István Zoltán Zárdai. 2024. The Ethics of LLMs at Universities: A Case for Restriction and Regulation. PDF
‘Disruptive technologies’ is a euphemism for new technologies released lacking adequate regulation, causing significant unemployment and costly, inefficient additional labour. So it stands with LLMs. They output lookalikes of authored writing. Most output remixes existing materials, effectively stealing, since lacking understanding and intention original meaning is not added.
LLMs enable low-cost, high-reward dishonesty. Students attempt to submit these products as their own texts. Some in education propose to use LLMs to allow students to generate text and then revise it. This is feasible in groups small enough to ensure that students can be monitored continuously and substantially rewrite the texts while reflecting on the editing process. However, such highlighting of what is important about rewriting tasks reveals that LLMs do not add value to such activities.
The recommended policy is to restrict and regulate LLMs. This protects jobs; it ensures that students have to develop skills that enhance their agency and their character development; and alleviates the problem of massive scale intellectual property misuse. Resistance to AI is urgent: no company promoting it will back down from doing so, no matter the risks of their products. This is a case of a game where one agent will never cooperate, and both agents endorse incompatible value sets. Consequently, no group potentially adversely affected should have qualms about intense opposition and has a duty to lobby for regulation and conditional supervised rollout of LLMs.
Keywords
academic writing, critical thinking, AI, ethics, regulation
AWCT 2024