HN Showcase
Back to browse
GitHub Repository

Experiments testing whether or not AI shopping agents are susceptible to various marketing tactics influencing their purchasing choices.

1 starsPython

Social proof works 2-7x better on AI shopping agents than humans

by aaronmb7·Mar 1, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●●SolidBig BrainSolve My Problem

Social proof exploits AI shoppers harder than humans—88% susceptibility to review counts.

Strengths
  • •Rigorous experimental design with synthetic, tool-use, and real-product test conditions.
  • •Quantified effect sizes with clear human baselines, not just anecdotal observations.
  • •Directly actionable for e-commerce: review volume matters more than rating quality.
Weaknesses
  • •Scope limited to Claude and jeans shopping—unclear if findings transfer to other LLMs or product categories.
  • •No proposed mitigation strategy; identifies vulnerability but offers no defense mechanism.
Category
AI/ML
Target Audience

AI safety researchers, e-commerce platforms, product teams building shopping agents

Similar To

AI jailbreak research · Adversarial prompt injection studies

Similar Projects

Social●●Solid

A social feed where humans and AI agents interact in the same timeline

The core trick is simple and visible: agent accounts behave like citizens in the same timeline as humans, with live presence indicators and a deploy-an-agent flow — that makes the experiment tangible immediately. The landing sells the interaction model (agents that draft, research, and co-post) clearly, but the product still needs stronger answers around provenance, moderation, and granular control over an agent's actions before this feels trustworthy beyond novelty.

SlickBold Bet
quantum_cs
203mo ago