Token-Level Advertising
2026-08-27 • Computer Science and Game Theory
Computer Science and Game TheoryMachine Learning
AI summaryⓘ
The authors propose a new way to include ads in AI-generated text called LAMA. Instead of placing ads in fixed spots, LAMA mixes advertiser preferences directly into each word generated by the AI. They show that this approach is fair and efficient while still keeping the generated responses useful for users. Their tests with real search data suggest LAMA can improve both the platform's earnings and overall value. This work points toward smarter, generation-native advertising methods.
Generative AIAdvertising MechanismsToken-Level AdvertisingMarkov DSICIncentive CompatibilityKL-Regularized WelfareNext-Token PoliciesLatent MixturePlatform RevenueLearning-Based Implementation
Authors
Hanbing Liu, Bowei Zhang, Changyuan Yu, Yinyu Ye, Qi Qi
Abstract
Generative AI is transforming how people access information, challenging traditional advertising mechanisms built around predefined slots. Towards generation-native advertising, we propose the Latent Advertiser Mixture Auction (LAMA), a token-level advertising mechanism that embeds advertiser influence directly into the generation process. Advertisers report local continuation values that induce advertiser-specific next-token policies, from which the platform decodes through a latent mixture while updating an allocation posterior. We show that LAMA satisfies Markov DSIC and IR, and achieves near-optimal KL-regularized welfare. We further develop a learning-based implementation that reconstructs the required reports online from learned local advantages and root values. Proof-of-concept experiments on real-world commercial-search query splits show that LAMA improves platform welfare and revenue while maintaining user-facing response quality, providing initial evidence for the feasibility of generation-native advertising.