Primjena i problematika umjetne inteligencije u književnosti

master's thesis

• Title (english): Application and issues of artificial intelligence in literature
• Author: Maja Tomiša
• Mentor: Tomislav Ivanjko (mentor)
• Committee member: Ana Barbarić (predsjednik povjerenstva)
• Committee member: Monika Batur (član povjerenstva)
• Committee member: Tomislav Ivanjko (član povjerenstva)
• Granter: University of Zagreb; Faculty of Humanities and Social Sciences; (Department of information and Communication sciences); Zagreb
• Defense date and country: 2024-09-16, Croatia
• Scientific / art field, discipline and subdiscipline: SOCIAL SCIENCES; Information and Communication Sciences; Library Science

Abstract

Modeli generativne umjetne inteligencije svakoga dana postaju sve dostupniji poslovnim i privatnim korisnicima te se, kako eksperimentalno tako i s ciljem kreiranja proizvoda za krajnjeg korisnika, počinju koristiti i u stvaranju književnih djela. Stavovi stručnjaka iz područja računalnih znanosti, filologije i raznih umjetničkih područja podijeljeni su u pogledu budućeg utjecaja umjetne inteligencije na stvaralaštvo. Cilj je ovoga rada analizirati načine na koje modeli generativne umjetne inteligencije obrađuju i generiraju prirodni jezik kako bi se odredile granice njihovih mogućnosti i potencijalne prijetnje autorskome radu, pravima i etičkim obvezama prema krajnjim korisnicima.
U radu se donosi pregled razvoja tehnologije koju se naziva umjetnom inteligencijom te njezini glavni mehanizmi obrade podataka, odnosno data mining i text mining, te povratne neuronske mreže čije je nelinearno obrađivanje podataka omogućilo stvaranje kvalitetnijih izlaznih vrijednosti. Kako je fokus istraživanja stavljen na književnost, posebno se posvećuje pažnja računalnom jezikoslovlju te radu s prirodnim jezikom u smislu mogućnosti generiranja, prevođenja, analize te stilskih obilježja teksta koje model umjetne inteligencije može ponuditi. Postupci obrade, razumijevanja i stvaranja na prirodnome jeziku popraćeni su analizama i usporedbama s načinima na koje uče ljudi, usvajaju i koriste jezik te kako nastaju ideje i razmatranjima što je to zapravo kreativnost. Neovisno o kvaliteti ponuđenih rezultata javlja se problem autorskoga prava na podatke koji se koriste za treniranje te generirane tekstove. Trenutno se pokušavaju osmisliti pravedna rješenja podjele prava ili kompenzacije izvorima i korisnicima, a to uključuje i etička pitanja pravednog korištenja, zloupotrebe modela ili rezultata.
Modeli generativne umjetne inteligencije sposobni su ponuditi tekstove koji su jezično smisleni, ali umjetnički bezvrijedni. Brzina generiranja teksta profitabilna je ako čitateljska publika ne zahtijeva autorsku intenciju, što je rijetko slučaj, a cijeli proces zadavanja cilja, usmjeravanja i korigiranja rezultata još uvijek zahtijeva ljudski nadzor.

Abstract (english)

Models of generative artificial intelligence are becoming more and more available to business and private users every day and, both experimentally and with the aim of creating products for the end user, they are also starting to be used in the creation of literary works. The views of experts from the fields of computer science, philology and various artistic fields are divided regarding the future impact of artificial intelligence on creativity. The aim of this paper is to analyze the ways in which generative artificial intelligence models process and generate natural language in order to determine the limits of their possibilities and potential threats to author's work, rights and ethical obligations towards end users.
The paper provides an overview of the development of technology called artificial intelligence and its main mechanisms of data processing, namely data mining and text mining, and recurrent neural networks whose non-linear data processing enabled the creation of higher quality output. As the focus of the research is on literature, special attention is paid to computational linguistics and work with natural language in terms of the possibility of generation, translation, analysis and stylistic features of the text that the artificial intelligence model can offer. The processes of processing, understanding and creation in natural language are accompanied by analysis and comparisons with the ways in which people learn, adopt and use language, as well as how ideas are created and considerations of what creativity actually is. Regardless of the quality of the offered results, the problem of copyright arises for the data used for training and the generated texts. Currently, attempts are being made to devise fair solutions for the distribution of rights or compensation to sources and users, and this includes ethical issues of fair use, misuse of models or results.
Generative artificial intelligence models are capable of offering texts that are linguistically meaningful but artistically worthless. The speed of text generation is profitable if the readership does not require the author's intention, which is rarely the case, and the whole process of setting the goal, directing and correcting the results still requires human supervision.

• Keywords: generativna umjetna inteligencija; književnost; text mining; kreativnost; autorsko pravo
• Keywords (english): generative artificial intelligence; literature; text mining; creativity; copyright
• Language: croatian
• URN:NBN: urn:nbn:hr:131:786070
• Study programme: Title: Information Sciences; specializations in: Archivistics Study, Library Science, Information Sciences (research), Information Sciences (teaching), Museology and heritage management, Informatology; Course: Library Science; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra bibliotekarstva (magistar/magistra bibliotekarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-09-17 09:12:46