Answers are grounded in this paper and cite their sources.
Analysis reveals AI's limitations in accurately capturing Gerard Manley Hopkins' relationship with London, suggesting ethical implications for pedagogy and scholarship.
Jude V. Nixon’s “Confident Confabulations: Gerard Manley Hopkins in London—AI + Cambridge University Press” compares two forms of text: his chapter “London,” for Cambridge University Press, against two sets of AI-produced essays exploring the roles London played in Hopkins’s life. Analyzing the responses to series of prompts on confluence between city and poet, Nixon examines the AI-generated texts to locate instabilities, “hallucinations” (factual errors), useful and authentic information, sourcing, and attribution, seeking to determine what information the LLMs got right, what they got wrong, and what they ignored, omitted, or simply did not know. While LLMs do make choices, they do not exercise judgment in the process by which ideas originate, evolve, and grow. Other questions persist: What are the ethical, pedagogical, and scholarly value to these productions? How might they impact the ways students learn and scholars write? How can the output be evaluated? Perhaps the most fundamental question with AI-generated texts is one of agency, not the deus ex machina but the homo ex machina —what cumulative effects might these have on the dis- or re-placement of the human?
No takes yet. Share an insight, caveat, or question.
Jude V. Nixon (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: