer tastes.

1.3: Consumer-level role of HMI: Ethics and privacy

Besides concerns over humans losing jobs to machines, there is a larger concern centring around the erosion of human self-determination/agency due to AI’s increasing invisibility and ubiquity, increased alienation, and privacy. For example, Netflix is currently using AI to predict what content customers prefer and design and develop this content (e.g., House of Cards). Another common instance is where AI-enabled assistants such as Amazon’s Alexa make consumer purchase decisions. This may reduce human effort and costs but raises concerns about consumer privacy.

2) Automated analysis of text, audio, images, and video

2.1: Country-level role of automated analysis: Economic inequality

AI benefits individuals residing in economically-disadvantaged countries. In addition, AI’s automated processing capabilities can also help optimise marketing communication efforts across various local (first) languages. For example, firms like Cogito and Chorus use real-time AI speech analytics to listen to conversations between salespeople and their customers and provide instant feedback on an array of dimensions that may impact their performance, such as their voice modulation, speaking tone and degree of empathy. Such solutions can potentially level the playing field for workers in developing countries. Another application addresses disparities in technology access between rich and poor consumers. Many customers in emerging markets such as India cannot afford a personal computer and thus typically watch videos on their phones. Netflix’s Dynamic Optimiser uses an AI-powered algorithm to enhance their viewing experience by analysing videos scene-by-scene and compressing data without affecting image quality, thus providing acceptable-quality streaming to the less well-off on mobile devices.

2.2: Company-level role of automated analysis: Glocalisation

AI technologies, in general, and automated analysis, in particular, are inherently global and universal. However, due to differences in language, culture, or location, text, audio, image, and video analytics, applications are often local. Thus, AI-based text, audio, image, and video analytics are likely to be more successful if locally applied, and AI-enabled automated analysis has the potential to offer powerful new localised insights for achieving this glocalisation. For example, YouTube employs NLP speech recognition software that autonomously produces translation (in the form of video captions) into dozens of different languages. A good example of this type of glocalisation is D-Labs, a new AI-fueled offering capable of extracting brand logos from images posted on Instagram. Considering the ubiquity of these types of social media postings, this automated image analysis can help marketers better adapt to local brand preferences.

2.3: Consumer-level role of automated analysis: Ethics and privacy

The relevance of concerns about ethics and privacy for AI-based automation of text, audio, images, and video is well known. Although AI-based automated analysis of text, audio, images, and video may not always be the cause, its need for massive data to fuel its machine learning algorithms increases the risk (and costs) of data breaches and makes such breaches more likely. To mitigate such risks while at the same time still harvesting the power of automated analysis, firms are becoming increasingly reluctant to store and process individual-level data and instead collect customer data but only store its statistical properties rather than the raw data itself. 

Implications of AI-led technological advancement across the three levels and two dimensions.

AI will likely lead to enhanced design and delivery of locally customised global offerings by enabling a greater understanding of customer behaviour across various local cultures. Such automated intelligence, integrated with existing information management systems to augment human analytical competencies, can transform firms into AI-fuelled organisations that employ systems of human–machine collaboration designed to harness and act on data-driven insights at a local level, while managing the privacy and ethics-related implications.