Wilbanks, D., Mondal, D., Tandon, N. & Gray, K. Large Language Models as Moral Experts? GPT-4o Outperforms Expert Ethicist in Providing Moral Guidance. Preprint at https://doi.org/10.31234/osf.io/w7236 (2024).
AI has demonstrated expertise across various fields, but its potential as a moral expert remains unclear. Recent work suggests that Large Language Models (LLMs) can reflect moral judgments with high accuracy. But as LLMs are increasingly used in complex decision-making roles, true moral expertise requires not just aligned judgments but also clear and trustworthy moral reasoning. Here, we advance work on the Moral Turing Test and find that advice from GPT-4o is rated as more moral, trustworthy, thoughtful, and correct than that of the popular The New York Times advice column, The Ethicist. GPT models outperformed both a representative sample of Americans and a renowned ethicist in providing moral explanations and advice, suggesting that LLMs have, in some respects, achieved a level of moral expertise. The present work highlights the importance of carefully programming ethical guidelines in LLMs, considering their potential to sway users’ moral reasoning. More promisingly, it suggests that LLMs could complement human expertise in moral guidance and decision-making.
Why it matters
The general consensus among most people has been that human values will forever be the domain of human beings, and not AI. This paper seems to suggest that moral judgement may not be off-limits to machines after all.
My key takeaway
I can’t say for certain that generative AI ‘is moral’ or that it has any understanding of what it means to ‘be moral’. But, if it’s outputs are judged by humans to represent a moral point of view, and further, that those perspectives are judged to be more moral than human judgements (according to human evaluators), then surely we have a responsibility to include generative AI into our decision frameworks? At least as another variable to consider.
Possible implication
Language models may be integrated into decision-making (sorry, decision-support) systems. For example, in high-stakes contexts like criminal justice, healthcare, social welfare, and so on.