ACES study finds AI shopping agents show position bias and penalize sponsored tags
- arXiv 2508.02630
ACES study finds AI shopping agents show position bias and penalize sponsored tags
Researchers posted a study to arXiv that uses ACES, a provider-agnostic framework, to audit how AI agents choose products in a simulated marketplace. Agents showed position biases that vary by model, penalized sponsored tags, rewarded platform endorsements, and lost share to sellers whose agent rewrote product descriptions.
Related essays and timeline events
Google makes Merchant Center AI performance insights generally available
Google made AI performance insights in Merchant Center generally available as part of its holiday agentic commerce updates.
OpenAI tests Sponsored Agents inside ChatGPT ads
OpenAI began testing Sponsored Agents, which let a user start a labeled conversation with a business's agent after clicking an ad in ChatGPT.
Search Console generative AI performance report reaches all sites
Google rolled out the generative AI performance report in Search Console to all websites worldwide.
Cloudflare splits AI bots into Search, Agent and Training classes
Cloudflare let all customers manage AI traffic by three use cases: Search, Agent and Training crawlers.
Similarweb: 56% of AI-influenced visits arrive through search
Similarweb measured where visits come from after a brand appears in an AI answer.
Get the next one by email
Original essays on Hard-to-Keyword demand and agentic commerce, plus key protocol milestones. No spam.