Modelagem do uso da tecnologia ChatGPT: moralidade, surpresas e singularidades

• Ano de defesa: 2024
• Autor(a) principal: Marson, Celise
• Orientador(a): Urdan, André Torres
• Banca de defesa: Urdan, André Torres , Vils, Leonardo , Veiga, Ricardo Teixeira , Cunha, Júlio Araujo Carneiro da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Nove de Julho
• Programa de Pós-Graduação: Programa de Pós-Graduação em Administração
• Departamento: Administração
• País: Brasil
• Palavras-chave em Português: ChatGPT; inteligência artificial; Chatbot; processamento de linguagem natural; modelo TAM; teoria IDT; teoria DMC
• Palavras-chave em Inglês: GPT Chat; artificial intelligence.; Chatbot; natural language processing; TAM model; IDT theory; DMC theory
• Área do conhecimento CNPq: CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO
• Link de acesso: http://bibliotecatede.uninove.br/handle/tede/3543

ChatGPT and similar services are the subject of intense debate worldwide, potentially serving as both a solution and a problem. In this dissertation, the research question is: What are the determinants of ChatGPT usage? The objective is to identify these determinants in higher education. The expectation is to contribute to teaching and learning, making educational activities more effective when involving ChatGPT. In the literature, I selected two technology usage models (Technology Acceptance Model, TAM; Diffusion of Innovation Theory, IDT) and one cognitive moral development model (DMC). I conducted an online survey with ChatGPT users among students in Brazil. In the first phase, I tested TAM and IDT using sample data and multiple linear regression. The variables UPgeral (perceived usefulness, from TAM) and COMgeral (compatibility, from IDT) have a positive effect on Weekly ChatGPT Usage. In the second phase, also using regression, I tested DMC as a precursor to usage with the same dataset, but no variable explained Weekly ChatGPT Usage. In the third phase, still using regression and the same sample, I tested TAM, IDT, and DMC together as precursors to usage. The UPgeral variable (perceived usefulness) has a positive and significant effect on Weekly ChatGPT Usage. Moreover, through Necessary Condition Analysis, ease of use and perceived usefulness are the primary conditions for Weekly ChatGPT Usage. Compatibility, complexity, observability, and testability are also necessary. Therefore, the sample lacks empirical support for the proposed effects of integrating TAM, IDT, and DMC. Thus, other concepts determine ChatGPT usage. For this reason, in the fourth phase, through exploratory qualitative research, I identified potential precursors to usage not covered in the previous phases, as well as evidence of the effects of the determinants investigated in phases one through three. I conducted interviews with specialists to explore perspectives, challenges, aspirations, and expectations regarding ChatGPT. The results highlighted the following determinants: accessibility, connectivity access, trust in technology, creativity, entertainment, expectations, prior experience, feedback and continuous improvement, perceived innovation, integration with existing systems, time optimization, personalization, workload reduction, perceived risk, satisfaction, and security. The interviews confirmed the presence of the quantitative variables, reinforcing and complementing the research findings. Therefore, it is necessary to complement the adopted models by incorporating variables that better capture the particularities and complexities of ChatGPT. Improving the understanding of ChatGPT is crucial for formulating strategies in the educational context (in particular) and in society (in general) in light of ChatGPT.