AI Search and Shopping: Will Consumers Really Let AI Agents Buy for Them?

Nearly 50% of shoppers now use an AI search engine such as ChatGPT, Google Gemini or Amazon Rufus at some point during the purchase journey. Whether it’s discovering products, comparing options or summarising reviews, AI search is quickly becoming part of how consumers research before buying.

But while AI is clearly influencing the discovery phase, will consumers really allow AI agents to take over the entire shopping process?

Payment processor Stripe recently shared a framework describing five stages of AI-assisted shopping. According to their model, we are currently around Level 2, where shoppers use AI tools for product research, comparisons and discovery.

The next step, Level 3, seems close. This would involve AI providing more personalised product recommendations based on a deeper understanding of the buyer’s preferences, purchase history and context. In many ways, we are already seeing early versions of this through recommendation engines and conversational AI.

However, Levels 4 and 5 represent a much bigger behavioural shift. At these stages, consumers would increasingly outsource the shopping process to AI agents. Rather than researching products themselves, the AI would identify options, make decisions and potentially complete the purchase on the user’s behalf.

While this vision is often discussed in tech circles, the reality may be further away — particularly for higher-involvement purchases.

One example is the relatively slow adoption of Amazon Rufus, Amazon’s AI shopping assistant. Despite being built directly into the shopping experience, many consumers still prefer browsing, comparing and researching products themselves.

There are several reasons why fully autonomous AI shopping may take longer to become mainstream.

1. People enjoy the shopping process

AI is at its most powerful when it automates repetitive or time-consuming tasks. However, shopping isn’t always something consumers want to automate.

For many people, comparing products, reading reviews and exploring options is part of the experience. It gives shoppers confidence in their decision and allows them to feel in control of the purchase journey.

If AI removes too much of this process, it risks removing something consumers actually value.

2. Shopping has emotional drivers

Shopping is often described as “retail therapy”, and there is a strong emotional element behind many purchases.

Discovering something new, finding a better deal or simply browsing products can create a small dopamine reward. These emotional triggers are part of why people enjoy shopping in the first place.

Because of this, consumers may not want to fully outsource the experience to an AI agent, particularly when purchases are personal or discretionary.

3. Not all buyers are decisive

AI systems are designed to optimise decisions based on data and probabilities. But not every shopper wants a perfectly rational recommendation.

Many buyers are naturally indecisive and prefer to explore multiple options before committing. They may revisit products several times, read different opinions and weigh up alternatives.

An AI personal shopper could feel too clinical for these types of decision-makers, who often value the exploration stage as much as the final outcome.

The future of AI shopping

AI will undoubtedly play an increasingly important role in commerce. The biggest impact in the near term is likely to be in research, discovery and product comparison, where AI can simplify large amounts of information.

However, completely handing over purchasing decisions to AI agents may require a bigger shift in consumer behaviour than many predictions assume.

For now, it seems more likely that AI will guide shoppers rather than replace them.