This research project aims to explore the dynamics of AI adoption within organizational contexts, focusing on the interplay of context and time in shaping users' attitudes and behaviors towards generalpurpose (GP) AI technologies. Through a comprehensive investigation involving empirical studies and an extensive review of existing literature, our objective is to formulate a GPAI Endurance Theory capturing sustained interactions with GPAI across various contexts and extended timeframes, transcending the limitations of conventional adoption theories' static nature. By delving into the recursive impact of previous AI usage on subsequent adoption, analyzing contextual factors, and considering temporal variations, we seek to provide organizations with valuable insights to facilitate the successful integration and long-term acceptance of GPAI-based components within their information systems (IS). This understanding will ultimately contribute to a deeper understanding of human-AI interactions in organizational settings and inform the future design and development of GPAI-based IS.
(text, images, audio) based on its training data and user interactions [
The exploration of individual technology acceptance is not a new endeavor; researchers have
already delved into this, producing well established models like the Technology Acceptance
Model (TAM) [
The underlying principles of GPAI technologies, such as GenAI, indeed revolve around their capacity to address a wide range of tasks across various contexts. As inherently versatile systems, they can be applied repeatedly to tackle diferent challenges. This characteristic necessitates a nuanced approach to studying the adoption and usage of such technologies, taking into account the dimensions of context and time. The adoption and usage behavior of GPAI technologies may exhibit significant variability across diferent contexts and over time through factors such as fatigue efects or fluctuations in learning curves, to cite only them.
Put together, these new dimensions suggest much more than an increment on existing adoption models and will bring us to the formalization of a new paradigm for intelligent technology adoption. Instead of adoption, we plan to investigate the introduction of a novel, more suited, concept called "AI endurance", which will refer to the ability of human users to interact with and utilize general-purpose intelligent technologies over extended periods and across various contexts efectively. It will encompass traditional aspects of technology adoption such as user satisfaction, engagement, and usability over time but would also consider how users’ experiences and interactions with GPAI technologies evolve over prolonged usage, including factors such as user fatigue, cognitive load, learning curve, and adaptation to evolving GPAI capabilities or interfaces. In essence, it will reflect the sustainability of users’ engagement and satisfaction with AI systems throughout their usage journey. In addition to achieving a clear understanding of workers’ usage behaviors, acceptance, and adoption of GPAI technologies, this research is also aimed at informing the design of GPAI-supported IS. It will achieve this by delineating a set of requirements for the design and engineering of GPAI-supported IS, grounded on knowledge about users’ behavior and attitude, sustaining their usability across various contexts and over time. These requirements will subsequently undergo a validation phase to ensure their eficacy and relevance.
Numerous studies have investigated technology acceptance and adoption models, with the
Technology Acceptance Model (TAM) serving as a foundational precursor in this domain. The
TAM posits that actual system usage hinges on behavioral intention, influenced by perceived
usefulness and ease of use [
In response to these limitations, AI-specific models were developed, grounded in their
predecessors [
As highlighted, AI technologies are unique – diferentiating from conventional technologies by their intelligent capabilities. It is consequently inadequate to treat the acceptance of AI technologies with the same frameworks applied to traditional technologies. To address this, researchers have endeavored to develop AI-specific adoption, acceptance, and aversion models and theories, often building upon existing frameworks such as TAM or UTAUT. However, a common trait among these models is their context specificity, tailored to measure AI acceptance within particular contexts. Yet, as AI technologies continue to evolve into more generalist forms, capable of application across diverse contexts such as GenAI, the challenge arises in assessing acceptance and usage within varied environments. This highlights the inherent dependency of GPAI acceptance on usage context, with acceptance and adoption within one context potentially influencing adoption in another, thus shaping the overall adoption landscape of GPAI technologies.
Another noteworthy observation emerges from our preliminary investigations: as AI tends to be more generalist and used within various contexts, it is reasonable to assume that the adoption of a GPAI system may exhibit a recursive behavior. This phenomenon suggests that adoption and usage at time may influence further adoption at time + 1, potentially explaining many decisions regarding the utilization of GPAI technology. Specifically, technology adoption, including AI technologies, is expected to display temporal variations characterized by phenomena such as fatigue, learning curves, and the potential influence of past negative experiences with one technology on the adoption of others. These aspects are applicable to GPAI systems as well. However, a distinctive characteristic is that GPAI technologies imply the prolonged use of the same system to address a variety of tasks across diverse contexts. This raises fundamental questions: what is the duration of these efects? How long do negative experiences with GPAI technologies persist in impacting current adoption? Moreover, how do factors like fatigue and learning curves influence users’ willingness to engage with or attitudes toward GPAI technologies? In essence, our inquiry seeks to understand the overarching impact of time on GPAI adoption dynamics.
By integrating considerations related to context and time, our objective is to elevate the measurement of AI adoption and acceptance to a higher level than previously achievable. This entails moving beyond the limitations of context-specific assessments and accounting for past experiences with AI technologies. Consequently, we aim to develop a novel adoption theory, rooted in existing technology acceptance and aversion frameworks, capable of assessing overall GPAI adoption and use across varying contexts and over time. Termed the "GPAI Endurance Theory", this framework represents a significant advancement in understanding the sustained engagement with and acceptance of AI technologies.
To summarize, this thesis project aims to address three key research questions concerning the adoption of AI technology incorporated within information systems: • RQ1: What is the role played by the usage context in defining a user’s overall GPAItechnology adoption? • RQ2: How do overall GPAI-technology adoption evolve through time? • RQ3: How to measure an overall GPAI-technology adoption – called "GPAI Endurance" – considering the impact of context and time?
The primary contribution anticipated from this whole research endeavor is the development of a GPAI Endurance Theory, which aims to elucidate how users respond to prolonged usage of GPAI technologies across diferent contexts. This theory will be instrumental in understanding the dynamics of user engagement with GPAI systems over time, considering contextual, temporal, and other socio-technical factors.
However, to realize this overarching goal, several intermediary contributions are envisioned. These contributions are essential steps in the progression towards the formulation of the GPAI Endurance Theory.
Our research will initially delve into the impact of context in shaping technology adoption, with a specific focus on GPAI technology adoption. A critical observation arising from this exploration is the absence of a formal theorization regarding the usage context for AI technology in existing literature. This gap hinders our ability to accurately gauge the influence of context on technology adoption. Therefore, our first research objective is to clearly and formally define what "usage context" means and its constituting factors in the context of GPAI technology adoption. We aim to create a precise definition to help evaluate how this concept influences users’ overall behavior in adopting and using GPAI technology. Achieving this will allow us to empirically investigate the research question RQ1: What impact does context have on GPAI adoption and use? Defining "usage context" and understanding its impact on GPAI adoption and usage is our first key contribution.
The second intermediate contribution of this research involves investigating the influence of time on GPAI technology adoption and usage patterns, that is, answering RQ2. Our plan is to delve into the long-term coexistence between workers and GPAI-based system components to comprehend how time impacts GPAI adoption. This investigation will entail analyzing the behavior of various workers across diferent organizations who will be asked to use a GPAIsystem over an extended period. The anticipated outcome of this research is to discern any shifts in attitude towards the usage of GPAI systems, pinpoint the reasons behind these shifts, and determine the duration of their impact on GPAI adoption. This endeavor aims to provide a comprehensive understanding of how time afects GPAI adoption, including an analysis of the duration of adverse experiences with the GPAI systems and their implications for future adoption of the same system. Additionally, we aim to explore whether behavioral phenomena such as fatigue or cognitive overload also influence GPAI system adoption and to what extent.
The outcomes of these preliminary investigations will lead to a deeper understanding of the underlying dynamics of GPAI experiences and future GPAI adoption, as well as the enduring efects of GPAI system usage on adoption and use over time. These foundational insights are crucial for the development of the third and most significant contribution – the GPAI Endurance Theory. This theory represents the apex of the project, aiming to establish a new paradigm for AI usage within organizations’ IS. Drawing upon answers to RQ1 and RQ2, the GPAI Endurance Theory seeks to ofer a comprehensive and innovative perspective on human-AI interactions within organizational contexts. By synthesizing insights from existing technology adoption models and empirical observations, the GPAI Endurance theory aims to provide a holistic understanding of the enduring dynamics of AI adoption.
The major contribution of this theory will be a model capable of assessing the level of GPAI endurance of a person or a group of people within an organization, based on captured contextual factors, knowledge on enduring efects, and other socio-technical factors shown to be relevant by existing acceptance models. This theory and its model will answer RQ3.
Another anticipated contribution pertains to design science. By addressing RQ1–3, we will gather valuable insights into user reactions towards various characteristics of GPAI systems or components. This will enable us to compile a list of potential requirements for a successful and widely adopted GPAI-based IS. Following the completion of this requirements list, a validation phase will ensue. During this phase, professional developers will assess the feasibility of the requirements, while business users will evaluate their relevance, completeness, and adequacy. This constitutes the "engineering" aspect of the thesis, aimed at guiding the further development of future successful GPAI-based IS.
As demonstrated, these contributions are primarily aimed at assisting organizations in developing informed and sustainable GPAI-based IS components by gaining a better understanding of the attitudes of their workforce. By identifying potential levers to increase the endurance of GPAI adoption and usage, organizations can ensure long-term success for their innovations and foster a successful coexistence between their workforce and GPAI-based components.
To ensure the success of this research project, we intend to employ a mixed-methods approach for data collection, analysis, and knowledge-building, encompassing both empirical and theoretical data.
For each contribution, we will start with a thorough literature review aimed at examining
existing research and establishing the groundwork for our studies. This literature review
phase will primarily involve an in-depth exploration of existing literature on acceptance and
aversion theories, including TAM [
Following the literature review, we will conduct several empirical studies to investigate the phenomena of interest.
First, we aim to identify and formalize the components of a comprehensive usage context that
may influence GPAI adoption and understand how these factors impact the adoption and usage
behavior of GPAI-technologies by workers. To achieve this, a methodological triangulation
approach will be utilized, as outlined by [
The second phase of the research project will focus on understanding the dynamic efect of
time on generalist AI technology adoption. A longitudinal study spanning two years will be
conducted, following the methodology proposed by [
Finally, the GPAI Endurance Theory will be modeled based on the findings from the initial
studies and existing theories using Structural Equation Modeling (SEM), with a specific focus
on Confirmatory Factor Analysis (CFA). SEM is chosen for its ability to measure and model the
impact of latent constructs such as time and context, as it allows for the integration of multiple
variables and their relationships [
As previously mentioned, an additional anticipated contribution involves compiling a list of requirements for future GPAI-based IS. These requirements will be derived from the interviews and focus groups conducted across various phases of the research. Eventually, we intend to present the list of requirements to both business experts and IS developers to evaluate their representativeness, validity, and feasibility, representing a validation of these requirements which may in turn lead, in future research perspectives, to the development of a protoype of GPAI-based IS.
In conclusion, the research project will employ a combination of observational, qualitative, and quantitative methodologies to gather insights and develop the GPAI Endurance Theory, providing a comprehensive understanding of the factors influencing generalist AI adoption across usage contexts and over time and a list of requirements for desirable and successful GPAI-based IS.
In conclusion, this study aims to explore the behavior of information systems users in adopting GPAI technologies. We will investigate the impact of two key variables on the use of these technologies: context (the conditions under which the technology is used) and time.
Our research project will first delve into the role of context in adopting GPAI technology. We plan to conduct a study to clearly define the concept of “context” in the context of AI technology usage. This will involve employing a triangulation methodology to gather and analyze observational, qualitative, and quantitative data from interviews and surveys.
Subsequently, we will examine the role of time in the adoption and usage of GPAI technologies. This longitudinal study will span over two years, aiming to capture potential fatigue efects, the impact of the learning curve, and the influence of past negative or positive experiences – among others – on the decision to use, adopt, and reuse GPAI technologies.
These investigations will pave the way for the third phase of this research project: deriving a GPAI Endurance Theory. This theory seeks to formalize the impact of context and time on the adoption and usage of GPAI technologies, drawing on insights from the earlier studies and existing literature. The ultimate goal is to develop a GPAI Endurance Model using Structural Equation Modeling, particularly Confirmatory Factor Analysis.
We will finalize this research by compiling and validating a list of requirements extracted from the preceding studies. These requirements will be critical in guiding the informed design and engineering of GPAI-based IS.
Overall, the aim of this research project is to comprehensively capture and formalize the inlfuence of contextual and temporal factors, along with other socio-technical factors, to elucidate the adoption and sustained use of GPAI-based technologies across information systems by an organization’s workers. This understanding will enable organizations to undertake consistent and successful GPAI-based projects with due consideration for their workforce.
I would like to thank the supervisor of this project, Prof. Corentin Burnay, for his support and reviews on this paper.