Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advertising in AI systems: Society must be vigilant

Menghua Wu, Yujia Bao

TL;DR

The paper analyzes how commercial incentives could shape generative AI outputs, arguing that unlike static ads, generative advertisements can dynamically influence each interaction via sponsorship-conditioned content. It proposes a framework treating AI systems as advertising platforms, with four design principles—Faithfulness, Utility, Privacy, and Provenance—and an implementation approach that uses tool-call patterns to integrate and audit sponsor content. It also outlines two debiasing strategies, direct debiasing and multi-sampling with aggregation, to recover ad-free outputs, and discusses open questions around regulation, user autonomy, transparency, and evaluation. The work aims to guide researchers and platforms toward transparent, user-centered management of commercial content in AI systems, balancing monetization with safety and trust by formalizing metrics and provenance for sponsorship effects.

Abstract

AI systems have increasingly become our gateways to the Internet. We argue that just as advertising has driven the monetization of web search and social media, so too will commercial incentives shape the content served by AI. Unlike traditional media, however, the outputs of these systems are dynamic, personalized, and lack clear provenance -- raising concerns for transparency and regulation. In this paper, we envision how commercial content could be delivered through generative AI-based systems. Based on the requirements of key stakeholders -- advertisers, consumers, and platforms -- we propose design principles for commercially-influenced AI systems. We then outline high-level strategies for end users to identify and mitigate commercial biases from model outputs. Finally, we conclude with open questions and a call to action towards these goals.

Advertising in AI systems: Society must be vigilant

TL;DR

The paper analyzes how commercial incentives could shape generative AI outputs, arguing that unlike static ads, generative advertisements can dynamically influence each interaction via sponsorship-conditioned content. It proposes a framework treating AI systems as advertising platforms, with four design principles—Faithfulness, Utility, Privacy, and Provenance—and an implementation approach that uses tool-call patterns to integrate and audit sponsor content. It also outlines two debiasing strategies, direct debiasing and multi-sampling with aggregation, to recover ad-free outputs, and discusses open questions around regulation, user autonomy, transparency, and evaluation. The work aims to guide researchers and platforms toward transparent, user-centered management of commercial content in AI systems, balancing monetization with safety and trust by formalizing metrics and provenance for sponsorship effects.

Abstract

AI systems have increasingly become our gateways to the Internet. We argue that just as advertising has driven the monetization of web search and social media, so too will commercial incentives shape the content served by AI. Unlike traditional media, however, the outputs of these systems are dynamic, personalized, and lack clear provenance -- raising concerns for transparency and regulation. In this paper, we envision how commercial content could be delivered through generative AI-based systems. Based on the requirements of key stakeholders -- advertisers, consumers, and platforms -- we propose design principles for commercially-influenced AI systems. We then outline high-level strategies for end users to identify and mitigate commercial biases from model outputs. Finally, we conclude with open questions and a call to action towards these goals.

Paper Structure

This paper contains 36 sections, 4 equations, 4 figures.

Table of Contents

  1. Introduction
  2. How might commercial incentives be incorporated into AI systems?
  3. Opportunities for generative advertising
  4. Product recommendation
  5. Product placement
  6. Paid visibility
  7. Advantages and challenges of generative advertising
  8. Specificity and suitability
  9. Long term impact on platform quality
  10. Ease of regulation
  11. How do we design an AI system that serves commercial content?
  12. Who are the stakeholders, and what are their requirements?
  13. Advertiser
  14. Consumer
  15. Platform
  16. ...and 21 more sections

Figures (4)

  • Figure 1: (A) Classic advertisements are independent of the platform's primary media. Sponsored content (green) does not impact ordinary content (blue), and they can be inserted anywhere. (B) Generative advertisements are AI-generated responses which have been conditioned on both the user query and sponsored information. They do affect the primary outputs of AI systems.
  • Figure 2: Flow of services in an AI system. A) Abstract overview. B) A coding assistant may serve as a platform that connects software vendors to software consumers (engineers).
  • Figure 3: Example of AI system with advertisements. Generative advertisements derive their content from both traditional information sources, as well as additional sponsored content. These sources may be accessed via standard tool-call patterns and synthesized by a language model into an overall response.
  • Figure 4: Common AI systems prominently display static advertisements for premium features (green). Screenshots were captured in May 2025.