When you interact with an AI shopping assistant online, the experience feels fast, efficient, and seamless. But what if that very perception of speed is subtly changing how you think about time, making you less patient and less likely to choose the environmentally friendly option? A new series of experiments suggests that interacting with AI agents can distort consumers’ internal sense of time in ways that discourage sustainable purchasing decisions.
The research, published in the Journal of Retailing and Consumer Services, involved 1,462 Chinese consumers across four controlled experiments. The team found that when people interacted with an AI assistant rather than a human one, they became more physiologically activated and felt less connected to their future selves. This combination made them less willing to wait for greener options like eco-friendly shipping or carbon offsets. However, the story is more nuanced than that: the same time-warping effect could actually promote sustainability when the delay itself was framed as an ongoing benefit.
What prompted the investigation
The research was led by Han Wang of Xishuangbanna Vocational and Technical College in China, alongside colleagues including Songyu Jiang of Rajamangala University of Technology Rattanakosin in Thailand and Gomaa Agag of Nottingham Business School in the United Kingdom.
The team started from a simple observation: AI systems are widely perceived as fast, efficient, and analytical. Previous research had already shown that when people know they are dealing with an AI rather than a human, they associate the interaction with rapid processing. This association can speed up something psychologists call the “internal clock,” a mental timing mechanism that governs how long a given period of time feels. When this internal clock runs faster, waiting periods feel longer than they actually are, and people become more impatient.
The concept of an “internal clock” comes from a well-established psychological theory. The idea is that your brain has a kind of mental pacemaker. When you are excited or stimulated, this pacemaker speeds up, producing more “ticks” in a given period. As a result, a five-minute wait can feel like ten minutes. When you are calm and relaxed, the opposite happens, and time seems to fly.
The researchers wanted to know whether this AI-induced time distortion could spill over into real purchasing decisions, specifically ones involving sustainability. Many green options ask consumers to accept a delay or trade-off: slower shipping that produces fewer carbon emissions, for example, or a carbon-offset certificate that arrives weeks after a purchase. If AI interactions make delays feel longer and more costly, they could quietly push consumers away from these environmentally responsible choices.
Two psychological pathways
The team proposed that AI interactions affect sustainable choices through two connected psychological pathways working in parallel.
The first pathway involves what psychologists call “state arousal,” essentially how activated or energized a person feels at a given moment. The researchers hypothesized that interacting with an AI agent would increase this activation because people associate AI with speed and efficiency. That heightened arousal would then accelerate the internal clock, making any waiting period feel longer and more burdensome.
The second pathway involves “future self-continuity,” which is the degree to which a person feels psychologically connected to who they will be in the future. If you feel strongly connected to your future self, you are more likely to make sacrifices now for long-term benefits, like choosing a more sustainable product. But when your internal clock speeds up and the future feels more distant, that connection weakens. You start to feel like your future self is almost a different person, and their concerns seem less relevant.
Together, these two processes create a chain: AI interaction increases arousal and reduces connection to one’s future self, which accelerates the internal clock, which makes delays feel longer, which makes consumers less patient, which makes them less likely to choose the greener option.
How the experiments worked
All four experiments were conducted between March and May 2025 using Credamo, a large Chinese online research panel. Participants were adults aged 18 and older who made at least one online purchase per month. Each person participated in only one of the four studies to prevent any learning effects. The experiments were set in a simulated music retail platform called “YinyueGo,” a context chosen because China’s music industry is heavily digitized and relies extensively on AI-driven recommendation systems.
In Study 1, 310 participants imagined purchasing limited-edition tour merchandise. They interacted with either an AI assistant (labeled with an algorithmic description and a robot icon) or a human music advisor (with a personal photo and greeting). After a standardized 20-second loading screen, both groups saw the same two delivery options for an eco-friendly T-shirt: eco-shipping that took four weeks but was free and low-carbon, or standard shipping that arrived in three days for an extra fee. Participants who interacted with the AI agent chose the eco-shipping option only 45.2% of the time, compared to 60.0% for those who interacted with a human agent. Statistical analysis confirmed that this gap was driven by higher arousal and lower future self-continuity in the AI group.
Study 2 added a twist by testing whether making the AI seem more human-like could reduce these negative effects. In a 2-by-2 design with 350 participants, the AI agent was presented in either a mechanical, impersonal style or a warm, friendly style with a name, first-person language, and empathic phrases. When the AI was presented in its cold, mechanical form, it significantly reduced the likelihood that participants would add a carbon offset to their concert ticket. But when the AI was designed with human-like warmth and personality, this negative effect essentially disappeared.
Study 3 examined another design feature: the speed at which the digital interface itself operated. With 376 participants in a 2-by-2 design, the researchers varied whether the platform’s visual and audio elements moved quickly or slowly, independently of whether the agent was AI or human. A fast-paced interface amplified the AI’s negative impact on sustainable choice (in this case, choosing an eco-vinyl pre-order), while a slow-paced interface neutralized it. Slow pacing appeared to calm participants’ arousal levels and restore their sense of connection to the future.
Study 4, with 402 participants, introduced a reversal. Instead of asking consumers to endure a delay as a cost (waiting for an eco-product to arrive), the scenario framed the delay as an ongoing benefit: a longer-term green streaming subscription plan that provided continuous environmental contributions over time. In this context, the AI-induced time distortion actually worked in favor of sustainability. Because the extended time period felt even longer and more substantial, participants interacting with the AI agent were more likely to choose the longer green plan. The same psychological mechanism that created impatience in the first three studies created patience when the passage of time was framed as a reward rather than a cost.
What this means for businesses
The findings carry practical implications for companies designing AI-powered shopping experiences, particularly those trying to promote sustainable products and services.
First, companies should be aware that the default perception of AI as fast and efficient can work against green initiatives that require consumers to accept any kind of waiting. Simply labeling an assistant as “AI” may be enough to shift consumer preferences away from slower, more sustainable options.
Second, design choices matter. Two specific modifications appeared effective at reducing this bias. Making AI agents warmer and more human-like through personalized language, friendly avatars, and empathic phrasing reduced the arousal spike and restored consumers’ connection to their future selves. Similarly, slowing the pacing of the digital interface, through calmer transitions and less frenetic visual elements, counteracted the speed associations that AI triggers.
Third, the framing of time is key. When a green option requires consumers to wait for a product (a delay as a cost), AI interactions tend to reduce uptake. But when the green option involves an extended period of ongoing benefits (a delay as a reward), AI interactions can actually increase uptake. Companies promoting long-term sustainability subscriptions or loyalty programs that accumulate environmental benefits over time may find that AI assistants are natural allies rather than obstacles.
There are important caveats. All four experiments were conducted with Chinese consumers in a music retail setting, and while the researchers conducted a supplementary replication with UK participants in a different product category, the generalizability of these findings to other cultures, industries, and real-world (as opposed to simulated) purchasing environments remains an open question. The studies also relied on hypothetical choices rather than actual purchases, which is common in experimental research but limits conclusions about real spending behavior.
