Abstract

This dissertation proposes a new metaontological model as a way to overcome the most relevant and challenging ethical problems of today. It argues that there is a teleological dissonance between what is expected of an AI and what it can offer, constructively. It proposes that such dissonance arises from a philosophical tradition in which there is a predilection for extracting a fragment of reality, to the detriment of valuing the analysis of complexity itself, as given. Thus, what results from this whole movement is the liminality of the current technological era, in which we are going through a period of complex transformation and potentiality, when much of what was stable and predictable is becoming the opposite: unstable and unpredictable. The first part of the dissertation addresses the path from the past to the present. The second part considers what the future of AI will be like, if taken from the perspective of overcoming current problems, without configuring them as fictional projections. We will make such projections based on existing technological resources or others that are considered viable and feasible to be used in algorithmic constructs. And finally, the third part deals precisely with liminality, this present that refers not only to temporality, but also to spatiality - in terms of limits and scope The conclusion argues that an ethical and responsible AI, free from current problems, requires a deep understanding of human complexity - and the respective representation of the existential modal - so that it is possible to apply the most appropriate values in all stages of technological development. The final consideration, which summarises the entire argument, is about the elaboration of a dynamic metaontological protocol, capable of capturing and emulating the relational complexity of reality. This feat is presented as necessary, but perhaps not sufficient, for the creation of this AI of the future that will surpass current demands.