Overcoming the risks of using neural networks for text generation: cognitive, ethical, pedagogical aspects
E. V. Biricheva
Institute for Philosophy and Law Ural Branch Russian Academy of Sciences, Yekaterinburg, Russia
Keywords: neural network, text generation, chatbot, artificial intelligence, AI, ethical issues, pedagogical issues, digitalization, purpose-oriented approach
Abstract
Introduction. The using of contemporary artificial intelligent systems (AIS), which has penetrated into diverse professional and everyday practices, in addition to positive effects, may also bring forth negative consequences. The spheres of science and education seem to deserve special attention, since the uncontrolled use of generative neural networks may carry risks for the creative competencies development and give rise to a number of ethical problems. In this regard, the article aims at the systematization of challenges posed by the use of neural networks for text generation in science and education, as well as the philosophical understanding of the grounds for overcoming different cognitive, ethical and pedagogical risks. The methodology concentrates on a phenomenological analysis of the ontological status of the contemporary AIS, possible problems in the field of development of human cognitive and communicative abilities and the reasons for the unfair use of generative neural networks. Reliance on the theoretical positions of the purpose-oriented approach allows to mark the boundaries of acceptable usage of AI tools and to outline the trajectories of the most effective overcoming of the identified problems along the axis of target settings of the involved actors. Discussion. The cognitive and communicative features of human are shown in the context of growing up, development of thinking and language ability in comparison with the tools of the AIS. Ethically unacceptable areas of application of the generative neural networks are systematized, taking into account their “weaknesses”. The analysis is made for conditions of experiencing dilemmas in which one may make an unethical choice, based on which we highlight the principles of minimizing such situations in the sphere of AIS applying. Specific actions are proposed to take in order to regulate the usage of neural networks in education. Conclusion. The article systematizes the problems and ways of their comprehensive overcoming in the context of the onto-gnoseological foundations of interaction between human and the AIS in the “hybrid” reality.
|
