Modele językowe sztucznej inteligencji w relacjach komputer człowiek z perspektywy analizy transakcyjnej

Abstract

The end of 2022 and the year 2023 saw the release and widespread adoption of publicly available, advanced artificial intelligence (AI) language models. These models can not only hold conversations with users, but also search and analyze information, translate text, create graphics, write essays, and more. The possible applications are growing, and the models are constantly evolving and improving. As a result, the issue of human-computer interaction has entered a new, previously unknown, level. Now, it is possible to engage in true interaction with a tool. It can also be said that in a conversation, it will exhibit certain traits that could be classified as personality traits.

In this paper, several AI language models were invited to participate in an experiment involving the completion of an egogram, a tool used in transactional analysis to determine the profile of the subject's ego states at the level of functional analysis. The results obtained indicate that currently available AI can be persuaded to participate in such an experiment, can answer all of the egogram statements, and these are not random answers. Based on the results, it is possible to determine the AI's ego state profile.