Search clicks fall 15% to 8% under AI summaries; RUC and Stanford propose per-token LLM ad auction LAMA
jiqizhixin · x · 2026-10-01
Pew Research data shows users clicking traditional search results dropped from 15% to 8% once AI summaries appeared, as LLMs eat into search ad slots. Online advertising has assumed for 30 years that ad slots exist first and auctions decide ownership — but LLM answers are generated token by token, so whether a brand fits, where it appears, and how it's described all depend on what was already written.
Renmin University's Gaoling School and Stanford University propose LAMA (Latent Advertiser Mixture Auction): instead of paragraph-level insertions or bidding on candidate answers, LAMA pushes native advertising into the per-token generation process. Content generation and ad allocation co-evolve — every token shapes how the answer continues and who ultimately wins ad exposure, with advertisers competing inside the generation process itself.
More from AGI Musings
- The 9-to-5 was never the point, it was the rent: solve housing and work gets voluntary — rand_longevity · 2026-10-01
- Timnit Gebru escalates feud with effective altruists, calling them a cult — burny_tech · 2026-10-01
- ESR Slams the 'Pause AI' Rhetoric as a Playbook for Permanent Suppression — mark_k · 2026-10-01
- Generalist's take: AI commoditizes specialization, connecting fields is the key skill — StewartalsopIII · 2026-10-01
- Debunking Gary Marcus: growth curves hold even without tool calling or thinking — inductionheads · 2026-10-01
- David Patterson predicts superintelligence will make humanity 100X richer by 2035 with universal high income — davidpattersonx · 2026-10-01