Beyond the Hype: Navigating the Ethics of AI-Driven Marketing

Introduction

The advancement of artificial intelligence (AI) and its growing influence have sparked excitement and concern, particularly in marketing. AI-driven practices such as microtargeting, which uses data analytics to target individuals with tailored messaging, and hypernudging, which employs AI algorithms grounded in programmatic advertising to deliver persuasive prompts (Yeung, 2017), have emerged as powerful tools for shaping consumer behaviour. While some hail these practices as revolutionary, promising unprecedented personalisation and efficiency, they have also raised ethical questions regarding their impact on individual autonomy — an individual’s capacity for self-governance and the ability to make decisions based on their values, beliefs, and preferences (Dworkin, 1988).

The discourse surrounding the ethical implications of AI-driven marketing has been marked by polarising narratives and hype cycles, from dystopian visions of AI as an existential threat to techno-optimistic views of AI as a panacea for all of society’s ills. Amidst these competing narratives, nuanced discussions about the efficacy and impact of AI-driven marketing have often been obscured, with hype generating inflated expectations and serving various agendas.

This essay seeks to provide a balanced perspective on the ethical implications of AI-driven marketing. By examining case studies like Cambridge Analytica and the Ad Council’s COVID-19 vaccination campaign and the underlying psychology, behavioural economics, and mechanics of programmatic advertising, it aims to shed light on the complex interplay between these practices and autonomy.

The essay argues that while AI-driven marketing raises valid concerns about manipulation and the erosion of autonomy, much of the current discourse is driven by simplistic narratives that fail to capture the full complexity. It advocates for developing ethical frameworks and design principles that balance the benefits of personalisation with respect for personal agency and privacy while cautioning against deterministic views that discredit our collective ability to make informed decisions.

Roadmap. The essay proceeds in six sections. Section 1 places generative AI and AI marketing in the Gartner hype cycle and shows how each phase distorts the public conversation. Section 2 traces the economic and political incentives that perpetuate those distortions across media, industry, and government. Section 3 unpacks the mechanics of programmatic advertising and the psychological mechanisms — System 1 thinking, Cialdini’s principles, behavioral economic biases — that microtargeting and hypernudging exploit. Section 4 examines two competing conceptions of autonomy (Kantian and substantive) and uses them to anticipate the strongest scholarly objections to a “balanced” view, including arguments from Zuboff, Crawford, and Faraoni that the asymmetry of power between platforms and individuals renders procedural protections insufficient. Sections 5 and 6 ground the analysis in two empirical case studies — Cambridge Analytica’s overhyped political microtargeting, and the Ad Council’s successful COVID-19 vaccination campaign — before Section 7 proposes a concrete ethical framework with seven design principles. The conclusion returns to the autonomy question and argues that the substantive view should anchor the next generation of regulation.

Ultimately, the goal is to contribute to a more nuanced and evidence-based discussion about the ethics of AI-driven marketing, moving beyond the hype and focusing on the real challenges and opportunities presented by these tools. This essay seeks to chart a path toward a future in which AI benefits can be harnessed while safeguarding human dignity.

Generative AI and AI Marketing Hype Cycles: Inflated Expectations and Consequences

The hype cycles surrounding generative AI and AI for Marketing share striking similarities in generating inflated expectations, obscuring nuanced discussions, and serving various agendas. Both cycles prioritise sensationalism over substantive analysis, leading to distorted public perceptions of the technologies’ capabilities and potential consequences. Using the ‘Gartner hype cycle’ model, which is a graphical representation of the maturity, adoption, and social application of specific technologies, we can seek to explain where generative AI and AI in marketing are in their hype cycles.

In the context of generative AI, the current hype cycle appears to be in the “peak of inflated expectations” phase, also referred to as an “AI boom” or “AI spring” (Gartner, 2023; Hajkowicz et al., 2023). This phase is characterised by rapid progress and heightened expectations, as exemplified by recent advancements in generative AI led by OpenAI (Heaven, 2023). The release of large language models like GPT-3.5 (used in ChatGPT) and GPT-4 has further fueled the AI boom, garnering attention for their ability to generate high-quality, human-like text (OpenAI, 2023).

However, the generative AI hype cycle has also been shaped by two polarising narratives: the apocalyptic doomers characterised as “a furor” of “online pseudo experts” making dire warnings about existential risks, while others, especially those in the VC-backed “effective accelerationism” movement, hype the technology’s transformative potential while ignoring safety concerns (Andreessen, 2023; King & Prasetyo, 2023). Prominent AI leaders and tech companies have perpetuated these narratives to serve their interests, such as generating investment and shaping favourable regulatory landscapes (Roose, 2023; Merchant, 2023; Heaven, 2023). While some experts criticise the focus on hypothetical risks as a distraction from AI’s real-world impacts (Heaven, 2023), the doomer narrative continues to dominate public discourse, obscuring more immediate concerns like bias, privacy violations, and job displacement (Wong, 2023).

In contrast, according to Gartner, the AI-driven marketing hype cycle is currently in the “trough of disillusionment” phase, which is characterised by waning interest as implementations fail to deliver expected results (Gartner, n.d.). Despite the initial hype, studies show that most advertisements have a limited impact on consumer decisions (Hsu, 2019), and marketers face challenges such as data availability and skills gaps (Gartner, n.d.). This disconnect is reflected in high Chief Marketing Officer turnover rates and low marketing department Net Promoter Scores (Graham, 2022; Neff, 2023). The hype cycle, fueled by rapid industry evolution and pressure to adopt new technologies, leads to resource misallocation and erodes trust in the profession (Hsu, 2019). However, as marketers start to use novel AI implementations, such as emotion AI, influence engineering, and generative AI (Gartner, n.d.), there is potential for the AI-driven marketing hype cycle to move towards the slope of enlightenment phase, where the benefits of the technology become more widely understood and accepted.

This disconnect between the capabilities marketers claim to have, and the reality of their impact has significant consequences, leading to wasted resources and increasing scepticism of advertising claims. However, the application of AI in marketing is giving marketers renewed optimism that they will be able to affect their companies’ bottom lines meaningfully. As AI-driven marketing practices evolve, there is a growing recognition of the need for ethical frameworks prioritising user autonomy, transparency, and accountability.

While AI and marketing hype cycles share common elements, their specific narratives, stakeholders, and potential consequences differ. The AI hype cycle focuses on long-term existential risks and transformative potential, influencing public perception and policy discussions around regulation and development. The marketing hype cycle, on the other hand, emphasises short-term business outcomes and data-driven optimisation, distorting resource allocation and eroding consumer trust.

It is important to note that hype cycles are not deterministic and that technologies can experience multiple hype cycles, followed by periods of disappointment and criticism, known as “AI winters” (Crevier, 1993, p. 203). Some may argue that hype cycles serve a valuable purpose in driving innovation and attracting investment. However, the consequences of inflated expectations and oversimplified narratives can be severe, leading to a lack of preparedness for real challenges, development of poorly aligned systems, and the erosion of public trust. Promoting nuanced, evidence-based discussions and working towards more grounded and responsible approaches to developing and deploying these technologies is essential to mitigate these risks.

The Perpetuation of Hype Cycles: Media, Public Discourse, and Incentives

The media and public discourse play a significant role in perpetuating AI and marketing hype cycles. Sensationalised coverage often emphasises the most dramatic aspects of these technologies, such as AI’s potential to surpass human intelligence (Galeon & Reedy, 2017) or the success of the most outrageous marketing campaigns (Heinz: A.I Ketchup, n.d.). The public’s fascination with these stories amplifies this sensationalism, leading to the rapid spread of misinformation and exaggerated claims through social media (Hoff, 2020; Boutin, 2020).

The modern attention economy, where companies vie for user attention and monetise it instead of relying on traditional monetary exchange, exacerbates these issues. It incentivises media outlets to generate quick, easily digestible content that appeals to the public’s desire for simple narratives (Meyer, 2018). This can lead to oversimplified narratives that obscure the true complexities of AI and marketing practices (Floridi, 2020; De Bruyn et al., 2020), hindering the development of appropriate governance and regulatory frameworks (Calo, 2017) and leading to wasted resources and a lack of accountability for campaigns’ true impact (Allenby et al., 2021; Vosoughi et al., 2018).

Economic and political incentives also contribute to the perpetuation of hype cycles. AI companies benefit from hype as it attracts investment, talent, and consumer interest (Malik, 2020; Bughin, 2017). By framing AI development as a response to existential risks or a race against competitors, these companies create a sense of urgency that justifies rapid, unconstrained development and discourages regulatory oversight (Heaven, 2023). This aligns with the tech industry’s aversion to regulation and preference for self-governance, which can lead to the development of biased, opaque, or poorly aligned AI systems (Crawford, 2021) and the concentration of power and wealth in a few dominant tech companies (West et al., 2019).

Similarly, marketers benefit from the hype around data-driven advertising as it allows them to justify larger budgets and maintain relevance (Big Data, Analytics, and the Future of Marketing & Sales, 2015), often at the expense of consumer privacy (Zuboff, 2019). The prioritisation of hype over nuanced discussion in marketing can erode consumer trust and amplify misinformation (Bechmann & Nielbo, 2018).

Political actors also exploit hype cycles to garner support for policies benefiting their nations’ tech industry or justify protectionist measures (Kriechbaum et al., 2021). Politicians can claim voters were misled by nefarious advertising to discredit opponents, as seen in the Cambridge Analytica scandal and the Brexit referendum. By perpetuating hype around these techniques’ effectiveness, political actors sow doubt about electoral legitimacy, undermine trust in democratic processes, and distract from substantive discussions about factors shaping voter behaviour. This further polarises the public and hinders the development of effective governance frameworks prioritising accountability, fairness, and the protection of individual rights (Yeung et al., 2020).

While some may argue that hype cycles are inevitable in technological innovation and political discourse, prioritising hype over nuanced discussion can have far-reaching consequences. These include the development of biased or poorly aligned AI systems (Crawford, 2021), the concentration of power in a few dominant tech companies (West et al., 2019), the erosion of consumer trust, and the amplification of misinformation (Bechmann & Nielbo, 2018). To mitigate these risks and promote more grounded and responsible approaches to the development and deployment of these technologies, it is crucial to facilitate diverse, inclusive, and transparent discussions (Cath et al., 2018) and develop governance frameworks that prioritise accountability, fairness, and the protection of individual rights (Yeung et al., 2020). This can be achieved by encouraging journalists to seek diverse perspectives and provide context and analysis (Allenby et al., 2021) while cultivating greater media literacy among the public (Vosoughi et al., 2018).

Ultimately, breaking free from the hype cycles that dominate discussions of AI and marketing requires a concerted effort from all stakeholders to promote nuanced, evidence-based perspectives and resist the temptation to oversimplify complex issues. By recognising and critically examining the role of media, public discourse, and the economic and political incentives that drive these hype cycles, we can work towards a more responsible approach to the development and deployment of these powerful technologies, ensuring that their benefits are realised while mitigating potential risks and negative consequences.

The Mechanics and Ethics of Microtargeting and Hypernudging in Programmatic Advertising

Microtargeting and hypernudging leverage psychology, behavioural economics, and programmatic advertising to influence individual decision-making.

Programmatic advertising relies on automated bidding processes and real-time auctions to deliver personalised ads to individuals based on their online behaviour, demographic characteristics, and inferred preferences. AI algorithms analyse vast troves of data to identify patterns and predict which ads are most likely to resonate with specific individuals, enabling marketers to deliver highly targeted messages at scale.

From a psychological perspective, these practices often target System 1 thinking, which is fast, intuitive, and emotionally driven, leveraging cognitive biases and heuristics to influence decisions without engaging the more critical and reflective System 2 (Kahneman, 2011). Cialdini’s six principles of persuasion — reciprocity, commitment and consistency, social proof, authority, liking, and scarcity — are also exploited by tailoring messages and offers to align with these principles (Cialdini, 1984).

Microtargeting and hypernudging also capitalise on behavioral economic phenomena such as loss aversion, social proof, and the default effect. As Thaler and Sunstein (2008) describe, a “nudge” is any aspect of the choice architecture that predictably alters behaviour without forbidding options or significantly changing economic incentives. These practices can be seen as digital nudging, using data-driven insights to design personalised choice environments that guide individuals toward desired actions.

However, the use of nudges in programmatic advertising raises ethical concerns. While they can promote positive behaviours, they can also manipulate individuals into making purchases or sharing personal data that may not be in their best interests (Thaler & Sunstein, 2008). The line between persuasion and manipulation is often blurred, and the lack of transparency can undermine autonomy and informed decision-making.

Moreover, the impact of these practices on individual decision-making is complex. While some argue that constant exposure to personalised messages can reduce critical thinking and increase susceptibility to manipulation (Faraoni, 2023), others suggest that consumers are not passive recipients and retain the ability to resist or reject persuasive attempts (Darmody & Zwick, 2020).

It is crucial to recognise the potential for misuse of these techniques. Government agencies have been reported to purchase individuals’ information from data brokers to circumvent legal restrictions on surveillance and data collection (U.S. Office of the Director of National Intelligence, 2022), highlighting the risk of these techniques being co-opted for purposes that may undermine individual privacy and civil liberties. This underscores the need for robust ethical frameworks and regulatory oversight to mitigate the risks of misusing microtargeting and hypernudging.

Furthermore, the impending deprecation of third-party cookies by major web browsers is likely to exacerbate these concerns by further consolidating power among platforms with vast reserves of first-party data (Mellet & Beauvisage, 2020). As these companies gain even greater control over the digital advertising ecosystem, the potential for misuse may increase. This development reinforces the importance of developing ethical frameworks and regulatory measures that prioritise transparency, accountability, and the protection of user autonomy in the face of increasingly sophisticated persuasive technologies.

Ultimately, the mechanics of programmatic advertising enable microtargeting and hypernudging at an unprecedented scale, offering benefits in terms of personalisation and relevance but also raising significant ethical concerns. As these practices continue to evolve, it is crucial to develop robust frameworks for transparency, accountability, and user empowerment to ensure that the power of persuasive technology is harnessed responsibly and in a manner that aligns with the broader principles of individual privacy, autonomy, and democratic values discussed throughout this essay.

The Philosophy of Autonomy and the Implications of AI-Driven Marketing

AI-driven marketing practices like microtargeting and hypernudging have brought the philosophical concept of autonomy to the forefront of debates surrounding persuasive technology ethics. Autonomy, an individual’s capacity for self-governance and self-determination, is a central value in liberal democratic societies.

Different conceptions of autonomy can inform the development of ethical frameworks for AI-driven marketing. Wolff’s Kantian perspective suggests that practices bypassing or undermining rational decision-making may threaten autonomy. In contrast, Bird’s more expansive view implies that practices exploiting psychological vulnerabilities or manipulating emotional responses may undermine autonomy by interfering with authentic self-expression.

The distinction between procedural and substantive autonomy is particularly relevant when assessing AI-driven marketing practices. While procedural autonomy may be satisfied if individuals are aware of targeting and can opt out, this formal autonomy may be illusory if targeting is pervasive and sophisticated enough to undermine authentic choice effectively. A substantive conception of autonomy would be more sensitive to how microtargeting and hypernudging can shape desires, beliefs, and values over time.

Anticipating the strongest counterargument. The most serious scholarly objection to this essay’s “balanced” position comes from Zuboff (2019) and Faraoni (2023), who argue that the asymmetry between platforms (with vast first-party data, behavioural prediction models, and the ability to A/B test interventions on millions of users in real time) and individuals (with bounded cognition, no audit visibility, and no comparable counter-tooling) is itself sufficient to defeat any procedural autonomy claim. On this view, the empirical evidence that any single ad has a small effect on any single consumer is the wrong unit of analysis: what matters is that the aggregate system shapes the choice architecture within which all consumer decisions are made, and individuals cannot meaningfully opt out of that architecture. Crawford (2021) extends this to argue that the very practice of treating consumers as data subjects to be optimised against constitutes a form of harm regardless of conversion outcomes.

This objection has real force, and a serious response cannot just point to small effect sizes. The reply this essay endorses is twofold. First, even granting Zuboff and Crawford’s structural argument, the policy implication is not “ban personalization” but “redesign the choice architecture to be auditable, contestable, and substantively bounded” — which is exactly what the substantive-autonomy framework proposed below is meant to do. Second, the small-effect-size literature (Bail et al., 2020; Kalla & Broockman, 2018) does not refute the structural critique; it refutes the specific claim that any one campaign meaningfully swung a specific election. Those are different propositions. Conflating them — which is what most of the post-Cambridge Analytica discourse does — produces both bad policy and bad scholarship. The substantive view can hold both: that the structural power asymmetry is real and demands intervention, and that the individual campaign was probably less decisive than its critics claimed.

The implications for liberal democratic values are significant. If AI-driven marketing undermines autonomy by manipulating preferences and decisions, it may erode liberal democracy’s foundations. AI algorithms’ opacity and complexity may threaten the ideal of informed consent, and the use of these practices in politics raises concerns about manipulating public opinion and skewing electoral outcomes.

The hype cycles surrounding AI and marketing create a false sense of inevitability, obscuring human agency and responsibility in shaping these technologies. Preserving autonomy will require a commitment to human agency, accountability, and democratic deliberation.

A substantive conception of autonomy that accounts for the quality and authenticity of decisions should guide the development of ethical frameworks for AI-driven marketing. This approach would justify stricter limits on microtargeting and hypernudging to protect against manipulation and undue influence, recognising the subtle ways these practices can shape beliefs and values over time. It would also emphasise the need for transparency, accountability, and user empowerment to ensure that individuals can make autonomous choices in the face of increasingly sophisticated persuasive technologies.

Case Study: Cambridge Analytica and Political Microtargeting

The Cambridge Analytica scandal revealed the potential for data misuse and lack of transparency in political microtargeting. The firm harvested personal data from millions of Facebook users without consent to build psychographic profiles and target voters with personalised political ads during the 2016 US presidential campaign (Rosenberg et al., 2018). While the scandal generated significant hype around the effectiveness of psychographic targeting, with claims that these techniques played a decisive role in Donald Trump’s election, this hype obscured more fundamental issues of data misuse and lack of transparency.

Experts have cast doubt on the actual efficacy of Cambridge Analytica’s techniques. The Trump campaign primarily phased out its data by the general election, relying instead on more accurate Republican National Committee data (Garrett, 2018). Moreover, the case highlights how narratives of manipulation can diminish genuine voters’ autonomy. By focusing on psychographic targeting’s supposed power to sway elections, such narratives risk portraying voters as passive subjects, easily manipulated by data-driven techniques. However, research suggests that political microtargeting effects are often overstated, and voters’ preexisting beliefs and preferences play a more significant role in shaping their behaviour (Bail et al., 2020).

Drawing on insights from psychology and behavioural economics is essential to develop a more nuanced understanding of the case and political microtargeting. Motivated reasoning research suggests people are more likely to accept information confirming their beliefs and reject challenging information (Kunda, 1990), limiting political microtargeting effectiveness. Additionally, the concept of “filter bubbles” (Pariser, 2011, p. 10) suggests personalised online content can create echo chambers reinforcing users’ beliefs, exploiting broader patterns of selective exposure and confirmation bias.

Other case studies offer valuable insights into political microtargeting efficacy. Hersh (2015) found that campaigns’ use of big data and microtargeting had limited effects on voter turnout and persuasion, with traditional factors like party affiliation playing a more significant role. Similarly, a meta-analysis by Kalla and Broockman (2018) found that the average effect of online political ads on voter choice was small, with most studies finding null or negligible impact.

While some argue that the Cambridge Analytica scandal demonstrates the power of data-driven techniques to manipulate voters, a closer examination reveals a more complex picture. The case highlights the need for a nuanced understanding of political microtargeting’s effectiveness, considering the interplay of individual, social, and technological factors shaping voter behaviour. Moving beyond simplistic narratives of manipulation and examining the empirical evidence, we can develop more responsible approaches to data and AI in political campaigns, prioritising transparency, accountability, and respect for voter autonomy.

The Cambridge Analytica case underscores the importance of the role of hype cycles in obscuring nuanced discussions and the need for interdisciplinary perspectives to assess the impact of AI-driven marketing practices on individual autonomy. As we navigate the ethical implications of these technologies, it is crucial to ground our discussions in empirical research and resist the temptation to oversimplify complex issues. Only by engaging with the nuances and complexities of these cases can we develop effective and responsible frameworks for using data and AI in marketing and politics.

Case Study: The Ad Council’s COVID-19 Vaccination Campaign and AI Targeting

The Ad Council’s “It’s Up to You” campaign, launched in 2021, illustrates how AI-driven personalisation can be harnessed to promote public health goals. By leveraging AI algorithms and data analytics to tailor messaging based on demographics, geography, and online behaviour, the campaign effectively engaged hesitant audiences and contributed to increased COVID-19 vaccination rates (Briggs et al., 2023; Montgomery, 2021).

The campaign’s success, which included a 2% increase in total vaccinations and an estimated 21,000 hospitalisations and 3,500 deaths averted (Briggs et al., 2023), demonstrates the potential benefits of AI-driven personalisation in public health contexts. However, it raises important ethical considerations regarding individual autonomy and privacy (Gasser et al., 2020).

To balance the benefits of personalisation with respect for individual autonomy, the Ad Council prioritised transparency and empowering individuals with knowledge. By directing users to the GetVaccineAnswers.org website, where they could find accurate information and make informed decisions (Montgomery, 2021), the campaign sought to avoid the pitfalls of more coercive persuasion forms.

The “It’s Up to You” campaign also highlights the importance of a multi-faceted approach to public health messaging, combining AI-powered personalisation with community outreach through trusted messengers (Montgomery, 2021). This approach recognises that while AI can help deliver relevant messages, the success of public health campaigns often relies on the trust and credibility of local leaders and advocates.

Other examples of AI-driven personalisation in public health, such as the UK’s National Health Service chatbots (Nadarzynski et al., 2019) and AI-powered symptom checkers and triage tools (Fraser et al., 2018), offer further insights into the potential benefits and risks of these techniques. While they can improve access to information and support, they also raise concerns about data privacy and the need for ongoing evaluation to ensure effectiveness and fairness.

Critics of AI-driven personalisation in public health messaging may argue that it is a form of manipulation, exploiting personal data to influence individual decision-making. They may also raise concerns about the potential for these techniques to exacerbate existing inequalities, as marginalised communities may be less likely to have access to or trust in digital technologies. However, proponents argue that AI-driven personalisation can be valuable for promoting public health and engaging hard-to-reach populations when implemented responsibly and transparently.

The Ad Council’s COVID-19 vaccination campaign underscores the importance of nuanced approaches that balance the benefits of AI-driven personalisation with respect for individual autonomy and privacy. As we navigate the ethical implications of these technologies in the public health sphere, it is crucial to prioritise transparency, empowerment, and collaboration with trusted community partners.

By examining the Ad Council’s campaign in the context of the broader discussions surrounding AI hype cycles, the mechanics and ethics of microtargeting and hypernudging, and the philosophy of autonomy, we can develop a more comprehensive understanding of the opportunities and challenges presented by AI-driven personalisation in public health. This case study contributes to the urgent call for developing ethical frameworks and evidence-based assessments to guide the responsible deployment of these powerful tools for the greater good.

Towards Ethical Frameworks for AI Marketing

As AI-driven marketing practices evolve, developing ethical frameworks that prioritise user autonomy, transparency, and accountability is paramount. These frameworks should be grounded in empirical research and insights from relevant disciplines to effectively address AI’s complex ethical implications in marketing.

To strengthen ethical AI marketing frameworks, it is essential to propose design principles prioritising transparency, meaningful consent, and the prevention of manipulation. Fundamental principles include ensuring users are fully informed about AI systems, providing transparent data collection and usage information, and designing interfaces that facilitate informed decision-making and respect user preferences (Bublitz, 2020).

Specific regulatory measures and industry practices are crucial to operationalising these frameworks. Mandatory algorithmic audits can ensure transparency and accountability, allowing for the identification and mitigation of biases, errors, or manipulative practices. Strengthening data privacy protections and requiring explicit, informed consent for data collection and usage can empower users to make autonomous decisions about their personal information. Clear guidelines for obtaining and managing user consent, such as standardised interfaces and industry-wide opt-out platforms, can ensure users are fully informed and empowered. The potential role of industry self-regulation in creating ethical codes of conduct and independent oversight bodies should also be explored.

A comprehensive ethical framework for AI marketing should encompass the following fundamental principles:

1. Transparency and disclosure: Ensuring clear and accessible information about the presence and functioning of AI systems, data collection, usage, and sharing practices, as well as standardised interfaces for obtaining informed consent.
2. Accountability and oversight: Implementing mandatory algorithmic audits, establishing independent oversight bodies, and defining clear lines of responsibility and liability for AI system developers and deployers.
3. Fairness and non-discrimination: Proactively identifying and mitigating biases in AI systems and data sets, ensuring equitable access to AI-driven personalisation benefits, and preventing discriminatory targeting or exclusion of vulnerable groups.
4. Respect for user autonomy: Designing interfaces that facilitate informed decision-making, providing meaningful choices and control over data collection, usage, and personalisation, and avoiding manipulative or deceptive practices that undermine user agency.
5. Privacy and data protection: Strengthening data privacy regulations and enforcement mechanisms, requiring explicit and informed consent for data collection and usage, and implementing privacy-by-design principles in AI system development.
6. Interdisciplinary collaboration: Fostering collaboration among AI developers, marketers, ethicists, and policymakers, integrating insights from relevant disciplines into AI system design, and promoting public dialogue and engagement on the ethical implications of AI in marketing.
7. Continuous evaluation and adaptation: Regularly assessing and updating ethical frameworks in light of technological advancements, monitoring AI systems for unintended consequences or emerging risks, and maintaining flexibility to adapt regulations and practices based on empirical evidence and stakeholder feedback.

Implementing such a framework will undoubtedly present challenges, requiring coordination, resources, and political will. However, by proactively addressing AI’s ethical implications in marketing and committing to evidence-based solutions, we can work towards a future where AI benefits businesses and consumers while safeguarding transparency, accountability, and respect for human autonomy.

Conclusion

This essay has explored the relationship between AI-driven marketing and autonomy, examining the hype cycles that generate inflated expectations, obscure nuanced discussions, and serve various agendas. By drawing on interdisciplinary perspectives from psychology, behavioural economics, and philosophy, we have highlighted the need for a more balanced understanding of the efficacy and impact of microtargeting and hypernudging.

Empirical evidence suggests that the effectiveness of these techniques is sometimes less dramatic than claimed (Soltani et al., 2019; Matz et al., 2017). Meta-analyses and studies have found relatively small effects on consumer behaviour and purchasing decisions, underscoring the need to avoid oversimplifying the power of AI-driven marketing and to examine their impacts on a case-by-case basis.

Even if the impact of AI-driven marketing techniques is less extensive than sometimes portrayed, they still warrant careful consideration due to their potential to influence individual decision-making and shape societal outcomes. On one hand, these techniques can be harnessed for beneficial purposes, such as public health campaigns (Briggs et al., 2023) or efforts to enhance informed decision-making and democratic participation (Darmody & Zwick, 2020). Personalized interventions promoting healthy lifestyle choices (Mohan et al., 2021) and targeted outreach increasing voter turnout and political knowledge among underrepresented groups (Endres & Kelly, 2018) demonstrate the potential for AI-driven marketing to drive positive behavioural changes. However, the increasing sophistication of microtargeting and hypernudging also raises concerns about the potential for misuse, such as the erosion of privacy, autonomy, and democratic values (Darmody & Zwick, 2020; Mellet & Beauvisage, 2020). The impending deprecation of third-party cookies by major web browsers will likely exacerbate these concerns by further consolidating power among platforms with vast reserves of first-party data, increasing the potential for misuse and underscoring the need for robust ethical frameworks and regulatory oversight.

Avoiding misuse of AI-driven marketing requires a commitment to nuance and a recognition that its trajectory is not predetermined but the product of deliberate choices made by researchers, developers, policymakers, and industry stakeholders. To realise the potential benefits of these technologies while mitigating their risks, we must prioritise the development of robust ethical frameworks and evidence-based impact assessments.

This essay proposes a multi-faceted ethical framework encompassing principles of transparency, accountability, fairness, and respect for user autonomy. Implementing such a framework will require interdisciplinary collaboration, stakeholder engagement, and a willingness to adapt based on empirical research and evolving technological capabilities.

Critics may argue that focusing on ethical considerations and regulatory oversight could stifle innovation and limit the potential benefits of AI-driven marketing. However, a proactive approach to addressing these issues is essential for building trust, promoting responsible innovation, and ensuring that these technologies are developed and deployed to prioritise individual rights, participatory democracy, and societal cohesion.

As we navigate the future of AI-driven personalisation, we are responsible for shaping these technologies to uphold ethical principles and preserve human agency. This requires resisting the allure of hype, engaging with the nuances and complexities of these issues, and remaining committed to developing guidelines and regulations that protect individual rights while enabling responsible innovation.

Ultimately, the path forward lies in cultivating a culture of transparency, accountability, and user empowerment within the digital marketing ecosystem. By establishing clear frameworks that safeguard privacy, autonomy, and democratic values while promoting the beneficial applications of AI-driven marketing, we can work towards a future in which these technologies are harnessed for the betterment of both businesses and consumers.

References

• Andreessen, M. (2023, October 16). The Techno-Optimist Manifesto. Andreessen Horowitz.
• Bail, C. A. et al. (2018). Exposure to opposing views on social media can increase political polarization. PNAS, 115(37).
• Bechmann, A., & Nielbo, K. L. (2018). Are we exposed to the same “news” in the news feed? Digital Journalism, 6(8).
• Big Data, Analytics, and the Future of Marketing & Sales (2015). McKinsey & Company.
• Bird, C. (1999). Liberalism, Autonomy and the Politics of Neutral Concern. Midwest Studies in Philosophy, 7(1).
• Briggs, R., Friedhoff, S., & Lundberg, E. (2023). Saving Lives with AI. Management and Business Review, 3(1&2).
• Bublitz, C. (2020). AI Systems and Respect for Human Autonomy. Front. Artif. Intell 4:705164.
• Calo, R. (2017). Artificial intelligence policy. UC Davis Law Review, 51.
• Cath, C., Wachter, S., Mittelstadt, B., Taddeo, M., & Floridi, L. (2018). AI and the ‘Good Society’. Science and Engineering Ethics, 24(2).
• Cialdini, R. B. (1984). Influence: The Psychology of Persuasion. Harper & Row.
• Crawford, K. (2021). Atlas of AI. Yale University Press.
• Crevier, D. (1993). AI: The Tumultuous History of the Search for Artificial Intelligence. Basic Books.
• Darmody, A., & Zwick, D. (2020). Manipulate to empower. Big Data & Society, 7(1).
• De Bruyn, A. et al. (2020). Artificial intelligence and marketing. Journal of Interactive Marketing, 51.
• Dworkin, G. (1988). The Theory and Practice of Autonomy. Cambridge University Press.
• Endres, K., & Kelly, K. J. (2018). Does microtargeting matter? Journal of Elections, Public Opinion and Parties, 28(1).
• Faraoni, S. (2023). Persuasive Technology and Computational Manipulation. Front. Artif. Intell 6:1216340.
• Floridi, L. (2020). AI and Its New Winter. Philosophy & Technology, 33(1).
• Galeon, D., & Reedy, C. (2017). Kurzweil Claims That the Singularity Will Happen by 2045. Futurism.
• Garrett, M. (2018, March 18). Trump campaign phased out use of Cambridge Analytica data. CBS News.
• Gartner Hype Cycle Research Methodology. (n.d.). Gartner.
• Gasser, U. et al. (2020). Digital tools against COVID-19. The Lancet Digital Health, 2(8).
• Graham, M. (2022). Average CMO Tenure Holds Steady at Lowest Level in Decade. WSJ.
• Hajkowicz, S. et al. (2023). AI adoption in the sciences. Technology in Society, 74.
• Heaven, W. D. (2023). How existential risk became the biggest meme in AI. MIT Technology Review.
• Hersh, E. D. (2015). Hacking the electorate. Cambridge University Press.
• Hoff, R. (2020). The public perception of AI. JCOM, 19(07).
• Hsu, T. (2019, October 28). The Advertising Industry Has a Problem. NYT.
• Kahneman, D. (2011). Thinking, Fast and Slow. FSG.
• Kalla, J. L., & Broockman, D. E. (2018). The Minimal Persuasive Effects of Campaign Contact in General Elections. APSR, 112(1).
• King, S., & Prasetyo, J. (2023). Assessing generative AI through the Gartner Hype Cycle. Frontiers in Education, 8.
• Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108(3).
• Matz, S. C. et al. (2017). Psychological targeting as an effective approach to digital mass persuasion. PNAS, 114(48).
• Mellet, K., & Beauvisage, T. (2020). Cookie monsters. Consumption Markets & Culture, 23(2).
• Merchant, B. (2023). The ‘AI Apocalypse’ Is Just PR. The Atlantic.
• Meyer, R. (2018). The Attention Economy is a Malthusian trap. The Atlantic.
• Montgomery, D. (2021, April 26). How to Sell the Coronavirus Vaccines. The Washington Post.
• Nadarzynski, T. et al. (2019). Acceptability of AI-led chatbot services in healthcare. Digital Health, 5.
• Neff, J. (2023). Marketing’s Image Problem Worsens. Ad Age.
• OpenAI et al. (2024). GPT-4 Technical Report (arXiv:2303.08774).
• Pariser, E. (2011). The Filter Bubble. Penguin.
• Roose, K. (2023). AI Poses ‘Risk of Extinction’. NYT.
• Rosenberg, M., Confessore, N., & Cadwalladr, C. (2018). How Trump Consultants Exploited the Facebook Data of Millions. NYT.
• Thaler, R. H., & Sunstein, C. R. (2008). Nudge. Yale University Press.
• U.S. Office of the Director of National Intelligence. (2022). Report on Commercially Available Information.
• Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380).
• West, S. M., Whittaker, M., & Crawford, K. (2019). Discriminating systems. AI Now Institute.
• Wolff, R. P. (1970). The Autonomy of Reason. Harper & Row.
• Wong, M. (2023). AI Doomerism Is a Decoy. The Atlantic.
• Yeung, K. (2017). ‘Hypernudge’. Information, Communication & Society, 20(1).
• Yeung, K., Howes, A., & Pogrebna, G. (2020). AI Governance by Human Rights–Centered Design. In Dubber et al. (eds), The Oxford Handbook of Ethics of AI.
• Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs.