AIAI

The AIAI project delves into the complexities of authorship and interpretation in the context of generative artificial intelligence (AI) like ChatGPT, DallE, or Midjourney. Drawing from poststructuralist traditions and empirical methods, it explores the ambiguity surrounding interpreting AI-generated texts and images, questioning the very notion of authorship in this landscape. The study unfolds in three stages: (1) evaluating existing theories of authorship, (2) conducting expert interviews with AI-using artists and writers, and empirical research involving participants co-creating with AI, and (3) synthesizing the findings to create a new framework for understanding authorship and interpretation of works created with the help of AI tools. The results may potentially impact copyright law and diverse humanities fields reliant on these concepts.

The main aim of the project is to study authorship and interpretation of works created with help of AI tools. The main goal is to create a new framework that redefines the current understanding of authorship and interpretation in the context of generative artificial intelligence.

AI: autorství a interpretace

Projekt AIAI se zabývá složitostí autorství a interpretace v kontextu generativní umělé inteligence (AI), jako je ChatGPT, DallE nebo Midjourney. Vychází z poststrukturalistických tradic a empirických metod a zkoumá nejednoznačnost interpretace textů a obrazů generovaných umělou inteligencí a zpochybňuje samotný pojem autorství v tomto prostředí. Studie se odvíjí ve třech fázích: (1) vyhodnocuje existující teorie autorství, (2) provádí expertní rozhovory s umělci a spisovateli používajícími AI a empirický výzkum zahrnující účastníky spoluvytvářející s AI a (3) syntetizuje výsledky s cílem vytvořit nový rámec pro chápání autorství a interpretaci děl vytvořených pomocí nástrojů AI. Výsledky mohou potenciálně ovlivnit autorské právo a různé humanitní obory závislé na těchto konceptech.

Hlavním cílem projektu je studium autorství a interpretace děl vytvořených pomocí nástrojů umělé inteligence. Hlavním cílem je vytvořit nový rámec, který nově definuje současné chápání autorství a interpretace v kontextu generativní umělé inteligence.

Project Stucture

The research questions in the project will be grouped into three work packages (WP), each corresponding to a cluster of interconnected problems.

WP1. Determining the status of the author of AI-generated texts and images.

Research questions: Can we understand the process of creating texts and images with the aid of AI in terms of authorship? Are theories of authorship developed in the humanities – whether hermeneutic, psychoanalytic, structuralist or post-modern – suited to understanding the problem of authorship of AI-generated texts and images? Can ideas developed within the field of copyright law help us to understand the ontological status of the author of AI-generated material? How do (human) artists who create works with the aid of AI tools understand the problem of authorship in the process?

WP2. Understanding the status of interpretation of AI-generated texts and images.

Research questions: Does it make sense to apply methods of interpretation devised for human-made texts and images to AI-generated material? What is the status of interpretation of AI-generated material both as a practice and as a product? Can this kind of interpretation be likened to the interpretation of other non-human generated material, such as animal art? How do experts at interpretation (e.g., literary scholars or art critics) understand the process of interpretation of AI-generated work?

WP3. Developing a new framework for understanding authorship and interpretation of AI-generated works.

Research questions: What theories of authorship and interpretation should look like to encompass the changes brought about by the advent of AI-generated content? Should theories of authorship concentrate more on the process of creation or the beliefs of the person engaged in the process? Should theories of interpretation focus on the work itself, the author (whoever or whatever it is), the person interpreting, or the act of interpretation itself?

Work Plan

The first two work packages will be divided into three phases – (1) theoretical research; (2) expert interviews; (3) empirical research. The first year of the project will focus on the first two phases of WP1 and WP2 – the theoretical research (explained in 4b below) will help to design a frame for discussing the topic with external experts. These interviews, in turn, will lead to the construction of the empirical part of the research, which will involve testing the awareness of authors in a controlled, experimental environment.

The empirical research will be planned, prepared and performed in year two of the project during two meetings conducted by the whole project team (i.e., both Czech and Polish teams) in Warsaw and Prague. This will allow two iterations of the same research in two different millieus and languages (Czech and Polish) to ensure a better understanding of both general and country/culture-specific elements in the results.

The third year of the project will be entirely devoted to WP3, which will include a summary of the results of the two previous phases and the establishment of a new theoretical framework for interpretation and authorship of works created by or with the aid of AI tools.

Each year, we will organize two physical project meetings, one in Prague and one in Warsaw. These meetings will serve for a close collaboration of the research teams, but will also involve workshops with external participants: (1) In the first year, we will invite experts with practical experience in the field and jointly conduct some of the expert interviews. (2) In the second year, we will jointly perform some of the participant studies. (3) In the third year, we will invite relevant members of the scientific community to present and discuss our findings.

Year 1

WP1 and WP2 theoretical research will yield an in-depth theoretical understanding of the problem of authorship and interpretation of AI-generated works. This will allow us to design questions for external experts and to formulate hypotheses informing the empirical part of the research. In this phase, we will write no less than two theoretical scientific papers concerning the usefulness of current theories of interpretation and authorship for tackling problems related to AI-generated and -assisted works. We will also rework parts of the expert interviews into at least one article meant for a non-scientific outlet.

Year 2

WP1 and WP2 empirical research will allow us to comprehend these theoretical problems in the practical sense, i.e., to understand the perspective of the authors of works and of interpretations. The results will be published in at least two scientific papers.

Year 3

WP3 will result in developing a new understanding of authorship and interpretation with reference to AI-generated works. This framework will help to pose new questions in the area and thus continue empirical research. The results will be published in at least two scientific papers. We will also seek to promote our results in non-scientific outlets, from print media to popular podcasts.

Detailed description of the project