Week 9: Advertising and AI, Current Trends

Week 9 looks at how artificial intelligence (AI) and the mobile, always-connected media environment are changing advertising. The chapter starts with a familiar problem in a new form: digital advertising clutter and the many ways consumers avoid ads. It then shows how online and offline behaviour are now mixed in hybrid pathways to persuasion, how AI makes personalization possible (and why it can backfire through privacy concerns and dataveillance), and how computational advertising has developed from simple banner buying into data-driven, algorithm-based systems. The second half covers generative AI in advertising (AI-generated copy, images and models, disclosure, algorithm aversion and authenticity), advertising in the metaverse, AI in scheduling and buying, and the impact of AI on agencies, jobs and strategy. The chapter ends with a short overview of current trends in the Belch and Belch tradition and a mini case.

Digital advertising clutter in an age of mobile media and AI

Clutter is not new. Earlier chapters discussed clutter in television and radio breaks, and zipping and zapping as ways viewers escape commercials. In digital media the problem has grown and changed shape. Every screen, feed, app and search result page can carry advertising, and AI systems can produce and place ads at a scale and speed that no human team could match. The result is a media environment in which consumers are exposed to more commercial messages than they can possibly process.

Digital clutter differs from traditional clutter in three ways:

• Mobile screens are small and personal. An ad on a smartphone competes for a very limited space and often interrupts a goal-directed activity such as reading, messaging or searching. Ads on a personal device can therefore feel more intrusive than ads on a shared television screen.
• Ads blend with content. Native ads, sponsored posts and influencer content look like editorial or social content, which makes it harder for consumers to separate commercial from non-commercial messages.
• AI increases volume and variation. Generative tools make it cheap to create many versions of an ad, and automated buying systems can place them everywhere the target audience goes. More versions and more placements can raise perceived clutter even when each single ad is relevant.

Ad avoidance: cognitive, affective and behavioral

Clutter is an important driver of ad avoidance. Advertising research usually distinguishes three components of avoidance, which match the classic think, feel and do structure of attitudes.

Typical antecedents of avoidance in digital media are perceived goal impediment (the ad gets in the way of what the user wants to do), perceived clutter, prior negative experiences with ads, and concerns about privacy and intrusiveness. These antecedents matter for managers because they point to remedies: ads that do not block the user's goal, that respect frequency limits and that feel relevant rather than creepy are less likely to be avoided.

Ad blocking creates a dilemma for the industry. Publishers lose income, which threatens free content; advertisers lose reach among often younger and more digitally skilled users. Responses include asking users to whitelist a site, offering paid ad-free versions, moving budget to formats that are harder to block (sponsored content, influencer marketing, in-app ads) and, more constructively, improving ad quality so that fewer users feel the need to block. AI plays a double role here: it can increase clutter by multiplying ads, but it can also reduce perceived clutter by improving relevance, capping frequency and choosing better moments to show an ad.

Hybrid pathways to persuasion: integrating online and offline

Consumers no longer move through a purely online or purely offline journey. They see a social media ad, check reviews on their phone in a store, visit the store, and buy later through an app. Persuasion therefore follows hybrid pathways in which online and offline advertising and online and offline behaviour influence each other. This builds on the omnichannel and customer journey thinking from Weeks 1 and 3.

• Webrooming: researching a product online and then buying it in a physical store, for example because the consumer wants to touch the product, avoid shipping costs or take it home immediately.
• Showrooming: examining a product in a physical store and then buying it elsewhere, often online. This pattern was introduced in the customer experience lecture.
• ROPO (research online, purchase offline): the broader measurement idea that online advertising and search create sales that are completed offline and therefore do not show up in online conversion data.
• Cross-media synergy: the combined effect of several media is larger than the sum of their separate effects, for example when a television ad makes a later online banner more meaningful, or when an outdoor poster prompts a search.

The managerial implications are clear. Campaign objectives and measurement should cover the whole journey, not one channel. Offline touchpoints (stores, packaging, events, outdoor) and online touchpoints (search, social, apps) should carry a consistent message, which is the core idea of IMC. Measurement must link channels, for example through store visit data, loyalty cards, promo codes or matched market tests, because the last online click rarely tells the full story.

Personalization with AI

AI makes it possible to adapt advertising to individual consumers on the basis of their data: browsing history, location, purchases, app use and inferred interests. Personalization promises higher relevance and therefore less avoidance, but it also raises privacy concerns.

Personalized versus tailored advertising

The literature distinguishes between tailored and personalized advertising, and the reading list includes Tripathi et al. (2025), who study consumer receptivity to tailored and personalized digital advertising.

The personalization-privacy paradox

Consumers say they value privacy, yet they often share data and respond to relevant ads. They want the benefits of personalization (relevance, convenience, discounts) but dislike the data collection that makes it possible. This tension is called the personalization-privacy paradox. In practice, the outcome depends on how the consumer weighs perceived benefits against perceived risks in a specific situation.

Two concepts explain when personalization backfires:

• Perceived intrusiveness: the feeling that an ad interferes with the consumer's own thoughts or activities, or reveals that the advertiser knows too much. An ad that refers to a sensitive topic or a recent private search can feel "creepy".
• Personalization reactance: based on psychological reactance theory, consumers who feel that their freedom to choose is threatened by an overly personal or manipulative ad react against it, for example by rejecting the message or the brand.

Factors that reduce these risks include transparency about why an ad is shown, consumer control over data (opt-in, preference settings), trust in the brand, and the degree of personalization. Moderate, explainable personalization tends to be safer than highly specific personalization that the consumer did not expect.

Dataveillance

Strycharz, Zhang and Segijn (2025) discuss dataveillance in interactive advertising. As the title of their article indicates, they examine how continuous data monitoring transforms the interactions between consumers and advertisers and what unintended consequences this has.

Dataveillance changes the relationship between consumer and advertiser. The advertiser can observe consumers continuously, while consumers often do not know what is collected. Possible unintended consequences discussed in this literature include feelings of being watched, chilling effects (consumers change their behaviour because they feel monitored), lower trust and more avoidance. For advertisers this means that the data that make personalization possible can also undermine the relationship that advertising is meant to build. Regulation such as the European GDPR and the move away from third-party cookies push the industry towards first-party data and consent-based personalization.

Computational advertising: past, present and future

Past

Early online advertising looked like print: publishers sold banner space on a fixed price basis (cost per thousand impressions), often directly to advertisers. Search advertising introduced a key computational idea: ads linked to keywords, sold by auction and ranked by relevance and bid. Cookies made it possible to follow users across sites and retarget them.

Present

• Programmatic advertising: the automated buying and selling of ad space through software platforms, often in real time. Advertisers use demand-side platforms, publishers use supply-side platforms, and exchanges match them in auctions that take milliseconds.
• Predictive targeting: machine learning models predict which users are most likely to click, convert or respond, and target them, including lookalike audiences that resemble existing customers.
• Dynamic creative optimization (DCO): ads are assembled automatically from components (headline, image, offer, call to action) and the combination is adapted to the viewer and context; the system learns which combinations work best.
• Recommender systems: algorithms that suggest products or content on the basis of past behaviour and the behaviour of similar users. On retail and streaming platforms recommendations and advertising increasingly merge, for example in retail media.

Future and research directions

The next phase combines computational advertising with generative AI: systems that do not only select and place existing ads but also create new content for each context. Important research directions include the effects of algorithmic decisions on consumers (fairness, bias, manipulation), transparency and explainability of targeting, privacy-preserving methods that work without third-party cookies, measurement in a fragmented and partly privacy-protected environment, and the changing role of human creativity and judgment. This links back to the Week 3 view of digital advertising as algorithmic persuasion.

AI in advertising scheduling and buying

Week 6 introduced scheduling decisions (continuity, flighting, pulsing). AI turns many of these decisions into continuous, automated processes:

• Automated bidding: the platform sets the bid for each auction to reach a goal set by the advertiser, such as maximum conversions or a target cost per acquisition.
• Budget pacing: algorithms spread the budget over the day or the campaign period so that money is not spent too early or left unspent, and shift spending to moments with better expected results.
• Dayparting and frequency capping: systems learn at which hours and days ads work best and limit how often one user sees the same ad, which also reduces perceived clutter.
• Budget allocation across channels: models move budget between platforms and campaigns on the basis of predicted performance.

The manager's role moves from setting individual bids to setting objectives, constraints and data inputs, and to checking whether the algorithm optimizes for the right outcome. An algorithm that optimizes for cheap clicks can, for example, hurt brand building if no one checks the quality of the placements.

Generative AI in advertising

Baek, Pantano, Plangger and Kim (2026) discuss the evolution, challenges and opportunities of AI-generated advertising. Applications in advertising practice include:

• Copy: headlines, product descriptions, social posts and many variants for testing.
• Images and video: product visuals, backgrounds, storyboards and complete spots, produced faster and at lower cost than traditional shoots.
• AI-generated models: synthetic people who appear in ads instead of real models, and fully virtual influencers.
• Ideation and research: brainstorming, summarizing consumer insights, creating draft personas.

Human-AI co-creation

Most current practice is not full automation but human-AI co-creation: people set the strategy, brief and judge the output, while AI generates options and speeds up production. Yang, Dong, Chu and Rheu (2026) explore how advertising professionals perceive this human-AI value co-creation. Liao (2025/2026) stresses the importance of human interactivity in AI use in advertising and develops a new scale to measure it. The common theme is that the value of AI in advertising depends on how humans and AI work together, not only on what the technology can do.

AI disclosure, advertising value and authenticity

Should brands tell consumers that an ad was made with AI? Disclosure supports transparency and may be required by platform rules or regulation, but it may also change how consumers judge the ad. Bui (2025/2026) examines how disclosing AI involvement in advertising affects perceived advertising value and purchase intentions. Advertising value is the consumer's overall judgement of the worth of an ad, classically built from informativeness, entertainment, irritation and credibility; Ye et al. (2025) provide a meta-analysis of the antecedents and consequences of advertising value.

Two related ideas help explain consumer reactions:

• Algorithm aversion: the tendency of people to prefer human judgement or human work over algorithmic output, even when the algorithm performs equally well. Feng (2025) charts AI evolution and aversion in advertising.
• Authenticity: consumers value ads, brands and people that seem genuine and "real". AI-generated content can be seen as less authentic, especially when it shows people, emotions or experiences that never happened.

Khodakarami, Peter and Cornelis (2026) explore the effects of AI-generated versus real models in advertising, focusing on female consumers. This connects to well-known questions about idealized beauty in advertising, social comparison and representation, which Campbell et al. (2025) also address in their work on diversity representation in advertising. For managers, the practical issues are honesty (not presenting synthetic people as real customers), representation (not creating an even narrower beauty ideal) and fit (AI imagery may suit a playful tech brand better than a brand built on craftsmanship).

Defining ads in the metaverse

The metaverse refers to persistent, shared, immersive virtual worlds in which users interact through avatars. Examples include gaming platforms with large virtual worlds, social virtual reality spaces and branded experiences in augmented reality. Advertising in such spaces does not look like a banner or a spot, which raises the question of how to define an ad there.

• Virtual worlds and branded virtual spaces: brands build their own stores, games or event venues inside a platform, where users can explore, play and socialize.
• Avatars: users' digital representations. Brands sell or give away virtual clothing and accessories (direct-to-avatar), so the brand becomes part of the user's self-presentation.
• Virtual influencers: computer-generated characters with social media profiles who promote brands. They give brands full control, but raise questions about disclosure and authenticity similar to AI-generated models.
• In-world placements: virtual billboards, sponsored items or product placements within the environment, comparable to advertising in video games (Week 10).

Defining ads in the metaverse is difficult because the borders between content, game, experience and advertisement disappear. This creates problems for recognition (users, especially young ones, may not see that something is commercial), disclosure, measurement (what counts as an exposure in a 3D world?) and regulation.

The impact of AI on advertising agencies, jobs and strategy

AI affects the whole advertising industry, not only individual campaigns.

• Agencies: production tasks that used to take weeks and many people (versions, resizing, translations, simple visuals) can now be done in hours. This puts pressure on business models built on billable hours and pushes agencies towards strategy, creative ideas, data and technology consulting. Some clients bring AI-based production in-house.
• Jobs: routine tasks are automated, while new roles appear (prompt and AI workflow specialists, data scientists, AI governance). Human skills such as strategic thinking, cultural insight, taste, ethical judgement and client relationships become more important, not less.
• Strategy: when everyone can produce content cheaply, content itself is no longer a source of advantage. Brand distinctiveness, first-party data, a clear positioning and trust become the main differentiators. Strategy also has to cover new questions such as AI disclosure policies, use of synthetic models and visibility in AI-driven search and assistant answers.
• Risks: errors and "hallucinations" in AI output, copyright and intellectual property questions, bias in data and output, brand safety and the risk that all brands start to look alike.

Current trends in advertising and IMC

Belch and Belch present advertising and IMC as a field that keeps changing because of technology, media fragmentation and changing consumers. In that tradition, the main current trends can be summarized as follows:

1. From mass media to addressable media: more advertising is bought per user or household, including connected TV and digital outdoor.
2. Growth of digital, social and mobile: budgets keep moving to search, social, video and retail media, with mobile as the main screen.
3. Consumer control: consumers decide what they watch and when, use ad blockers and paid ad-free options, and create content themselves.
4. Data, accountability and measurement: clients demand proof of return on investment, while privacy rules make tracking harder.
5. Influencers and content: brands work with creators and produce content that people choose to watch (see Week 10).
6. AI throughout the process: from insight and creative to buying, scheduling and measurement.
7. Ethics, regulation and responsibility: privacy, transparency, representation and sustainability claims are under growing scrutiny.

Mini case: Lumora Skin (fictional)

Lumora Skin is an invented mid-priced skincare brand that sells online and through drugstores. Its marketing team wants to use AI to scale a new campaign for a moisturizer. The plan has four parts: (1) generative AI creates hundreds of ad variants with AI-generated models of different ages and skin types; (2) DCO assembles the variants for each viewer; (3) personalized ads refer to the user's recent product searches and local weather ("dry air in your city today"); and (4) programmatic buying with automated bidding targets users predicted to be likely buyers.

After two weeks, online click-through rates are good, but social media comments call the models "fake" and some users complain that the ads "know too much". Drugstore partners report higher sales that the online dashboard does not show. A review of the campaign shows: perceived intrusiveness and personalization reactance caused by the weather and search references; authenticity concerns about AI-generated models in a category where real skin results are central; and a ROPO effect that the measurement missed. Lumora's response: replace AI models with real customers in hero ads while keeping AI for background visuals and variants, add a clear AI disclosure where AI is used, lower the personalization level and give users a "why am I seeing this ad" explanation, set frequency caps to reduce clutter, and add store sales data to the measurement.