The experimental setup: Wired tested multiple AI platforms including Perplexity AI, ChatGPT, Google’s Gemini, Anthropic’s Claude, and Amazon’s Rufus to assist with holiday shopping for five recipients of varying ages and interests.
- The shopping list included gifts for recipients ranging from 6 months to 49 years old, each with distinct interests like baking, music, and teenage fashion
- Perplexity AI’s new “Buy with Pro” subscription feature, priced at $20 monthly, was among the tools evaluated
- The test aimed to assess how well AI could handle real-world gift shopping scenarios
Performance analysis: Each AI assistant demonstrated distinct strengths and limitations when tasked with gift recommendations.
- Perplexity AI and Amazon’s Rufus frequently suggested items that were either impractically expensive or oddly misaligned with the stated preferences
- ChatGPT provided thoughtful gift ideas but lacked the ability to provide direct product links
- Claude maintained ethical standards by avoiding scraped review data but couldn’t offer direct shopping links
- Google’s Gemini generated generic, uninspiring gift suggestions that failed to demonstrate meaningful personalization
Practical outcomes: The AI-assisted shopping experiment yielded mixed results in terms of actual gift procurement.
- Some successful purchases were made based on ChatGPT’s recommendations, particularly for baking-related items
- The time spent testing various AI tools led to delayed shipping, with many gifts scheduled to arrive after Christmas
- Users still needed to manually refine search prompts and complete purchases themselves, as none of the AI tools could autonomously execute transactions
Future implications: This early-stage experiment reveals both the current limitations and future potential of AI-assisted shopping.
- Companies like Perplexity are likely gathering valuable data from these shopping interactions to improve their AI models
- The experience highlights the gap between current AI capabilities and the vision of fully automated shopping assistance
- While AI shopping assistants show promise, they currently serve better as brainstorming tools than complete shopping solutions
The experiment suggests that while AI shopping assistants will continue to evolve, they remain several iterations away from truly transforming the holiday shopping experience. Users seeking immediate shopping assistance may find traditional methods more efficient for now, though the rapid pace of AI development could soon change this dynamic.
I Used AI to Do All of My Holiday Shopping