Artificial intelligence algorithms are increasingly making decisions for consumers, from curating vacation destinations to recommending auto loans. New research indicates that receiving advice from a computerized agent alters human time perception, making people feel that future delays are longer than they actually are. This distorted perception pushes consumers to make more impatient financial decisions. The study was published in the Journal of Consumer Psychology.
Psychologists use the term intertemporal choice to describe decisions that involve a trade-off between different times. A common example is choosing between a smaller reward today and a larger reward a month from now. Humans routinely struggle with these choices, often heavily discounting the value of future rewards because the wait feels unpleasant.
How people experience the passage of time is highly subjective. Psychologists use an “internal clock” theory to explain this phenomenon. The theory suggests that the human brain possesses a cognitive timer that sets the pace of our subjective experience. When a person is relaxed, the internal clock ticks slowly, making time feel like it is flying by.
When a person feels fear or observes fast-moving objects, the internal clock speeds up. Because the brain’s internal timer is ticking away rapidly, the person perceives that a great deal of time has passed. Consequently, the individual experiences the wait as dragging on for an extended period.
Consumers generally associate artificial intelligence with incredibly fast, efficient data processing. The researchers hypothesized that interacting with a fast-processing computer agent would speed up a consumer’s internal clock while they waited for a response. Once the internal clock accelerated, any upcoming wait time for a reward would seem subjectively longer, driving the consumer to choose immediate gratification.
Yuanyuan (Jamie) Li, a researcher at the Southern University of Science and Technology in China, led the investigation. Li collaborated with researchers Shan Lin, Han Gong, Xiang Wang, and Chris Janiszewski. The team conducted a series of online experiments and analyzed real-world financial data to test their hypothesis.
In their first experiment, the researchers asked participants to imagine booking a vacation. Half of the participants requested advice from a human travel specialist, while the other half interacted with an automated chat bot. Both groups waited exactly twenty seconds for a response.
Before seeing their travel options, participants were offered a rebate choice. They could receive thirty dollars in cash immediately or thirty-five dollars in four weeks. Participants interacting with the computer agent were significantly more likely to choose the immediate cash. When surveyed, the automated group reported that the four-week wait felt further away than the human-advised group did.
To confirm that the perception of speed was causing the impatience, the researchers ran a second experiment that manipulated the computer’s reputation. They told half the participants that the computer agent was programmed to spend extra time analyzing data to ensure a high-quality recommendation. By breaking the association between computers and raw speed, the researchers entirely eliminated the impatience effect.
The team then designed an experiment to test the internal clock theory directly. They reasoned that if a computer agent provides an instantaneous recommendation, the consumer’s internal clock never has a chance to run. Participants in this study asked for financial investment advice and received it either immediately or after a fifteen-second wait.
As predicted, the computer agent only induced impatience when participants had to wait for the response. The fifteen-second delay allowed the accelerated internal clock to distort the consumer’s perception of time. When the response was immediate, participants showed no difference in their preferences compared to those consulting a human advisor.
Another experiment demonstrated that the phrasing of time delays matters. Participants imagined buying a mobile phone and navigating a promotional rebate. The researchers offered the delayed rebate as either a time interval, such as “one month,” or a specific calendar date, such as “October 17.”
The researchers found that specific calendar dates anchor the human mind, leaving less room for subjective distortion. When the delay was presented as a calendar date, the identity of the agent had no impact on the consumer’s choice. The computer-induced impatience only surfaced when the delay was presented as a flexible time interval.
The researchers also explored situations involving recurring payments and rewards. They asked participants to choose between federal food assistance programs, offering either a larger weekly payout for nine weeks or a smaller weekly payout for fifteen weeks. Because receiving money repeatedly is a positive experience, a longer time horizon is desirable.
Waiting for a computer agent made the difference between the nine-week and fifteen-week periods feel larger. As a result, participants interacting with the artificial intelligence preferred the longer reward program. The opposite occurred when the researchers asked participants to select auto loans.
Because consumers view recurring loan payments as a negative experience, they generally want the loan period to end quickly. When a computer curated the auto loan options, participants perceived the loan duration as dragging on for a longer period. This pushed them to choose shorter-term loans with higher monthly payments over longer-term loans with lower monthly payments.
To see if this pattern held up outside the laboratory, the researchers analyzed auto loan data from the third quarter of 2022. They compared general industry data against loans originated through LendingTree, a financial services platform that uses an algorithm to curate loan options. They adjusted the data to account for different consumer credit scores across the two data sets.
The real-world data mirrored the laboratory results. Consumers using the computer-curated platform consistently selected shorter loan durations than the industry average. While the researchers note that secondary data cannot definitively prove causation, the results align perfectly with their experimental findings.
The authors acknowledge that the marketplace contains many variables that could influence real-world consumer behavior. Customers using financial technology apps might be more price-sensitive than those visiting brick-and-mortar banks, or they might prefer to aggressively minimize their loan durations for other reasons. Algorithms might also trigger feelings of uncertainty about the future, prompting consumers to lock in immediate gains.
Future investigations into artificial intelligence could explore complex decisions that mix both costs and benefits over time. For now, the research highlights a hidden psychological cost of automated convenience. As companies shift to computerized customer service, they risk making their customers feel that the future is further away than ever.
The study, “Time is shrinking in the eye of AI: AI agents influence intertemporal choice,” was authored by Yuanyuan (Jamie) Li, Shan Lin, Han Gong, Xiang Wang, and Chris Janiszewski.
