Patricia Rossi is an Associate Professor in Marketing at SKEMA Business School. She is a Researcher at the Market Interactions SKEMA Centre. Her research interests include Decision Making, Sensory Marketing, Sustainable Marketing, and Artificial Intelligence. She is currently involved in a research project to investigate how awareness of behavioural biases and the use of robo-advisors can impact financial decision-making.
Could you tell us more about your field of research, particularly decision-making and artificial intelligence?
Overall, my interests are related to consumer behaviour and consumer psychology. I am interested in everything that helps us understand why we do the things we do or why we make the choices we make. I will give you an example. In a recent research project (“Delegation of purchasing tasks to AI: The role of perceived choice and decision autonomy”), my co-author Mariyani Ahmad Husairi and I investigated whether consumers’ autonomy perception could impact artificial intelligence (AI) adoption in a purchase context. Nowadays, AI is involved in almost everything we do. This technology is supposed to make our lives easier by taking the lead on some tasks, like our purchases. But do we always want AI to take the lead? The answer is no. With the growth of AI, consumers do not always make their own purchase decisions, which negatively impacts their autonomy perception. Despite the potential benefits of AI making decisions for us, we still want to be in charge of our choices. We found that there are two dimensions of consumer autonomy in a purchase context: choice autonomy and decision autonomy. Our studies consistently show that the lower the perceived autonomy, the lower the AI adoption likelihood. Stated otherwise, when consumers perceive they have choice and decision autonomy, they are more likely to adopt an AI-enabled technology than when they feel a lack of autonomy.
What fascinates me about this research area is that, in many situations, the decisions we make are not necessarily logical or optimal. If we want to change that and have a positive impact on people’s lives, we need first to understand the reasons behind their behaviours.
What results surprised you the most?
There are a few surprising findings. First, in a purchase context, autonomy can be divided into two dimensions. There is choice autonomy and decision autonomy. Choice autonomy refers to autonomy in narrowing down possible choices. Decision autonomy refers to actively making the final purchasing decision. In a purchase situation, a consumer may have full autonomy in both dimensions, no autonomy in either dimension, or full autonomy in just one dimension. What is interesting, however, is that both dimensions of autonomy matter to consumers.
The second surprising finding is that consumers prefer to keep their autonomy even when the purchase decision is complex. Our initial thought was that in complex purchase situations, the use of AI could be beneficial. For example, if a consumer had to make a choice based on 20 (instead of only 3) important attributes, would this consumer be more willing to give up their autonomy? When complexity factors like limited attention, number of product attributes, and cognitive effort are involved, we tend to rely more on AI recommendations. To our surprise, even when faced with such complexities, consumers preferred to retain their choice and decision autonomy.
A third surprising finding is that consumers want to keep their autonomy in most situations, except for when they believe AI can help them purchasing things that are required to perform identity-relevant activities. For example, fishing, baking, and running are identity-relevant activities. Prior studies have highlighted that when technology takes over functions that are relevant to one’s identity, the result is technology aversion. This is mainly because we want to perform those activities ourselves. In contrast, our studies show that when an activity is important to us, we accept to relinquish our purchasing autonomy to AI if it ends up helping us to perform such activity. Our research shows that an avid runner is willing to let AI purchase his running shoes in comparison to a casual runner. These effects occur because AI-enabled purchasing tools, in this case, complement consumers’ identity-related goals while allowing the attribution of the outcome to themselves. For example, while AI purchases the running shoes, which saves time and energy for the runner, the activity of running itself is still performed by the person.
The findings of this study are applicable even when customers are not aware that AI is applied behind their decision support systems. For example, customers may not realize that online retailers such as Amazon use AI to narrow down options and generate recommendations. They may notice, however, that their options are restricted, leading to a perceived lack of choice autonomy. This, in turn, may lead to a retailer rejection or boycott. Taken together, the findings of our study show that customers’ desire to preserve their autonomy exceeds the need to reduce the time and effort resulting from complex decisions.
The outputs of this project can contribute to building a road where cooperation between humans and AI is possible. Since the ubiquitous presence of AI is a reality, we need to understand how we can get the most out of this technology.
What is the research you are currently carrying out?
Currently, with Mariyani Ahmad Husairi, Daniel Fernandes, and Alexandre Alles, we are trying to understand how awareness of behavioural biases that harm investment outcomes, such as the disposition effect, and the use of robo-advisors, can impact financial decision-making. This project started with a discussion about robo-advisors, an AI-based financial service that has been showing impressive growth in the market. Given their nature, robo-advisors could be a potential solution to avoid behavioural biases and obtain positive outcomes in financial investments. Yet, many consumers/investors resist using robo-advisors. In this project, we investigate whether awareness of behavioural biases can result in less biased human decision-making and increase in the adoption of robo-advisors. We also aim to investigate which type of message framing is effective in communicating about behavioural biases to impact the use of robo-advisors. So far, the findings of a few studies, including a field study, suggest that raising awareness of human biases in financial decision-making increases the likelihood of using robo-advisors and reduces the disposition effect. Now, we need to understand what kind of message framing will be more effective in doing so.
How does this study impact society at large?
With the growth of robo-advisors, more people will have access to more—and hopefully better—financial investments. This happens because the initial investment and the service cost of robo-advisors tend to be lower compared to those involved in having a fully dedicated human advisor. In addition, robo-advisors may contribute to better financial performance. Since robo-advisors are programmed to avoid behavioural biases, their use can have a positive impact on people’s financial well-being. Therefore, with this research, we expect to contribute to the understanding of what motivates people to accept and adopt AI in their decision-making, creating opportunities for a positive impact on their lives.
Read also: Berill Blair: “Climate services are still underutilised in many high-stakes decision environments”