Uso de chatbots na produção de resumos científicos: o que os resumos nos dizem da investigação? - uma revisão sistemática

Scientific abstracts are abbreviated, concise, and faithful versions of the intellectual content of an original document. They can have an explicit or implicit structure and have functions such as alerting, anticipating, substituting, and selecting information. Traditionally, they are produced by human operators, but with technological advancements, automated systems have also been used, especially since the second half of the 20th century. Initially, these automated systems were based on word frequency and position, but they have evolved with artificial intelligence. Chatbots, like Chat GPT, launched in 2022, have revolutionized text production, including scientific abstracts, by generating realistic content based on large volumes of pre-existing text. However, there are ethical concerns about authorship and the scientific accuracy of these abstracts since AI cannot be held responsible for the generated content. This study aims to analyse the use of AI tools in constructing scientific abstracts through article summaries, evaluating themes, contexts, methods, and conclusions of recent studies. Methodologically, a systematic literature review was conducted using the Web of Science, Scopus, and Ebsco Discovery Service databases, focusing on articles from the last three years. The existing studies, whose summaries were analyzed, demonstrate a significant need for further study and reflection on this topic. There is a noted difficulty in distinguishing AI-generated abstracts from human ones and controversy regarding the quality and ethics of using AI. Despite technological improvements, human validation remains crucial to ensure the accuracy and quality of abstracts, reinforcing the importance of human oversight and adaptation of AI-generated content.