Technology Impact Report: The Rise of ChatGPT in Education

TECHNOLOGY IMPACT REPORT: THE RISE OF ChatGPT IN EDUCATION
By (Name)
The Name of the Class (Course)
Professor (Tutor)
The Name of the School (University)
City and State of its Location
Deadline
Technology Impact Report: The Rise of ChatGPT in Education
Introduction
ChatGPT (Chat Generative Pre-trained Transformer) is an AI chatbot created by OpenAI and launched on November 30, 2022 (OpenAI, 2023). When asked, “What is ChatGPT?” ChatGPT replied, “Hello! I’m ChatGPT, a large language model developed by OpenAI. I’m designed to be able to understand and generate human-like language, and I can do things like answer questions, have conversations, and even generate text based on prompts” (ChatGPT, 2023). The technology has been hailed as a breakthrough in artificial intelligence (AI), and natural language processing (NLP) is expected to have far-reaching implications for how people use the web. This technology report examines two of the most controversial impacts of ChatGPT in the context of research, writing, and teaching in academics.
How ChatGPT Works
As a large language model (LLM), ChatGPT uses deep learning to process vast collections of text and, in turn, generate text in conversational form based on its GPT-3.5 language processing AI model (Browne, 2023; Shen et al., 2023). GPT-3.5 is an update on GPT-3 with 175 billion parameters and was trained on 570 GB of text (Montti, 2022). Research suggests that the dramatic increase in scale from GPT-2 to GPT-3 caused behavioral changes in the model, such as being “able to perform tasks it was not explicitly trained on, like translating sentences from English to French, with few to no training examples” (Montti, 2022).
In addition to being trained on enormous datasets, ChatGPT was trained using Reinforcement Learning from Human Feedback (RLHF) which enables ChatGPT to adjust its behavior based on input from human evaluators and output responses optimized for human user intention (Montti, 2022; OpenAI, 2023; Shen et al., 2023). Further, ChatGPT has been fine-tuned several iterations using Proximal Policy Optimization (PPO), a new class of reinforcement learning algorithms optimized for both usability and performance (OpenAI, 2023). Recent research indicates that ChatGPT, as a transformer, with its self-attention mechanism and advanced neural network, excels in a wide array of natural language tasks such as automatic summarization, machine translation, and question-answering (Shen et al., 2023).
ChatGPT in the Context of Education
Unlike modern search engines such as Google Search and Microsoft Bing, ChatGPT is fundamentally more direct and more coherent in its answers to user requests because rather than displaying a list of links to external web pages, it provides a direct and immediate response that flows like human dialogue (Abdul, 2023; Ortiz, 2023). This ability to produce human-sounding text allows users to not only engage in long natural-language conversations with ChatGPT but also use ChatGPT to generate well-written and well-structured essays, articles, editorials, proposals, technical reports, scientific papers, novels, and even poems. The numerous applications of ChatGPT in the production of textual material introduce new challenges in education as questions about academic (dis)honesty, originality, accuracy, and bias abound. Other concerns regarding the potential role of ChatGPT in education include job security, effective and appropriate curriculum and lesson planning, and the future of peer-reviewed research and publication.
However, it is imperative to recognize the many limitations of ChatGPT as a source of knowledge and education. First, ChatGPT is trained on a large but finite volume of data available up to the end of 2021, which means that ChatGPT has no information on developments from 2022 onwards (Abdul, 2023). By inference, this also means that ChatGPT is vulnerable to bias because OpenAI’s researchers and developers determine what data is used to train ChatGPT (Wu, 2022). Second, ChatGPT may produce seemingly credible but factually incorrect answers under what is known as the “hallucination effect” (Abdul, 2023; Montti, 2022; OpenAI, 2023; Shankland, 2023; Shen et al., 2023; Wu, 2022). Third, ChatGPT’s quality of the answer depends on the user’s quality of instruction or direction, as ChatGPT is sensitive to phrasing and will guess the user’s intention rather than ask for clarification (Abdul, 2023; Montti, 2022; OpenAI, 2023; Shankland, 2023; Shen et al., 2023; Wu, 2022). Fourth, while ChatGPT is programmed to decline inappropriate requests, sometimes it can be fooled by the user into providing harmful information (OpenAI, 2023; Shankland, 2023).
Impact #1: ChatGPT and Academic Integrity
One of the significant risks to the education industry posed by ChatGPT is the potential for academic misconduct (Khalil and Er, 2023: 3). Ultimately, it is likely to have a significant impact on academic integrity. The technology can produce automated text that can be used to produce convincing essays and other written works. Students can generate convincing answers to academic questions by feeding said questions to ChatGPT in an open-ended format. In this way, ChatGPT has the potential to enable students to quickly and easily create convincing documents that they could use in place of their original work. According to Khalil and Er (2023), ChatGPT-generated papers contain “…highly original content [unlikely to] …face plagiarism issues if” submitted for assignments (8). Commenting on the writing capability of ChatGPT, a high-school English teacher of 12 years evaluated that ChatGPT, in its current iteration, is superior to most students in writing proficiency (Herman, 2022).
In a sample of over 200 K-12 teachers, a recent survey found that over one in four have caught at least one student cheating using ChatGPT (McDade, 2023). In another recent survey, 500 out of 1000 college students stated that they knew about ChatGPT, with 64% of those admitting to having used ChatGPT for school work at least once. Niagara College reported that some of its students use ChatGPT to cheat and stated that the institution is on “high alert” (Hristova, 2023). A professor at North Michigan University reported that he had caught two students submitting essays written by ChatGPT (Nolan, 2023).
Echoing similar sentiments, a professor at the University of Pennsylvania’s Wharton School highlighted that ChatGPT would pass the final exam of the school’s MBA program with a grade of B to B- (Terwiesch, 2023: 2). Similarly, when put to the test on multiple exams, ChatGPT reportedly scored 70% on the United States Medical Licensing Examination, 70% on a bar exam for lawyers, 78% on New York state’s high school chemistry exam’s multiple-choice section, and placed in the 40th percentile on the Law School Admission Test (Shankland, 2023). These events point to an uncertain future for the education sector as their role of producing quality graduates is essential to maintaining a well-functioning society.
On the other hand, certain researchers and scholarly journals, such as Nurse Education in Practice and Oncoscience, have started using and crediting ChatGPT as a study author in official publications (Stokel-Walker, 2023: 620). In contrast, the peer-reviewed journal Science has updated its policy to forbid the use of AI-generated content explicitly and to declare that an AI program, includin...