Within the recent years, the term “generative AI” has started to appear more commonly in the media. What once was a feature in science fiction or cyberpunk settings in fictional stories, the ability to make a program answer any question summarizing information found online by just writing a few keywords into a program has now become an easily accessible and somewhat common tool in our everyday lives called ChatGPT. On one hand, such tools are helpful time savers, allowing people to generate ideas faster, however, the increasing accessibility of it has achieved ChatGPT negative reputation, especially in regards of misinformation, academic honesty and copyright infringement.

The basic functions of ChatGPT are known to most due to its popularity – it is a chatbot, which offers various functions, such as answering questions, suggesting topics, rephrasing information and other similar text-based features. On a surface level, it is a highly efficient tool, however, it is not all-powerful either. Generative AI works on the principle of reusing information that is already available to it. A user inputs a query and the tool “generates” an answer based on the keywords used in the question. ChatGPT has access to the internet through web scraping, where it had automatically extracted data from various websites and stored in its own databank. With this principle, a few issues arise – for example, what happens, when the user asks for information that the tool does not have information on? It turns out, the tool simply creates its own information by reusing and mixing unrelated information. Because of this, the tool often can offer blatant misinformation. Obviously, when using ChatGPT while being consciously aware it’s a generative tool, not a true database, such mishaps should not pose any danger to its users. Unfortunately, a lot of people decide to use ChatGPT as a search engine, which can lead to some issues, especially if it’s used for important research.

Another issue is copyright and academic honesty. If asked, ChatGPT may not only provide ideas for stories, but also write short ones by itself. But, of course, the program does not have the capability to create anything unique and instead uses the web scraped data to mix it together into a new combination. A widely known and discussed problem is, who is considered the true author of the text? If a student is tasked to write an essay and asks ChatGPT to write it for them, is the author of that essay that the tool will provide the student, ChatGPT or the various authors whose texts were scraped from the web and repurposed in the student’s request? Furthermore, if students become used to ChatGPT writing everything for them, it is possible they will never learn how to write by themselves, therefore becoming unable to write their own senior exam essays, their higher education thesis, then later the reports and reviews in their workplaces, or even such simple things as important emails and messages to their superiors and loved ones.

Undeniably, when used right, ChatGPT can provide some value as well and there are many studies already showing that a rational usage of the tool can even boost work efficiency, however, as things stand now, many take the tool for granted and are becoming too reliant on it. And although there are many tools being created that can be used to identify the usage of any kind of generative AI in written texts, they, just like ChatGPT, aren’t perfect and may misidentify texts, which already happens in academic settings, marking essays written by students as AI and ChatGPT assisted essays as completely original. It is also important to remember that the true extent of what ChatGPT can and can’t do is still unknown and can only be found by using it and allowing the tool and its users to make mistakes. However, it is quite obvious that the more ChatGPT is used and updated and the more information it acquires, the harder it will be to differentiate between texts written by humans and generative AI, threatening to completely change how we view academic honesty and copyright due to its advancements.

Prepared by Austėja Malinauskaitė, an intern at the LieDM Association, bachelor’s student, Kaunas University of Technology, Faculty of Social and Humanities and Arts