Machine marketing: rethinking the customer in the age of generative AI

Probably not a substantial claim: may not pass S3 (load-bearing).

LLMs generate outputs by predicting likely token sequences based on patterns learned from large-scale training data.

“At the core of these systems are large language models (LLMs), which generate outputs by predicting likely token sequences based on patterns learned from large-scale training data (Brown et al. 2020;Radford et al. 2018;Vaswani et al. 2017).”

• cites Brown et al. 2020 sourceSupported
• cites Radford et al. 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Vaswani et al. 2017 sourcePartly supported

  Section 3 (p. 2) describes the Transformer decoder generating symbols one at a time from previously generated symbols, and Section 3.4 (p. 5) describes predicted next-token probabilities. Section 5.1 (p. 7) reports training on millions of sentence pairs. The paper establishes these points for its Transformer sequence-transduction models, but does not make the broader claim about LLMs as a class. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Recent literature in marketing research has largely focused on how consumers interact with AI systems and how these AI systems change customer journeys and referrals.

“Recent literature in marketing research has largely focused on how consumers interact with AI systems and how these AI systems change customer journeys and referrals (e.g., Aggarwal et al. 2024;Blut et al. 2024;Gelbrich et al. 2025;Liu et al. 2024).”

• cites Aggarwal et al. 2024 sourcePartly supported

  The section “A Conversational Generative Engine” describes users interacting with an AI-powered search system through queries and follow-ups. But the paper focuses on improving website visibility in generative-engine responses; it does not establish that marketing research has largely focused on consumer interactions, or examine changes to customer journeys and referrals. The indexed passages provide no page numbers. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites Blut et al. 2024 sourceNot checked — no full text of the source
• cites Gelbrich et al. 2025 sourcePartly supported

  The Data Collection section describes a review of marketing and related research on how customers respond to robots, chatbots, algorithms, and other automated agents compared with human agents. That supports the consumer-interaction part of the claim, but the passages do not establish that marketing research has largely focused on this topic or address changes to customer journeys and referrals; searches for those terms found no relevant passage. Page numbers were not available in our copy. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

Recent advances in GenAI have substantially expanded the capabilities of AI systems, enabling increasingly autonomous forms of human-AI interaction.

“Recent advances in GenAI have substantially expanded the capabilities of AI systems, enabling increasingly autonomous forms of human-AI interaction (Mienye et al. 2025).”

• cites Mienye et al. 2025 sourcePartly supported

  Section 2 (Related works; page unavailable) describes expanded LLM capabilities, including content generation and automated customer service. Section 8.3 (Ethical risk tiers; page unavailable) also discusses models used in autonomous systems. Neither passage establishes the claimed progression toward increasingly autonomous human–AI interaction. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

GEO aims to increase the likelihood that a firm's products, services, or content are retrieved, cited, and incorporated into AI-generated responses.

“Analogous to search engine optimization (SEO), GEO aims to increase the likelihood that a firm's products, services, or content are retrieved, cited, and incorporated into AI-generated responses.”

• cites Aggarwal et al. 2024 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Researchers have discussed deliberately implementing human-like cognitive biases in AI systems to more closely mirror human decision-making.

“Researchers have even discussed deliberately implementing human-like cognitive biases in AI systems to more closely mirror human decision-making (Besta et al. 2025;Hagendorff and Fabi 2023).”

• cites Besta et al. 2025 sourcePartly supported

  The Introduction (p. 1) says the authors deliberately prime AI agents with MBTI personalities and find behavioral biases aligned with those profiles. Section 4.3 (p. 7) reports differences in the agents’ strategic decisions that the authors relate to psychological traits. The paper does not describe deliberately implementing human cognitive biases specifically to mirror human decision-making; its intervention is personality priming. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Hagendorff and Fabi 2023 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Assessing the overlap between silicon and human samples is a common practice in current silicon sampling research.

“instead of assessing the overlap between silicon and human samples, which is a common practice in current silicon sampling research (Gao et al. 2025;Imschloss et al. 2025;Loru et al. 2025;Sarstedt et al. 2024)”

• cites Gao et al. 2025 sourceSupported
• cites Imschloss et al. 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Loru et al. 2025 sourceSupported

May combine two claims.

While LLMs sometimes mimic parameters of human information processing and behavior, this is not always the case.

“However, while LLMs sometimes mimic parameters of human information processing and behavior, this is not always the case (e.g., Kuric et al. 2026;Sarstedt et al. 2024).”

• cites Kuric et al. 2026 sourceSupported
• cites Sarstedt et al. 2024 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Tool-use capabilities enable LLMs to interact with external software systems, databases, search engines, and web services through APIs.

“In addition, tool-use capabilities enable LLMs to interact with external software systems, databases, search engines, and web services through APIs (Wang et al. 2024).”

• cites Wang et al. 2024 sourcePartly supported

  The Action goal section describes LLM-based agents using external APIs, including RESTful web services and software tools. It also describes querying databases with SQL and extracting content from web pages, but does not establish that database or search-engine access occurs through APIs. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

AI systems are shifting from passive tools to active market participants, evolving into AI customers that navigate offers independently, interact with firms directly, and shape purchase decision processes.

“These developments signal a shift from AI as a passive tool to an active market participant, ultimately evolving into AI customers that navigate offers independently, interact with firms directly, and shape purchase decision processes.”

Probably not a substantial claim: may not pass S4 (based).

AI systems now transcend simple information retrieval and delivery to support users across the entire consumer journey.

“AI systems now transcend simple information retrieval and delivery to support users across the entire consumer journey.”

Probably not a substantial claim: may not pass S4 (based).

Generative artificial intelligence is transforming consumer decision-making by enabling large language models to function as collaborative AI assistants or autonomous AI agents.

“Generative artificial intelligence (GenAI) is transforming consumer decision-making by enabling large language models to function as collaborative AI assistants or autonomous AI agents.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

As consumers increasingly delegate decisions to AI assistants and AI agents, these systems become AI customers that marketers must recognize as part of their target audience and that marketing scholars must incorporate into future market and consumer research.

“As consumers increasingly delegate decisions to AI assistants and AI agents, these systems become AI customers that marketers must recognize as part of their target audience and that marketing scholars must incorporate into future market and consumer research.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

AI agents can independently perform delegated tasks and execute actions within predefined constraints.

“While AI assistants act as active collaborators in consumers' decision processes, AI agents can independently perform delegated tasks and execute actions within predefined constraints.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

The growing role of AI customers necessitates machine marketing as a new subdiscipline dedicated to understanding, predicting, and shaping the behavior of AI-based decision-makers.

“This research note examines the implications of AI customers for marketing practice and scholarship, arguing that their growing role necessitates the emergence of machine marketing as a new subdiscipline dedicated to understanding, predicting, and shaping the behavior of AI-based decision-makers.”

Probably not a substantial claim: may not pass S4 (based).

AI assistants act as active collaborators in consumers' decision processes.

“While AI assistants act as active collaborators in consumers' decision processes, AI agents can independently perform delegated tasks and execute actions within predefined constraints.”

Probably not a substantial claim: may not pass S4 (based).

For both AI assistants and AI agents, consumers delegate part of their decision authority to an AI system.

“For both AI assistants and AI agents, consumers delegate part of their decision authority to an AI system, such as selecting an information source (e.g., brand-owned websites, comparison websites, product tests), choosing a retailer, or completing the checkout process.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

As AI customers operate through AI-mediated evaluation and delegated decision-making, they interpret product information, compare alternatives, infer consumer preferences, and potentially select products on consumers' behalf.

“As AI customers operate through AI-mediated evaluation and delegated decision-making, they interpret product information, compare alternatives, infer consumer preferences, and potentially select products on consumers' behalf.”

Probably not a substantial claim: may not pass S4 (based).

As AI systems increasingly evaluate, recommend, select, and transact on consumers' behalf, they evolve from decision-support technologies into AI customers that actively participate in marketplace exchanges.

“As AI systems increasingly evaluate, recommend, select, and transact on consumers' behalf, they evolve from decision-support technologies into AI customers that actively participate in marketplace exchanges (Fig. 1).”

Probably not a substantial claim: may not pass S4 (based).

AI customers act as intermediaries between the human consumer and the firm throughout the customer journey, from initial information search and preference building over check-out to post-purchase interactions such as refund negotiations.

“AI customers act as intermediaries between the human consumer and the firm throughout the customer journey, from initial information search and preference building over check-out to post-purchase interactions such as refund negotiations.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

AI customers blur the boundaries between tools and collaborators: they act (semi-)autonomously within bounded parameters but typically remain contingent on human intention and oversight.

“They blur the boundaries between tools and collaborators: they act (semi-)autonomously within bounded parameters but typically remain contingent on human intention and oversight.”

May combine two claims.

AI agents autonomously pursue user-defined goals, decompose complex objectives into subtasks, gather and evaluate information from multiple sources, iteratively refine decisions, and execute actions with minimal or no human intervention.

“AI agents extend these capabilities by autonomously pursuing user-defined goals.They can decompose complex objectives into subtasks, gather and evaluate information from multiple sources, iteratively refine decisions, and execute actions with minimal or no human intervention.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Purchases involving AI customers require firms to close sales and to initiate and manage delivery and payment processes in collaboration with AI interfaces.

“Purchases involving AI customers require firms to close sales, as well as initiate and manage delivery and payment processes in collaboration with AI interfaces.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Before the final purchase, AI customers will likely play a role in price adjustments.

“Before the final purchase, AI customers will likely play a role in price adjustments.”

Probably not a substantial claim: may not pass S4 (based).

The shift from SEO to GEO involves consumers moving from generic search terms to more nuanced prompts.

“the shift from SEO to GEO comes with changes in consumers' search strategy in that they move from generic search terms as in the search engine case (e.g.,”

May combine two claims.

Identifying preference structures may be relatively straightforward for search goods but becomes more challenging for experience and credence goods, where consumers rely on external cues such as sensory experiences to evaluate quality.

“While identifying preference structures may be relatively straightforward for search goods (e.g., selecting a mobile phone contract with the best value for money), it becomes more challenging for experience and credence goods, where consumers rely on external cues such as sensory experiences to evaluate quality (e.g., Weathers et al. 2007).”

• cites Weathers et al. 2007 sourceNot checked — no full text of the source

May combine two claims.

A recent industry survey indicates that 85% of respondents are open to integrating AI systems into their purchase journey, with 32% open to completely delegating decision-making to AI systems.

“A recent industry survey by Oteh et al. (2026b) indicates that 85% of respondents are open to integrating AI systems into their purchase journey, with 32% of respondents being open to completely delegating decision-making to AI systems (i.e., let AI choose what to buy).”

• cites Oteh et al. 2026bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Machine marketing will require understanding how AI customers process prompts, represent preferences, and make decisions.

“Just as SEO required understanding how search engines interpreted and ranked content, machine marketing will require understanding how AI customers process prompts, represent preferences, and make decisions.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Machine marketing considers AI customers as a separate target group that notably differs from human customers and thus requires uniquely focused marketing campaigns and tailored marketing strategies.

“That is, machine marketing considers AI customers as a separate target group that notably differs from human customers and thus requires uniquely focused marketing campaigns and tailored marketing strategies.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

As consumers and AI systems increasingly interact, AI customers will become more sensitive to human behavior, just as human consumers may adapt their behaviors considering AI-based recommendations and purchase decisions.

“As consumers and AI systems will increasingly interact, AI customers will become more sensitive to human behavior, just as human consumers may adapt their behaviors considering AI-based recommendations and purchase decisions.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Silicon samples may offer insights into understanding AI customers by mimicking their behavior, even if they do not mimic human behavior perfectly in all its facets.

“Silicon samples may offer insights into this regard by mimicking the behavior of AI customers, even if they do not mimic human behavior perfectly in all its facets.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

This development will reshape marketing strategies, metrics, and ethics in the digital marketplace.

“This development will reshape marketing strategies, metrics, and ethics in the digital marketplace.”

May combine two claims.

AI assistants support consumers throughout the decision process but require continuous human interaction, guidance, and confirmation.

“AI assistants support consumers throughout the decision process but require continuous human interaction, guidance, and confirmation.”

May combine two claims.

Although AI customers will often operate as decision-support systems, consumers may increasingly delegate decision tasks to them.

“Although AI customers will often operate as decision-support systems, consumers may increasingly delegate decision tasks to them.”

Probably not a substantial claim: may not pass S4 (based).

GEA is the paid counterpart to GEO, enabling firms to promote products or services through sponsored placements embedded in AI-generated responses or recommendations.

“By contrast, GEA is the paid counterpart to GEO, enabling firms to promote products or services through sponsored placements embedded in AIgenerated responses or recommendations.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

With increased decision authority, AI customers act with higher autonomy and can initiate actions or complete transactions within prespecified boundaries without human involvement.

“With increased decision authority, AI customers act with higher autonomy and can initiate actions or complete transactions within prespecified boundaries without human involvement.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Consumers may become dependent on AI-generated recommendations and delegate an expanding range of decisions.

“Conversely, consumers may become dependent on AI-generated recommendations and delegate an expanding range of decisions.”

Probably not a substantial claim: may not pass S4 (based).

As relationships deepen, AI customers may increasingly mirror aspects of their human counterpart's personality, values, and decision-making.

“As these relationships deepen, AI customers may increasingly mirror aspects of their human counterpart's personality, values, and decision-making.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Human customers would rely on their AI assistants and AI agents to search for favorable offers, negotiate lower prices, and exploit discounts.

“human customers would rely on their AI assistants and AI agents to search for favorable offers, negotiate lower prices, and exploit discounts”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Many consumers are hesitant to engage in sales failure recoveries and might therefore be willing to delegate related tasks, such as filing complaints or negotiating solutions, to AI agents.

“many consumers are hesitant to engage in sales failure recoveries and might therefore be willing to delegate related tasks, such as filing complaints or negotiating solutions, to AI agents.”

Probably not a substantial claim: may not pass S4 (based).

AI assistants and AI agents increasingly act as intermediaries between firms and consumers.

“Today, a similar shift is underway as AI assistants and AI agents increasingly act as intermediaries between firms and consumers.”

Probably not a substantial claim: may not pass S4 (based).

Making a purchase decision involving AI customers will be a dyadic process that may involve discussing choice options with an AI assistant.

“Making a purchase decision will then be a dyadic process that may involve discussing choice options with an AI assistant.”

Probably not a substantial claim: may not pass S2 (checkable).

Machine psychology would examine how AI customers perceive, decide, act, and interact in ways that resemble human behavior, rather than interpreting the internal representations and design components of the neural networks underlying LLMs.

“Researchers have proposed that such an endeavor necessitates the establishment of an academic area of machine psychology (Hagendorff, 2023), with important implications for the marketing discipline.”

• Also stated in: § Studying the behavior of AI customers · hedged

  “Rather than interpreting the internal representations and design components of the neural networks that underly LLMs, machine psychology would examine how AI customers perceive, decide, act, and interact in ways that resemble human behavior.”

• cites Hagendorff, 2023 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Through retrieval-augmented generation, LLMs can access and incorporate external information beyond their training data.

“Through retrieval-augmented generation (RAG), for example, LLMs can access and incorporate external information beyond their training data (Lewis et al. 2020).”

• cites Lewis et al. 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Firms would integrate AI customers into their dynamic pricing systems in an effort to capitalize on a consumer's willingness to pay.

“firms would integrate these AI customers into their dynamic pricing systems in an effort to capitalize on a consumer's willingness to pay.”

May combine two claims.

AI systems filter, recommend, and even decide on products and services, effectively becoming new gatekeepers of market access.

“These systems filter, recommend, and even decide on products and services, effectively becoming new gatekeepers of market access.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

AI customers' role extends beyond executing delegated tasks and fundamentally reshapes how consumers make decisions, the degree of autonomy they grant AI systems, and how human and AI preferences evolve through repeated interactions.

“Their role extends beyond executing delegated tasks and fundamentally reshapes how consumers make decisions, the degree of autonomy they grant AI systems, and how human and AI preferences evolve through repeated interactions.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

In the post-sales stage, machine marketing affects how consumers use products, engage in after-sales services, provide feedback, and ultimately dispose of products.

“In the post-sales stage, machine marketing affects how consumers use products, engage in after-sales services, provide feedback, and ultimately dispose of products.”

Probably not a substantial claim: may not pass S4 (based).

Across stages of the customer journey, AI customers must develop representations of their human counterpart's preferences, goals, and constraints in order to provide useful recommendations and make appropriate decisions.

“Across stages of the customer journey, AI customers must develop representations of their human counterpart's preferences, goals, and constraints in order to provide useful recommendations and make appropriate decisions.”

• AssumptionRests on an assumption: AI customers can develop representations of their human counterpart's preferences, goals, and constraints.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

The emergence of machine marketing closely parallels the rise of SEO and SEA several decades ago.

“The emergence of machine marketing closely parallels the rise of SEO and SEA several decades ago.”

May combine two claims.

Consumers' willingness to delegate choice tasks to AI systems depends on perceived autonomy and control; reduced perceived decision autonomy lowers the adoption likelihood of recommender systems.

“Specifically, reduced perceived decision autonomy lowers the adoption likelihood of recommender systems (Ahmad Husairi and Rossi 2024).”

• cites Ahmad Husairi and Rossi 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable).

AI customers are market actors that emerge through the integration of AI systems into consumer decision-making processes and span a continuum from supportive AI assistants to more autonomous AI agents.

“AI customers thus constitute market actors that emerge through the integration of AI systems into consumer decision-making processes and span a continuum from supportive AI assistants to more autonomous AI agents.”

May combine two claims.

Consumers may lack the capacity, motivation, or time to evaluate complex information or the vast number of available options and may therefore instruct an AI system to identify a specific product or recommend the best option within a given price range.

“For example, consumers may lack the capacity, motivation, or time to evaluate complex information or the vast number of available options (Chernev et al. 2015;Malhotra 1982;Shahid et al. 2025) and may therefore instruct an AI system to identify a specific product or recommend the best option within a given price range.”

• cites Chernev et al. 2015 sourceNot checked — no full text of the source
• cites Malhotra 1982 sourceNot checked — no full text of the source
• cites Shahid et al. 2025 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

AI customers will become an important target for companies to market their products and services.

“AI customers will become an important target for companies to market their products and services.”

May combine two claims.

The final purchase decision will be heavily influenced by AI-human collaboration, making machine marketing a counterpart of traditional marketing that targets human consumers.

“The final decision, therefore, will be heavily influenced by this AI-human-collaboration, making machine marketing a counterpart of traditional marketing that targets human consumers.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

The machine psychology perspective offers machine marketing frameworks that help anticipate how AI customers behave in complex environments, how consumers interpret and respond to them, and how human-AI relationships develop over time.

“The machine psychology perspective thereby offers machine marketing with frameworks that help anticipate how AI customers behave in complex environments, how consumers interpret and respond to them, and how human-AI relationships develop over time.”

Probably not a substantial claim: may not pass S2 (checkable).

This transition signals a broader move from marketing to humans toward marketing through and to machines.

“This transition signals a broader move from marketing to humans toward marketing through and to machines.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Machine marketing shifts attention from human-tailored value propositions to machine-readable offerings, from human-facing persuasion to AI-facing influence, and from direct consumer demand to demand that is filtered, translated, or reshaped by AI systems.

“Specifically, machine marketing shifts attention from human-tailored value propositions to machine-readable offerings, from human-facing persuasion to AI-facing influence, and from direct consumer demand to demand that is filtered, translated, or reshaped by AI systems.”