Embracing AI advisors for making (complex) financial decisions: an experimental investigation of the role of a maximizing decision-making style

Artificial Intelligence (from hereon: AI) is transforming human life at a rapid pace, becoming integral to various facets of individuals’ daily existence and driving substantial advancements across multiple domains. Consumers now routinely rely on AI agents for a variety of daily tasks (Logg et al., 2019), such as receiving entertainment recommendations on platforms like Netflix or Spotify, using voice assistants like Alexa or Siri at home (Acikgoz et al., 2023), and relying on AI-powered investment and trading tools like Betterment for investment decisions and portfolio management (Belanche et al., 2020). This widespread reliance is unsurprising, as algorithms consistently outperform humans in predictive tasks (Elkins et al., 2013), demonstrating greater accuracy in forecasting academic success and clinical diagnoses (Dawes, 2008; Dawes et al., 1989; Grove et al., 2000).

Despite their superior performance and increasing usage in general, there are still situations in which individuals often reject algorithms in their decision-making, a tendency known as algorithm aversion (Dietvorst et al., 2015). This aversion is particularly strong in situations involving uncertainty (Burton et al., 2020; Dietvorst and Bharti, 2020). This is especially relevant in high-stakes areas such as financial decision-making, where uncertainty is inherent, and individuals often display algorithm aversion due to skepticism about AI’s ability to understand the complexity of one’s personal situation (Filiz et al., 2022; Hentzen et al., 2021). As a result, they tend to dismiss AI tools even when these could improve outcomes (Downen et al., 2024). This aversion can result in suboptimal decisions that reduce expected utility (Dietvorst et al., 2015). For example, by rejecting AI-driven tools for financial decisions, consumers miss the opportunity to mitigate behavioral biases, while also foregoing easy ways to improve portfolio diversification, which could enhance overall financial performance (Back et al., 2023; Bhattacharya et al., 2012; D’Acunto et al., 2019). As a result, it is imperative to find ways to reduce consumers’ algorithm aversion in the financial decision-making context.

To date, the existing literature has identified algorithm aversion as a primary factor influencing consumers’ acceptance of algorithm-based decision tools. While researchers have explored many factors contributing to algorithm aversion (Mahmud et al., 2022), only a limited number of studies examine the role of psychological factors and cognitive tendencies. In particular, there is still limited insight into how different decision-making styles and mindsets influence the use of AI advisors. However, such an understanding is critical for both theory and practice, as decision-making styles and mindsets can be externally induced (Ma and Roese, 2014; Silber et al., 2024), potentially shaping individuals’ interactions with AI systems (Khan et al., 2024). Exploring how decision-making styles, particularly individuals’ tendency to maximize, influence the use of AI advisors in consequential decision-making domains like personal finance offers valuable insights for advancing theoretical models of AI interaction and guiding practical applications to enhance user adoption of AI recommendation agents.

A maximizing decision-making style aims at making the best possible decision, unlike satisficing, where individuals settle for an option that meets their minimum requirements (Schwartz et al., 2002). Previous research has demonstrated that maximizers invest more time and resources into their decision-making (Dar-Nimrod et al., 2009), aiming to explore and compare a wide range of options to achieve their goal(s) (Cheek and Schwartz, 2016). To achieve the best decision, maximizers prioritize autonomy, believing that full control over the decision-making process allows them to explore all available options, ensuring they do not miss a better alternative (Dar-Nimrod et al., 2009). However, this can lead maximizers into the “Sisyphus effect” where they restart the decision process even in similar situations, leading to decision fatigue and frustration (Carrillat et al., 2011; Dar-Nimrod et al., 2009). The exhaustive search for the best option is a cognitive burden, leaving maximizers mentally drained despite their efforts to improve decision quality (Carrillat et al., 2011; Misuraca and Teuscher, 2013).

While much of the existing research has focused on how maximizing behaviors affect well-being (Belli et al., 2022), less attention has been paid to how decision aids, such as AI tools, might alleviate cognitive load and, in turn, enhance the quality of maximizers’ decisions. By providing data-driven insights and reducing the number of options to evaluate, AI can help streamline maximizers’ decision-making process without compromising their need for thorough analysis (Dietvorst and Bartels, 2022; Haenlein and Kaplan, 2019). Although maximizers prioritize autonomy, relying on AI could allow them to maintain high decision standards while mitigating decision fatigue. Thus, our study seeks to explore how a maximizing decision-making style affects the usage of AI advisors, addressing the following research question: “What is the relationship between a maximizing decision-making style and the likelihood of using an AI advisor, and what are the underlying psychological processes that drive this relationship?”

Building on existing literature, this paper argues and finds that a maximizing decision-making style is positively associated with a preference for using AI advisors in financial contexts due to higher perceived effectiveness and reduced algorithm aversion. While maximizers are more neurotic and anxious (Purvis et al., 2011)—traits linked to increased algorithm aversion and reduced acceptance (Sharan and Romano, 2020; Van Esch et al., 2021)—we actually show that maximization increases the propensity to use AI advisors. This is because this decision-making style may increase the perception that algorithms are effective in analyzing large datasets to achieve the best outcomes, reducing algorithm aversion, and thus increasing the propensity to use AI (Dietvorst and Bartels, 2022; Haenlein and Kaplan, 2019). Given that AI excels at processing complex financial data and mitigating biases, it aligns well with maximizers’ pursuit of the best possible decisions in high-stakes financial environments (Schwartz et al., 2002).

Our paper makes both theoretical and practical contributions to the existing literature. Theoretically, we expand the literature on the effect of maximization on information seeking, by considering AI advisors as a source. While prior studies focused on maximizers’ individual search behaviors and emotional responses (Dar-Nimrod et al., 2009), we examine their reliance on external advisors in a financial context, finding a positive link between maximizing and the likelihood of using an AI advisor for financial decision-making, doing so addresses calls for research on the positive effects of maximizing (Misuraca et al., 2021). Additionally, by introducing maximization as a decision-making style that reduces algorithm aversion, we shift the focus of existing work that studies factors such as expertise and personality (Mahmud et al., 2022). We further explore this relationship through serial mediation by perceived effectiveness and algorithm aversion, responding to Castelo et al.'s (2019) call for more research in this regard.

Practically, our findings offer business practitioners actionable insights on how to mitigate algorithm aversion and increase the use of AI advisors in financial decision-making. Financial service providers such as banks, pension funds, and insurance companies can adopt these strategies to foster greater acceptance of AI, doing so could ultimately benefit consumers through being able to offer personalized services and streamlined processes at a reduced cost, enhancing financial outcomes (Nicholas, 2022). As AI can optimize services and reduce costs, it can strengthen the quality of financial advice while maintaining cost-effectiveness, positioning AI-based advisors as an effective alternative to humans (Shanmuganathan, 2020).

Algorithm aversion, the tendency to reject AI-driven decision-making tools in favor of human judgment, remains a significant barrier to the adoption of these technologies (Dietvorst et al., 2015). This aversion persists despite the superior performance of algorithms in numerous domains, including predictive accuracy in financial decision-making and other high-stakes environments (Germann and Merkle, 2023; Hollebeek et al., 2024). The extant literature identifies several psychological traits and cognitive tendencies as key contributors to this phenomenon (Mahmud et al., 2022). For instance, individuals with higher self-efficacy tend to prefer their own decisions over those suggested by algorithms (Araujo et al., 2020; Logg et al., 2019). Additionally, traits like neuroticism and anxiety contribute to algorithm aversion due to feelings of untrustworthiness in technology (Meuter et al., 2003; Sharan and Romano, 2020; Van Esch et al., 2021). In contrast, adopting a promotion-focused mindset can reduce this aversion (Khan et al., 2024). Table 1 summarizes the current research landscape in terms of known factors driving algorithm aversion and underscores the contribution of the present study.

A maximizing decision-making style refers to the extent to which individuals strive to make the best choices, varying along a continuum from accepting satisfactory options (satisficing) to seeking the best possible option (maximizing) (Schwartz et al., 2002). Initially conceptualized as a three-dimensional construct encompassing high standards, alternative search, and decision difficulty (Schwartz, 2004), recent research supports a two-dimensional construct focusing on high standards (i.e. the goal of choosing the best option) and alternative search (i.e. the strategy of seeking and comparing alternatives to achieve one’s ultimate goal) (Cheek and Schwartz, 2016). Decision difficulty (i.e. the difficulty of choosing an option) is regarded as a consequence or antecedent of maximizing rather than an intrinsic component (Cheek and Goebel, 2020). Accordingly, this paper conceptualizes maximizing as a two-dimensional construct, recognizing its manifestation as both a dispositional trait and a situational mindset characterized by a tendency to compare and a goal of obtaining the best (Ma and Roese, 2014). Figure 1 illustrates the overarching framework of our study. In the next sections, we will discuss each hypothesis in detail.

During the decision-making process, individuals with maximizing tendencies are inclined to invest more time and resources in their search (Dar-Nimrod et al., 2009), often resisting efforts to simplify or expedite the process, even when effective search strategies are readily available (Carrillat et al., 2011). While maximizers highly value autonomy in decision-making, focusing solely on maintaining complete control can, paradoxically, hinder their goal of improving the decision quality (Dar-Nimrod et al., 2009). This relentless pursuit of the best choice becomes particularly relevant when it comes to seeking advice from recommendation agents, especially those powered by AI, as such systems are capable of processing vast datasets and offer personalized, data-driven recommendations (Dietvorst and Bartels, 2022; Haenlein and Kaplan, 2019; Sheth et al., 2022). These capabilities resonate with maximizers’ desire for thoroughness and accuracy (Schwartz et al., 2002), potentially alleviating the cognitive burden associated with maximizing behavior (Misuraca and Teuscher, 2013; Polman and Maglio, 2022). This alignment is especially pertinent in high-stakes decision-making environments, such as financial decision-making, where the evaluation of multiple alternatives is crucial for making well-informed choices. In such contexts, AI-driven recommendations should be particularly attractive to maximizers, as they provide data-driven insights facilitating the pursuit of the most accurate and advantageous outcome (Back et al., 2023; Germann and Merkle, 2023).

Despite research suggesting an aversion to algorithmic input in decision-making processes (Dietvorst et al., 2015), those with maximizing tendencies may be inclined to consult an AI recommendation agent. This inclination aligns with the perception that algorithms employ a maximizing decision-making strategy, aiming to attain the best possible outcome (Dietvorst and Bartels, 2022). AI recommendation agents’ ability to assimilate extensive data and optimize information, avoiding biases and emotions, aligns with maximizers’ inherent drive for the best choice (Schwartz et al., 2002). Moreover, research suggests that maximizers, while valuing the autonomy provided by the decision-making process, may face challenges such as decision fatigue and information overload (Dar-Nimrod et al., 2009; Schwartz et al., 2002). These challenges arise from the substantial cognitive effort required by maximizing, which challenges one’s memory and compels one to repeatedly compare options (Polman and Maglio, 2022). This repeated comparison increases cognitive load and creates a self-reinforcing cycle, leading to consistently engaging in more option comparisons and exerting more effort in decision-making (Misuraca and Teuscher, 2013; Polman and Maglio, 2022). In this context, the efficiency of AI recommendation agents should be particularly appealing to maximizers engaged in financial decision-making, as these tools not only broaden the scope of available options, but can also alleviate the cognitive load associated with extensive searches for the best possible financial outcome. Consequently, we expect individuals primed to maximize to be more likely to use AI recommendation agents in their financial decision-making processes.

To better understand why maximizers might be more likely to consult AI recommendation agents in their financial decisions, it is important to examine the underlying psychological process. Individuals with a maximizing decision-making style strive for the best decisions and are inherently inclined towards thorough and comprehensive decision-making processes (Schwartz et al., 2002). This predisposition becomes even more critical in financial contexts, where minor variations in choices—such as investment strategies or savings plans—can have significant long-term financial consequences (Lynch, 2011). However, despite the considerable resources they invest in expanding their set of alternatives (Chowdhury et al., 2009; Dar-Nimrod et al., 2009), maximizers often believe that there may still exist better alternatives that they have not found yet (Sparks et al., 2012). To overcome this belief and improve their search process, maximizers should value tools that can accurately, efficiently, and reliably help them maximize their objective. Perceived algorithmic effectiveness—defined as the belief that algorithms perform tasks with accuracy, efficiency, and reliability—plays a pivotal role in this context, as it directly influences individuals’ decisions to use algorithmic advice in their decision-making (Castelo et al., 2019; Wirtz et al., 2018; Zhu et al., 2022). Given that algorithms inherently employ a maximizing decision-making strategy (Dietvorst and Bartels, 2022), we expect that individuals with a stronger maximizing decision-making style are more likely to perceive algorithms as effective tools for achieving their financial goals.

However, perceived effectiveness alone is unlikely to fully explain maximizers’ preference for AI, as decision-making involves both rational evaluations and emotional factors. Even when AI is seen as effective, high levels of algorithm aversion can reduce its usage (Castelo et al., 2019; Dietvorst et al., 2015). Thus, only when perceived effectiveness is accompanied by low algorithm aversion can maximizers be expected to fully embrace AI in their financial decision-making. Consequently, when maximizers view algorithms as effective and therefore experience lower levels of algorithm aversion, they are more inclined to rely on AI advisors in their financial decision-making, as these tools promise to enhance the accuracy and efficiency of their financial choices (Fernandes et al., 2014)—a relationship that we discuss as a process of serial mediation of perceived effectiveness and AI aversion in the next section. Note that in a robustness check, we will also demonstrate mediation by each variable in isolation.

Despite the potential efficiency advantages, algorithm aversion remains a significant barrier to the widespread utilization of AI recommendation agents. This aversion is the tendency for individuals to reject algorithms in favor of human advisors despite evidence of algorithms’ superior performance (Dietvorst et al., 2015). However, research shows that providing explanations for how an algorithm works (Cadario et al., 2021), and showing clear evidence of its effectiveness (Dietvorst et al., 2018) diminishes individuals’ propensity to reject algorithmic advice. When individuals perceive algorithms as effective, they are more likely to rationalize that relying on them is a sound decision, reducing aversion (Castelo et al., 2019; Zhu et al., 2022). This cognitive evaluation can occur because the perceived benefits outweigh biases against algorithms (Logg et al., 2019). In financial decision-making, this shift in perception is particularly important, as the high stakes amplify the need for accurate, data-driven recommendations (Back et al., 2023). Additionally, when algorithms are seen as effective, individuals tend to attribute any errors to situational factors, such as data quality, rather than inherent flaws in the algorithm, increasing their tolerance for errors and reducing overall aversion (Dietvorst et al., 2018). Hence, we suggest that individuals primed to maximize will perceive algorithms as more effective and consequently reduce their aversion towards them.

Individuals who exhibit algorithm aversion often favor less accurate human judgment over algorithmic decision-making, preferring to seek advice from human experts or rely on their intuition rather than trusting algorithmic systems (Dietvorst et al., 2015; Logg et al., 2019; Önkal et al., 2019; Shaffer et al., 2013). This aversion is rooted in the perception that algorithms are opaque, unpredictable, error-prone, unable to learn from mistakes, incapable of performing subjective tasks, and cannot account for humans’ unique characteristics (Castelo et al., 2019; Chandra et al., 2022; Dietvorst et al., 2015; Hollebeek et al., 2024; Longoni et al., 2019; Pradeep et al., 2018; Reich et al., 2023). Additionally, individuals might be reluctant to accept decisions proposed by AI because its maximizing strategy, which prioritizes the best outcomes while disregarding other important factors such as moral considerations, context sensitivity, and decision process transparency in the decision-making process, is seen as objectionable (Dietvorst and Bartels, 2022). Such findings align with Dietvorst et al. (2015) research, illustrating that algorithm aversion negatively impacts the likelihood of using AI, as individuals with such aversion are less likely to accept algorithmic advice, even when it is more accurate than human judgment. In financial decision-making, where accuracy and efficiency are paramount, the complexity of evaluating numerous financial options may prompt individuals—especially maximizers—to see AI as a necessary tool for navigating large datasets and making informed choices. Accordingly, we propose that individuals primed to maximize are more likely to perceive algorithms as effective, leading to reduced algorithm aversion and ultimately a higher likelihood of using an AI advisor in their financial decision-making processes.

In the following, we describe how we combine an exploratory pre-study with a main study to experimentally test our hypotheses on the role of a maximizing decision-making style in embracing AI for financial decisions, as well as the underlying psychological mechanisms.

Our pre-study investigates how individuals with different levels of maximizing tendencies engage with and experience AI advisors. The primary purpose of this pre-study is to gather insights that inform the design of the financial scenarios used in our main study, ensuring their relevance and alignment with participants’ experience. Participants are sourced through Prolific, focusing on US residents, and undergo a preliminary screening to confirm recent interaction with AI technologies. Specifically, we asked individuals if they had interacted with AI in the past three months. Based on their responses, 29 individuals with relevant experience ultimately participated.

Of the sample, 55.2% identify as female and 44.8% as male, which constitutes a slight underrepresentation of males compared to the general population (US Census Bureau, 2021). The average age of 35.21 years is slightly below the national average of 38.8 years, but this discrepancy may be attributed to our focus on individuals aged 18 and above to comply with the ethical guidelines of our university’s Institutional Review Board, while the US average is based on also including younger individuals ^[1]. In terms of education, 64.6% of participants hold a university degree (with 37.9% holding a bachelor’s degree, 20.7% a postgraduate degree, 3.4% a Doctorate degree, and 3.4% a professional degree), surpassing the national average of 35% who hold a bachelor’s degree or higher. Regarding employment, 58.6% of participants have jobs, which aligns with the national average of 58.6%, while 13.8% are unemployed, 10.3% are students, and 10.3% are self-employed. The median gross annual income category of $10,276 to $41,776 is below the national median of $50,000 to $74,999 (US Census Bureau, 2021).

After presenting participants with a definition of AI advisors as “artificial intelligence systems designed to analyze large volumes of data and offer personalized advice tailored to users’ preferences and needs to assist in their decision-making processes, such as chatbots, search engines, and recommendation algorithms,” they were asked to respond to five open-ended questions. These questions explored their recent experiences with AI advisors (“Please describe an instance in the past three months where you used an AI advisor to assist you in a decision-making process. What decision did you need to make?”, “What motivated you to consult an AI advisor for this decision?”, “How did you feel about the experience/assistance provided by the AI advisor during your decision-making process? And why?”, “Please share your reasons for whether you followed or did not follow the advice given by the AI advisor.”, and “How satisfied were you with the assistance or outcome provided by the AI advisor?”). Following the conceptualization of maximizing behavior by Cheek and Schwartz (2016), we treat it as a two-dimensional construct, comprising “high standards” and “alternative search”. To operationalize these dimensions, we utilized the 9-item Maximization Tendency Scale (Diab et al., 2008) for the “high standards” dimension (α = 0.831), and the 12-item Maximization Inventory (Turner et al., 2012) for the “alternative search” dimension (α = 0.925).

Using NVivo 20 software, we conducted a thematic analysis to identify and interpret patterns in participants’ open-ended responses (Braun and Clarke, 2006), specifically focusing on their AI experiences within financial decisions and their motivations for adoption. To compare decision-making styles, we used a median-split method based on participants’ maximizing tendency (median_{high standards} = 5.44; median_{alternative search} = 5.25). Participants scoring above the median were classified as maximizers (n = 14 for both high standards and alternative search), and those below as satisficers (n = 15 for both), enabling a clearer comparison of decision-making styles, consistent with prior research by Iyengar et al. (2006). Using hierarchical coding following King (2012), we grouped similar codes and identified key patterns in how these groups utilize AI for financial decisions. This method provided insight about differences in AI-driven financial decision-making between satisficers and maximizers.

We identified several key similarities between satisficers and maximizers in their use of AI for financial decision-making, as detailed in Table 2. Both groups valued AI for its efficiency and convenience, especially in helping streamline decision-making processes such as portfolio management and budgeting. For example, AI’s ability to process large sets of financial data quickly, from stock trends to retirement planning options, was consistently mentioned as a major benefit. In addition, both satisficers and maximizers trusted AI’s data-driven insights, relying on its objective analysis to assist in making financial decisions. Whether managing taxes or understanding the implications of stock dividends, participants from both groups considered AI as a valuable tool for handling data-intensive financial tasks.

Despite these similarities, there were also notable differences between the two groups in terms of their interaction with AI. In particular, satisficers were more likely to use AI for isolated financial tasks, such as investment selection or tax advice, and generally expressed satisfaction when AI provided practical and generic responses that solved immediate problems. In contrast, maximizers integrated AI more deeply into their financial lives, using it for continuous, strategic financial management. They relied on AI for real-time portfolio optimization, continuously monitoring market changes and adjusting their investment strategies accordingly. This group also demanded higher levels of personalization and relevance, often expressing dissatisfaction when AI’s advice didn’t meet their specific financial goals. Furthermore, while satisficers tended to verify AI advice with external sources or their personal judgment, maximizers adopted a more strategic verification approach, ensuring that AI’s recommendations aligned with their investment objectives before implementation.

The pre-study revealed behavioral differences between satisficers and maximizers in their use of AI for financial decision-making. Specifically, we found directional evidence suggesting that maximizers exhibited an overall more positive attitude towards AI usage. Additionally, the thematic analysis indicated that satisficers tend to use AI for straightforward, one-off tasks, whereas maximizers engage with AI for more complex, ongoing financial decisions. These thematic findings and the positive relationship between maximizing and AI usage provide initial insights into decision-making patterns. We leveraged these insights to design the main study’s financial scenario, focusing on how priming maximization tendencies impacts AI usage in complex financial contexts, and identifying the underlying psychological process.

Our main study utilizes an experimental approach to test hypotheses H1 and H2, by examining how situationally inducing maximizing affects the likelihood of using an AI advisor for financial decision-making, as serially mediated by the perceived effectiveness of algorithms and algorithm aversion. It is important to recognize that situational maximizing integrates the dimensions of high standards and alternative search, which are evaluated separately when maximizing is examined as a personality trait (Cheek and Schwartz, 2016; Ma and Roese, 2014).

We recruited 140 US participants from Prolific and randomly assigned them to a two-cell between-subject experimental design (priming of decision-making mindset: maximizing vs satisficing). The purpose of the priming was to activate these distinct decision-making mindsets to observe their influence on participants’ likelihood of using an AI advisor for financial decision-making. Following prior work (Silber et al., 2024), we utilized the priming questions by Ma and Roese (2014). Specifically, participants in the maximizing mindset condition were asked a series of five questions prompting them to report their “best choice” in various financial scenarios, such as “Which rewards package do you think is the best?” with options including a $75,000 annual salary with a $5,000 bonus, an $80,000 annual salary, or a $78,000 annual salary with a $2,000 bonus. Conversely, participants in the satisficing mindset condition were asked five questions prompting them to report a “good enough” choice (e.g. “Which rewards package do you think is good enough?”). These prompts encouraged participants to adopt a decision-making mindset influencing their response to the later scenario. In designing the financial scenario for the main study, we drew directly upon the pre-study. The pre-study provided critical insights into how individuals with varying levels of maximizing tendencies engage with AI advisors, ensuring that the task was both realistic and aligned with the study’s objectives. This approach enhanced the integrity and robustness of the findings. Results from a manipulation check confirmed the effectiveness of the experimental priming ^[2].

Next, drawing on the findings of the pre-study, we provided participants with the following scenario: “Imagine building an investment portfolio for your retirement that contains stocks, bonds, and other financial instruments” and asked them to report their likelihood of using an AI advisor for such a task. We assessed perceived algorithm effectiveness (α = 0.841) using a 3-item, 7-point Likert scale from Castelo et al. (2019), with items such as “I can see the benefits in algorithms that can perform this kind of task better than humans.” Additionally, we measured algorithm aversion (α = 0.866) using a 7-item, 7-point Likert scale from Germann and Merkle (2023), including statements like “On average, investment funds based on investment algorithms achieve higher returns than those managed by fund managers.” We also included standard socio-demographic questions such as age, gender, education, and income as control variables. Finally, given the specific context of our study, we also included AI literacy as a control variable (Hornberger et al., 2023), measured through 5 questions, each with four multiple-choice response options, focusing on key AI-related concepts. These included: intelligence of AI (“Why do AI systems behave intelligently?”), similarities of humans and AI (“How are humans and AI similar?”), decision-making (“How do AI systems make decisions?”), programmability (“What determines the behavior of AI systems?”), and learning from data (“Why can systems based on machine learning obtain good results?”). As the AI literacy scale is a formative measure, calculation of Cronbach’s alpha does not apply to this scale and is therefore not reported here.

We excluded 25 participants who provided incomplete, invalid, or implausible responses, such as failing attention checks or exhibiting straight-lining behavior in their answers. The final sample thus comprised 115 participants, with 46 males and 69 females, resulting in a slight underrepresentation of males compared to the nearly equal gender distribution in the US population (US Census Bureau, 2021). The average participant age was 41.3 years, which is somewhat higher than the US average of 38.8 years. This age difference is likely again due to our sampling criteria, which only included individuals aged 18 and older. Approximately 63% of the participants hold a university degree (38.3% have a Bachelor’s degree, 16.5% a Master’s degree, 5.2% a Doctorate degree, and 3.5% a professional degree), which is higher than the national average of 35% holding a Bachelor’s degree or higher. Further, 73.9% hold some form of employment, which is above the US average of 58.6%. Additionally, 11.3% are not looking for work, 8.7% are looking for work, and 6.1% are still studying. Finally, the median gross annual income category of $10,276 to $41,776 is lower than the national median of $50,000 to $74,999 (US Census Bureau, 2021). There are no significant differences between the two experimental conditions in terms of participants’ gender, age, or employment status (all ps > 0.100). However, we observed a significant difference in the income levels of participants between the conditions (p = 0.041), with those in the maximizing condition reporting a higher mean income category (M = 2.97, SD = 1.075) than those in the satisficing condition (M = 2.58, SD = 0.925). Given this difference, in our analysis we controlled for income to avoid any confounding effect of participants’ income level. The experimental groups were balanced in size, with 58 (57) participants in the maximizing (satisficing) mindset condition, thereby confirming successful random assignment.

We find a marginally significant positive direct effect of inducing a maximizing (Mean = 4.45, SD = 1.779) compared to a satisficing mindset (Mean = 3.84, SD = 1.859) on the likelihood of using an AI advisor (t(113) = 1.787, Cohen’s d = 0.333, p = 0.077), which supports H1. In addition, those primed to maximize (Mean = 5.08, SD = 0.659) perceive algorithms as significantly more effective than those primed to satisfice (Mean = 4.73, SD = 1.127; t(89.952) = −2.059, Cohen’s d = 0.921, p = 0.042). Moreover, individuals in the maximizing condition (Mean = 3.52, SD = 1.039) exhibit significantly lower algorithm aversion compared to those in the satisficing condition (Mean = 4.11, SD = 0.978; t(113) = 3.157, Cohen’s d = 1.009, p = 0.002).

To formally test the serial mediation of inducing a maximizing vs satisficing mindset on the likelihood of using an AI advisor by perceived effectiveness of algorithms and algorithm aversion, we applied Model 6 from Hayes and Preacher’s (2014) SPSS Process macro, utilizing 5,000 bootstrap samples. The results indicate a positive indirect effect of inducing a maximizing mindset on the likelihood of using an AI advisor through higher perceived effectiveness of algorithms and lower algorithm aversion (β = 0.1232; 95% CI = 0.089–0.2935). We found a significant mean difference between the maximizing and satisficing conditions in the likelihood of using an AI advisor as a direct effect, indicating that a maximizing mindset directly influences the likelihood of using AI. Hence, the absence of a direct effect in the mediation analysis indicates that this relationship is fully mediated by perceived effectiveness and algorithm aversion. In particular, priming a maximizing mindset leads to a stronger perception of the effectiveness of algorithms (β = 0.3553, p = 0.041) in comparison to priming a satisficing mindset. Additionally, we find that the perceived effectiveness of algorithms decreases individuals’ algorithm aversion (β = −0.4503 p = 0.000). Finally, algorithm aversion decreases an individual’s likelihood of using an AI advisor (β = −0.7699, p = 0.000). These findings support H2 and are graphically illustrated in Figure 2. Furthermore, even after we control for socio-demographic variables, including income, our findings remain qualitatively the same, with no changes to the significance of the key variables.

As a robustness check, we also conducted separate mediation analyses for each mediator in isolation instead of the previously reported serial mediation analysis that includes both mediators simultaneously. In particular, using Hayes and Preacher’s (2014) SPSS Process macro (Model 4) with 5,000 bootstrap samples, we first examined the mediating role of perceived effectiveness of algorithms. Our results revealed a positive indirect effect of inducing a maximizing mindset on the likelihood of using an AI advisor, mediated by higher perceived effectiveness of algorithms (β = 0.2400; 95% CI = 0.0176–0.4953). Specifically, compared to inducing a satisficing mindset, inducing a maximizing mindset led to a stronger perception of algorithm effectiveness (β = 0.3553, p = 0.041), and this increase in perceived effectiveness in turn significantly increased the likelihood of using an AI advisor (β = 0.6755, p = 0.002). Next, we analyzed the mediating role of algorithm aversion. Here, we find a positive indirect effect of inducing a maximizing mindset on the likelihood of using an AI advisor through reduced algorithm aversion (β = 0.5305; 95% CI = 0.2043–0.8893). Specifically, compared to inducing a satisficing mindset, inducing a maximizing mindset was associated with lower algorithm aversion (β = −0.5939, p = 0.021), and this reduction in algorithm aversion in turn significantly increased participants’ likelihood of using an AI advisor (β = 0.8933, p = 0.000).

In this study, we examined the association between a maximizing decision-making style and the likelihood of using an AI advisor and the psychological process underlying this relationship. In a pre-study, we explored dispositional maximizing using validated scales and assessed participants’ experiences with using AI. While both satisficers and maximizers valued AI for its efficiency, satisficers tended to use AI for isolated tasks, whereas maximizers integrated it more deeply into ongoing financial management. These insights shaped the experimental scenario of the main study, where participants were asked to imagine constructing a retirement investment portfolio with stocks, bonds, and other financial instruments, reflecting real-world tendencies observed in the pre-study. We hypothesized that maximizing is linked to a higher likelihood of using an AI advisor in a financial context, with this effect being serially mediated by the perceived effectiveness of algorithms and algorithm aversion. We experimentally induced a maximizing mindset and observed this direct effect. Furthermore, we found that the relationship between maximizing and the likelihood of using an AI advisor in a financial context is serially mediated by the perceived effectiveness of AI algorithms and algorithm aversion. In a robustness check, we also demonstrated mediation by each of these variables in isolation. Next, we discuss the broader theoretical and practical implications of our research.

Theoretically, our study contributes to the literature on decision-making and its intersection with AI adoption, with a focus on maximizing. Specifically, this research advances our understanding by examining how maximizing tendencies influence the selection and use of decision aids, such AI advisors in a high-stakes financial context—an area overlooked by previous literature. While prior research has primarily focused on the individual choice process and emotional responses of maximizers (Carrillat et al., 2011; Dar-Nimrod et al., 2009; Luan and Li, 2019; Misuraca and Teuscher, 2013), our findings establish a novel link between maximizing and an increased likelihood of using an AI advisor, thereby contributing new insights to the field. Our study also introduces a theoretical perspective on how AI tools align with the cognitive and psychological characteristics of maximizers. Although maximizers highly value autonomy, the substantial cognitive load involved in their exhaustive search for the best decision outcome often results in decision fatigue (Dar-Nimrod et al., 2009; Huang and Li, 2024). This cognitive strain highlights how decision aids can reduce the burden while enhancing the strategies maximizers use in decision-making. We identify AI advisors as a valuable resource for maximizers, offering efficiency and speed while complementing their desire for thoroughness and accuracy. By providing data-driven insights and facilitating more efficient decision-making, AI advisors enable maximizers to achieve high-quality outcomes without compromising their preference for a comprehensive decision process (Dietvorst and Bartels, 2022; Haenlein and Kaplan, 2019). This alignment highlights the theoretical importance of AI as a mechanism to address the cognitive demands of maximizers while maintaining their decision-making strategy and goal.

Our paper contributes to the literature by examining the relationship between a maximizing decision-making style and the propensity to use AI advisors in financial decision-making, while uncovering the underlying psychological mechanisms. Unlike prior studies that primarily associate a maximizing decision-making style with neuroticism and anxiety (Purvis et al., 2011; Sharan and Romano, 2020)—traits linked to heightened algorithm aversion (Meuter et al., 2003; Van Esch et al., 2021)—our findings demonstrate that maximization can, in fact, enhance the likelihood of adopting AI technology in financial contexts. This relationship arises from the alignment between maximizers’ preference for thoroughly evaluating all possible options and the data-driven and comprehensive decision-making approach offered by algorithms (Dietvorst and Bartels, 2022). By broadening the scope of algorithm aversion research to include decision-making styles, our study highlights maximizing as a constructive factor in reducing algorithm aversion. Additionally, addressing calls for further research into the positive aspects of maximizing (Misuraca et al., 2021), we advance the understanding of how this decision-making style can lead to greater acceptance and use of AI advisors in financial decision-making. Our research illustrates a novel pathway for mitigating algorithm aversion and expanding the theoretical perspective on AI adoption in consumer financial decision-making.

Finally, our paper contributes to the literature on algorithm aversion by uncovering novel pathways to mitigate this phenomenon. While most studies have concentrated on individual characteristics (Mahmud et al., 2022), our research broadens the scope by investigating decision-making styles. For instance, Logg et al. (2019) showed that egocentric bias and self-esteem cause individuals to favor their own decisions over algorithmic ones. Similarly, Araujo et al. (2020) found that individuals with higher online self-efficacy are more confident in using algorithms. Moreover, Sharan and Romano (2020) highlighted that neuroticism contributes to algorithm aversion due to perceived untrustworthiness and anxiety about technology, a view supported by Meuter et al. (2003) and van Esch et al. (2021). Finally, Khan et al. (2024) showed that adopting a promotion-focused approach can alleviate algorithm aversion. Building on this work, our study highlights maximizing tendency as a decision-making style that facilitates greater acceptance of AI technologies. Further, we delve into the psychological mechanisms underlying this relationship through serial mediation by perceived effectiveness and algorithm aversion, addressing Castelo et al.'s (2019) call for research to investigate non-performance-related factors in algorithm use.

From a practical perspective, our study provides valuable insights for business practitioners on strategies to mitigate individuals’ aversion to algorithms and increase their likelihood of seeking advice from AI advisors. Financial institutions, consultants, and service providers can play a key role in implementing these strategies to mitigate algorithm aversion and encourage the use of AI advisors for better financial decision-making (Northey et al., 2022). For instance, financial consultants could incorporate maximizing prompts during client consultations, encouraging clients to conduct thorough searches for the best solution. Financial services could integrate these strategies into advertising materials or client onboarding with messaging such as “maximize your financial potential with AI guidance,” or “AI-powered advice to help you make the best financial decisions”, framing AI tools as enhancing personalized and efficient decision-making (Ding et al., 2023; Ma and Roese, 2014). By guiding clients through detailed decision-making processes and framing AI advisors as tools to maximize outcomes, practitioners can reduce algorithm aversion and increase engagement with AI technologies (Kawaguchi, 2021; Mende et al., 2019). Financial service providers, specifically, could highlight the personalized and cost-effective nature of AI services (Nicholas, 2022) with phrases like “tailored to your unique personal needs” to encourage acceptance. These improvements, particularly relevant for individuals with lower incomes and net worth (Fulk et al., 2018), likely lead to better financial outcomes, as AI-driven advisors provide affordable, expert financial guidance, broadening access to high-quality financial planning. This approach not only fosters greater acceptance of AI-driven solutions, but can also enhance the quality of clients’ financial decisions (Back et al., 2023; D’Acunto et al., 2019).

Furthermore, the scalability of AI tools provides financial institutions such as banks with the opportunity to extend their reach to underserved and rural areas, promoting greater financial inclusion and accessibility (Gokhale et al., 2019; Niederberger, 2024). By adopting these strategies, companies can gain a competitive edge by offering AI-based advisors as a viable, cost-effective alternative to human advisors, which should ultimately lead to better financial outcomes for clients and higher overall customer satisfaction, thereby likely increasing customer loyalty and firm profitability (Blut et al., 2021; Huang and Li, 2024; Levitt, 2024; Shanmuganathan, 2020). Implementing these approaches will help financial institutions harness the full potential of AI, thereby reinforcing their position in an increasingly competitive financial marketplace.

Despite its contributions, our study has some limitations that could inform future research. First, our findings are based on data from the United States, and thus reflect its individualistic culture and technological landscape, which might limit the generalizability to other WEIRD (Western, Educated, Industrialized, Rich, Democratic) nations, since cultural characteristics significantly influence technology adoption (Belanche et al., 2020; Chi et al., 2023). In individualistic cultures like the US, individuals may prefer making decisions independently, viewing AI as undermining autonomy (Gillespie et al., 2023). In contrast, in collectivistic cultures like Japan and South Korea, which emphasize community well-being, individuals may see AI as a tool to enhance collective benefits through increased efficiency and productivity (Cui, 2022). Furthermore, maximizers in individualistic cultures (e.g. the US, Canada) may resist AI to maintain control, while those in collectivistic cultures (e.g. China, India) may embrace AI to enhance outcomes for the group (Sheth et al., 2022). Since our study does not account for these cultural differences, future research could examine how maximizing tendencies shape algorithm aversion and AI usage across cultural contexts (Mogaji et al., 2022).

Second, we focus on investigating maximizers’ likelihood of using an AI advisor in the financial decision domain, with a particular focus on investing for one’s retirement. Although individuals generally exhibit aversion to employing algorithms across various decision domains, this tendency and its underlying mechanism may depend on the nature of the task and situation. For instance, there is a notable reluctance to use algorithms for subjective tasks (Castelo et al., 2019), hedonic decisions (Longoni and Cian, 2022), and decisions involving morally relevant trade-offs (Dietvorst and Bartels, 2022). Future research could thus explore the relationship between a maximizing decision-making style and the tendency to use an AI advisor in different financial decision contexts that involve some of these factors, such as sustainable investing, making financial choices for someone else, or choosing insurance plans.

Despite these limitations, this paper provides valuable insights on maximizing and algorithm aversion, particularly examining their influence on the likelihood of using an AI advisor in a financial context. Significant advancements in AI technology and the widespread adoption of AI-based tools across various domains have led to substantial positive transformations in human life and improved individual decision-making processes. While some individuals have become increasingly familiar with and reliant on AI agents for various daily tasks (Logg et al., 2019), others continue to exhibit aversion towards algorithms (Dietvorst et al., 2015). Since algorithms frequently outperform human capabilities (Elkins et al., 2013), such algorithm aversion may hamper individuals from enhancing their decision-making processes and leveraging the advantages that AI can provide (Ameen et al., 2023; Dwivedi et al., 2023). Our study is among the first to demonstrate that situationally inducing maximizing enhances the perceived effectiveness of algorithms, which reduces individuals’ algorithm aversion, and subsequently increases their propensity to utilize AI financial advisors.