ChatGPT and Me

Artificial intelligence is a part of our lives as a fun, maybe scary tool, but certainly not as a genius overlord– they still have a long way to go before that. These processors, using data from the Internet, generate written text that was not created by any human. But artificial intelligence has existed in different forms for a long time, perhaps longer than we might expect.

In the 1975 article “Artificial Intelligence and Linguistics: A Brief History of a One-way Relationship” published in Volume 1 of the Berkeley Linguistics Society’s proceedings of their first annual meeting, author Rosenberg describes several artificial intelligence language processors as they manifested in the mid-20th century.

In one of the oldest processors, a Machine Translator (MT) from the 1940s, text in one language goes in, and translated text comes out, all on a physical cardboard disc. It was considered a failure due to its limited processing abilities of the intricacies of language as well as the tedious necessity of editing and pre-editing for a comprehensible result. Presently, Google Translate is extremely usable and is still constantly improving– even in the past decade there has been a noticeable difference in the quality of the translations it produces. Question answering systems functioned as offline search engines, producing preloaded information upon request. Prominent was BASEBALL (1961) which answered simple semantic questions about baseball games. SAD-SAM (1963) constructed family trees using inputted grammatically accurate sentences, for example “Harry lives with his Aunt Petunia, Uncle Vernon, and their son Dudley.” It could then answer questions such as “Who is Harry’s cousin?.” STUDENT (1964), when given a word problem in English generated an algebra equation and then solved it. Now, over half a century later,  AI creates quality text so competently that education institutions are worried about students passing it off as their own work.

The peril of using an AI to produce an essay is not just guilt weighing down on your conscience and depriving yourself of a critical thinking opportunity, but legitimately getting punished for plagiarism. Despite the bot creating unique text, “(When) checked by Turnitin (the most widely used plagiarism detection service), it would show very high degrees of similarity between the submissions of all students using the same set of keywords as prompts.” So while ChatGPT creates unique works, if students submit similar prompts, it produces noticeably similar essays. 

So when ChatGPT creates an essay, a script, or a short story, is it the true author, despite not being a human? When publishing work created by ChatGPT or any language producing processor, should you credit it as an author? In “Chatting and Cheating. Ensuring Academic Integrity in the Era of ChatGPT,” a paper published earlier this year from professors at the University of Plymouth, they utilized ChatGPT to actually write part of their article (revealed part way through the paper, a shocking plot twist worthy of The Matrix), giving it prompts such as “Produce several witty and intelligent titles for an academic research paper on the challenges universities face in ChatGPTand plagiarism.” In the acknowledgements of their paper they stated:

Although it has certainly made a significant contribution to the work, and has drafted much of the article, it was unable to provide agreement on submission to this journal and it has not reviewed and agreed the article before submission. Perhaps more crucially, it cannot take responsibility and be accountable for the contents of the article – although it may be able to assist with any questions raised about the work.

Sources 

Cotton, Debby, et al. “Chatting and Cheating. Ensuring Academic Integrity in the Era of Chatgpt.” EdArXiv, 10 Jan. 2023. Web.

Rosenberg, R. S. (1975). Artificial Intelligence and Linguistics: A Brief History of a One-way Relationship. Proceedings of the Annual Meeting of the Berkeley Linguistics Society, 1. Retrieved from https://escholarship.org/uc/item/0p8526q0

Photos:

IBM. (n.d.). 702 Data Processing System operator's console. IBM Archives. Retrieved January 25, 2023, from https://www.ibm.com/ibm/history/exhibits/mainframe/mainframe_2423PH702C.html 

Woman operating the card puncher, c.1940. (n.d.). U.S. National Archives and Records Administration. photograph. Retrieved January 25, 2023, from Card_puncher_-_NARA_-_513295.jpg.