Skip to main content Skip to navigation

The Death of the Short-Form Physics Essay in the Coming AI Revolution

Project Overview

The document explores the impactful role of generative AI, specifically models like ChatGPT and davinci-003, in the educational landscape, particularly in producing high-quality short-form essays in subjects like physics. A study highlighted within reveals that AI-generated essays can achieve first-class marks, comparable to those written by human students, prompting a critical reassessment of traditional assessment methods in education. This finding raises questions about the effectiveness of conventional short-form assessments and suggests that educational institutions may need to adapt their evaluation strategies to account for the advancements in AI technology. The document emphasizes the rapid evolution of AI and its potential to revolutionize educational practices, highlighting both the opportunities and challenges that come with integrating such technologies into learning and assessment processes.

Key Applications

AI-generated short-form physics essays using models like ChatGPT and davinci-003

Context: University-level Physics assessments, specifically the Physics in Society module

Implementation: AI models generated essays based on specific prompts related to physics questions, which were then marked by independent assessors.

Outcomes: AI essays achieved an average score of 71.2%, comparable to the module average of 71.5%, indicating high quality and potential for AI to replace traditional essay writing.

Challenges: Challenges include detecting AI-generated text, potential misuse by students, and the need for new assessment strategies.

Implementation Barriers

Technological and Educational

Current plagiarism detection tools focus on identifying copied text rather than distinguishing between AI-generated and human-written text. There is also a fear that AI-generated essays could undermine the integrity of assessments.

Proposed Solutions: Requiring in-person assessments or invigilated settings to ensure authenticity, and redesigning assessments to include formats that are less susceptible to AI generation.

Project Team

Will Yeadon

Researcher

Oto-Obong Inyang

Researcher

Arin Mizouri

Researcher

Alex Peach

Researcher

Craig Testrow

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Will Yeadon, Oto-Obong Inyang, Arin Mizouri, Alex Peach, Craig Testrow

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies