Against Machine Originality On Lovelace, LLMs, and ChatGPT

Abstract

This paper was originally written for Dr. Kino Zhao’s PHIL 302 course Topics in Epistemology and Metaphysics: Philosophy of Machine Learning. The assignment asked students to a write a term paper critically engaging with various course texts on artificial intelligence, computation, and machine learning. The paper uses APA citation style.

In this paper, I argue in defence of Lady Lovelace’s objection that machines have no pretensions to originate anything. I begin by examining Alan Turing’s response to this objection, focusing on his appeal to machine unpredictability and the child-machine thought experiment, both of which are intended to demonstrate that machines can, in principle, possess pretensions to originate. I argue, however, that machine “learning” ultimately remains derivative of human-provided structures. Using modern large language models (LLMs) like ChatGPT as a contemporary case study, I conclude that apparent machine originality is better understood as an extension of human design rather than a genuine act of originality.