PhD Sample Proposal: The impact of continuous use of generative artificial intelligence models on human cognitive behaviour

Background

The recent years have witnessed multiple applications of technology in different sectors. With artificial intelligence (AI) being at the technological forefront, it has now become an integral part of modern society. While earlier the use of AI started as a simple smartphone assistant, today its applications can be seen in complex decision-making systems (Dwivedi et al., 2021). AI tools are now commonly used for tasks such as content creation and coding. Due to the nature of these tasks, they once needed complete human intervention in the form of problem-solving and decision-making skills. However, with the introduction of generative AI models, these tasks can be completed with the help of technology (Kim et al., 2022).

Generative AI models, such as ChatGPT, DALL-E, Bing Chat and others, are designed to produce human-like responses (Zhao, 2023). Based on user prompts, these tools can generate both short and long-form responses. Apart from generating coherent text results, these tools are also trained to create images or solve complex problems in seconds. The revolutionary capabilities of AI have made it an invaluable asset in many fields. However, the ease with which these systems can produce high-quality results raises questions about their growing use in day-to-day applications. The widespread reliance on AI-powered tools may have potential effects on human cognitive abilities over time (Sun & Medaglia 2019). As AI systems become more refined and widely available, there is a growing need to understand their impact on human cognitive behaviour.

In the past, several studies have explored how AI-powered tools influence human cognition. For example, Marikyan et al. (2023) examined the impact of smart home technology on cognition. Similarly, Sachdeva et al. (2024) conducted a study to understand the impact of AI-powered tools on the cognition of employees working in the banking sector. However, generative AI, which is the latest addition to AI technology, represents a new relationship between human and computer interaction. As compared to other applications of AI, generative AI not only helps users in collecting information but also assists them in various creative and analytical processes that are traditionally performed by the human mind.

As the use of generative AI tools is becoming more integrated in educational, professional and personal settings, it is important to understand their long-term effects on human cognitive abilities. This research aims to investigate how continuous use of generative AI can influence different aspects of human cognition, such as creativity, critical thinking, problem-solving skills and memory function.

Problem Statement

The growing use of generative AI tools is known to have both positive and negative effects on human cognition. On the one hand, it eases the complexities involved in challenging tasks, on the other, it reduces the need to use human problem-solving skills in a difficult environment. Although generative AI has the potential to offer exceptional support for various cognitive tasks, there is a growing concern that depending on these tools for the long term may lead to the reduction of cognitive skills. Thus, this research makes an attempt to understand the comprehensive nature of generative AI and its potential impact on human cognitive processes in the coming years. Due to the comprehensive nature of the research area, the current research will focus on establishing a connection between the use of generative AI and different cognitive skills. The study will emphasise key cognitive skills such as creative thinking, problem solving, critical thinking, analytical understanding, memory function and information retention.

Aims and Objectives

The primary aim of this research is to investigate the impact of continuous use of generative AI tools on different aspects of human cognition. To achieve this aim, the study will make an attempt to get a detailed insight into the following objectives:

• Measure the impact of frequency of generative AI use on creative thinking and problem-solving abilities.
• Examine the growing dependence on AI-generated content and its effect on critical thinking and analytical skills.
• Evaluate the potential effects of generative AI use on memory function and information retention in users.

Preliminary Literature Review

Decades ago, the global marketplace witnessed a significant industrial revolution that led to the transformation of various manual tasks and processes that were beyond the physical capabilities of human beings. Today, AI holds a similar transformative potential that has the power to replace human contributions through augmentation in different industrial, intellectual and social areas (Dwivedi et al., 2021). The rapid progression of AI technologies presents both opportunities and challenges for human cognition. For instance, AI has proven to be useful in augmenting human intelligence by providing tools that can enhance problem-solving capabilities and information processing. However, there is growing concern among experts regarding the potential side effects of continuous use of AI on human cognitive behaviour. Some of the common areas where AI can have a significant impact on human cognitive behaviour include decreased attention spans, altered decision-making processes and changes in social cognition. As people of all age groups are using AI-powered tools daily, it is essential to understand how the use of AI impacts their cognitive behaviour.

Relationship between technology and different cognitive behaviours

For example, Risko and Gilbert (2016) noted that people use technical devices such as phones and laptops to use the internet, stay connected or play games. When powered by AI, these devices provide an ideal platform for users to express their cognitive processes. This is also known as cognitive offloading, where people rely on different technical applications instead of using their own spatial abilities (Finley et al., 2018). This means instead of memorising phone numbers, birthdates or shopping lists, people feed information on their devices and refer to it later. Finley et al. (2018) also noted that based on their preferences, people distribute cognitive demands between internal and external resources for problem-solving tasks. While using technology for these tasks decreases the rate of information retention, it makes it easier for people to recall and find information when needed. As a result, technology can act as an external memory system that potentially alters the way we process and store information.

However, Grinschgl and Neubauer (2022) argued that when it comes to cognitive offloading, there can be both advantages and disadvantages with the constant use of AI-powered technology. While it can make it easier for users to perform a task, constant use of technology may have detrimental long-term effects on human cognition. To understand the relationship between AI-influenced cognitive offloading and memory retention, Grinschgl et al. (2021) conducted multiple experiments on a similar group of participants using a pattern copy task. The study concluded that although cognitive offloading accelerates task processing, it may interfere with the formation of memory for the processed information. As a result, users may find it difficult to recall the offloaded information or may demonstrate signs of confusion in explicit learning conditions. This shows that prolonged use of generative AI may have an impact on the attention span of users, which may further result in reduced capacity for focused thinking. Additionally, Dalilian & Nembhard (2024) noted that the outcome of AI systems is not always accurate. Thus, completely depending on generative AI models for complex tasks may lead to inaccurate results.

Cognitive Dissonance Theory

The above studies form a foundation for understanding the impact of prolonged use of generative AI on human cognition. Linking these studies to cognitive dissonance theory can provide a better perspective on the relationship between generative AI and cognitive behaviour. In 1962, Festinger suggested the cognitive dissonance theory. This theory indicates that a negative disconfirmation may begin the reduction of alleged inconsistency between expectation and performance, which may potentially lead to satisfaction (Harmon-Jones & Mills, 2019). The cognitive dissonance theory can be further used to evaluate the behaviour of users before and after using generative AI models. Moreover, the theory also suggests that a state of dissonance can arise when an individual experiences two or more contradictory cognitions.

Furthermore, Cancino Montecinos (2020) noted that cognitive dissonance can be an outcome of both positive and negative emotions in an individual. As a result, whenever an individual experiences cognitive conflict, an aversive feeling may arise, irrespective of the type of situation and consequences for the individual. This feeling can motivate the individual to either feel more focused or experience reduced attention while dealing with a cognitive conflict. Since generative AI uses large language models to produce outcomes based on the user’s inputs, using these models frequently can increase the dependence of individuals on technology. This can lead to lowered productivity when individuals are expected to work without technical assistance and may be associated with negative emotions or discomfort. Based on the cognitive dissonance theory, users may take measures that can lead to a change in their attitude or behaviour to reduce dissonance arising due to the inability to use generative AI.

Proposed Methodology

To investigate the impact of continuous use of generative AI on human cognition, this study will employ a mixed-methods approach. It will combine a survey methodology with structural equation modelling (SEM). This approach will allow for the collection of both quantitative and qualitative data, providing a comprehensive understanding of the complex relationships between AI use and cognitive processes. A similar model has been implemented by Shin (2016) and Tolbert and Drogos (2019) to examine the structural relationships between different cognitive variables and the use of social media platforms. However, there is not enough evidence regarding the use of the SEM approach for evaluating the impact of generative AI on the cognitive variables of individuals.

For the purpose of this study, a survey will be conducted on a selected number of participants. The study will recruit a diverse sample of 300 participants aged 18-65. The participants will represent various educational backgrounds, professions and levels of experience with generative AI tools. These participants will be recruited through online platforms, universities and professional networks. The participants will further be divided based on their age, sex and demographics. Based on a comprehensive questionnaire, the participants will be surveyed to evaluate the four main constructs of the study that include creativity, critical thinking, problem-solving and memory retention. Additionally, the participants will have to provide data about the frequency and duration of generative AI use, along with the list of tools used.

The survey will include both closed-ended questions and open-ended questions to understand the experiences of different participants with generative AI. A 5-point Likert scale will be used to evaluate the rating of participants based on their use of generative AI models. On the other hand, responses to open-ended questions will be evaluated using thematic analysis to identify recurring themes. In addition to the survey, the participants will also be encouraged to take cognitive assessment tests. These tests will include different questions to measure the critical thinking and problem-solving skills of the participants. Furthermore, a mediation analysis will be used to measure the validity of different variables used in the study. Before using the SEM approach, a confirmatory factor analysis will be conducted to evaluate the validity and reliability of the measurement instruments. The SEM approach will further assist in examining the direct and indirect effects between different variables.

Research Contributions

This study aims to make several significant contributions to the field of psychology. By employing the SEM approach, this research will provide a comprehensive model to study the relationships between generative AI use and different human cognitive processes. It will offer a better understanding of both direct and indirect effects of continuous use of generative AI on cognitive behaviour. Due to the lack of research in this area, the study can act as a foundation for future work. However, in its preliminary stage, the research will focus on examining the key factors influencing cognitive behaviour with the continuous use of generative AI.

In the future, it has the potential to take into consideration the role of personal characteristics and individual strengths and weaknesses to get enhanced results. Recruiting participants from different professional and educational settings can play a key role in evaluating the use of generative AI in different contexts. Based on the results obtained, the study can help in maximising the benefits of generative AI use while mitigating its potential side effects on cognitive behaviour.

References

Cancino Montecinos, S. (2020). New perspectives on cognitive dissonance theory, Doctoral dissertation, Department of Psychology, Stockholm University, http://su.diva-portal.org/smash/get/diva2:1411016/FULLTEXT01.pdf

Dalilian, F., & Nembhard, D. (2024). Cognitive and behavioral markers for human detection error in AI-assisted bridge inspection. Applied ergonomics, 121, 104346, https://doi.org/10.1016/j.apergo.2024.104346

Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T., & Williams, M. D. (2021). Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy. International journal of information management, 57, 101994, https://doi.org/10.1016/j.ijinfomgt.2019.08.002

Finley, J. R., Naaz, F., & Goh, F. W. (2018). Memory and technology. Cham: Springer International Publishing, http://dx.doi.org/10.1007/978-3-319-99169-6

Grinschgl, S., & Neubauer, A. C. (2022). Supporting cognition with modern technology: Distributed cognition today and in an AI-enhanced future. Frontiers in Artificial Intelligence, 5, 908261, https://doi.org/10.3389/frai.2022.908261

Grinschgl, S., Papenmeier, F., & Meyerhoff, H. S. (2021). Consequences of cognitive offloading: Boosting performance but diminishing memory. Quarterly Journal of Experimental Psychology, 74(9), 1477-1496, https://doi.org/10.1177/17470218211008060

Harmon-Jones, E., & Mills, J. (2019). An introduction to cognitive dissonance theory and an overview of current perspectives on the theory. In E. Harmon-Jones (Ed.), Cognitive dissonance: Reexamining a pivotal theory in psychology (2nd ed., pp. 3–24). American Psychological Association, https://doi.org/10.1037/0000135-001

Kim, Y., Kang, S., Nam, Y., & Skalicky, S. (2022). Peer interaction, writing proficiency, and the quality of collaborative digital multimodal composing task: Comparing guided and unguided planning. System, 106, 102722, https://doi.org/10.1016/j.system.2022.102722

Marikyan, D., Papagiannidis, S., & Alamanos, E. (2023). Cognitive dissonance in technology adoption: A study of smart home users. Information Systems Frontiers, 25(3), 1101-1123, https://doi.org/10.1007/s10796-020-10042-3

Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in cognitive sciences, 20(9), 676-688, https://doi.org/10.1016/j.tics.2016.07.002

Sachdeva, C., Gangwar, V. P., Grover, V., & Gochhait, S. (2024, January). Cognitive Dissonance in Banking Employees: Exploring Factors Amid the Artificial Intelligence Revolution. In 2024 ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems, 1731-1735 https://doi.org/10.1109/ICETSIS61505.2024.10459558

Shin, D. H. (2016). Do users experience real sociability through social TV? Analyzing parasocial behavior in relation to social TV. Journal of Broadcasting & Electronic Media, 60(1), 140-159, http://dx.doi.org/10.1080/08838151.2015.1127247

Sun, T. Q., & Medaglia, R. (2019). Mapping the challenges of Artificial Intelligence in the public sector: Evidence from public healthcare. Government Information Quarterly, 36(2), 368-383, https://doi.org/10.1016/j.giq.2018.09.008

Tolbert, A. N., & Drogos, K. L. (2019). Tweens’ wishful identification and parasocial relationships with YouTubers. Frontiers in psychology, 10, 2781, https://doi.org/10.1016/j.giq.2018.09.008

Zhao, X. (2023). Leveraging artificial intelligence (AI) technology for English writing: Introducing wordtune as a digital writing assistant for EFL writers. RELC Journal, 54(3), 890-894, https://doi.org/10.1177/00336882221094089