Cognitive processes in imaginative moral shifts: How judgments of morally unacceptable actions change

AI contextual information shapes moral and aesthetic judgments of AI-generated visual art

Throughout history, art creation has been regarded as a uniquely human means to express original ideas, emotions, and experiences. However, as Generative Artificial Intelligence reshapes visual, aesthetic, legal, and economic culture, critical questions arise about the moral and aesthetic implications of AI-generated art. Despite the growing use of AI tools in art, the moral impact of AI involvement in the art creation process remains underexplored. Understanding moral judgments of AI-generated art is essential for assessing AI's impact on art and its alignment with ethical norms. Across three pre-registered experiments combining explicit and implicit paradigms with Bayesian modelling, we examined how information about AI systems influences moral and aesthetic judgments and whether human art is implicitly associated with positive attributes compared to AI-generated art. Experiment 1 revealed that factual information about AI backend processes reduced moral acceptability and aesthetic appeal in certain contexts, such as gaining financial incentives and art status. Experiment 2 showed that additional information about AI art's success had no clear impact on moral judgments. Experiment 3 demonstrated that an implicit association task did not reliably link human art with positive attributes and AI art with negative ones. These findings show that factual information about AI systems shapes judgments, while different information doses about AI art's success have limited moral impact. Additionally, implicit associations between human-made and AI-generated art are similar. This work enhances understanding of moral and aesthetic perceptions of AI-generated art, emphasizing the importance of examining human-AI interactions in an arts context, and their current and evolving societal implications.

How people reason with counterfactual and causal explanations for Artificial Intelligence decisions in familiar and unfamiliar domains

Few empirical studies have examined how people understand counterfactual explanations for other people's decisions, for example, "if you had asked for a lower amount, your loan application would have been approved". Yet many current Artificial Intelligence (AI) decision support systems rely on counterfactual explanations to improve human understanding and trust. We compared counterfactual explanations to causal ones, i.e., "because you asked for a high amount, your loan application was not approved", for an AI's decisions in a familiar domain (alcohol and driving) and an unfamiliar one (chemical safety) in four experiments (n = 731). Participants were shown inputs to an AI system, its decisions, and an explanation for each decision; they attempted to predict the AI's decisions, or to make their own decisions. Participants judged counterfactual explanations more helpful than causal ones, but counterfactuals did not improve the accuracy of their predictions of the AI's decisions more than causals (Experiment 1). However, counterfactuals improved the accuracy of participants' own decisions more than causals (Experiment 2). When the AI's decisions were correct (Experiments 1 and 2), participants considered explanations more helpful and made more accurate judgements in the familiar domain than in the unfamiliar one; but when the AI's decisions were incorrect, they considered explanations less helpful and made fewer accurate judgements in the familiar domain than the unfamiliar one, whether they predicted the AI's decisions (Experiment 3a) or made their own decisions (Experiment 3b). The results corroborate the proposal that counterfactuals provide richer information than causals, because their mental representation includes more possibilities.

Moral judgements reflect default representations of possibility

Moral judgement requires representing what is possible: judging that someone ought to do something implies that they actually can do that thing; and if they cannot do that thing, then it is not the case that they ought to have done it. Moral judgements are often made quickly and effortlessly, suggesting that they may rely on default, rather than deliberative, representations of what is possible. To investigate this possibility, we asked participants to make 10 different kinds of moral judgements either very fast or more slowly about 240 different actions across 12 contexts. We found that these moral judgements were more similar to one another when participants were forced to quickly assess the morality of immoral actions, suggesting a common default template for moral judgement that becomes more differentiated upon reflection. When making moral permissibility judgements quickly, participants were more likely to judge that improbable, irrational and impossible actions were not permissible, indicating that default representations of permissibility may be reflecting default representations of possibility. A direct investigation revealed a close relationship between default representations of possibility and fast judgements of moral permissibility. These findings demonstrate the role of default representations of possibility in moral cognition. This article is part of the theme issue 'Thinking about possibilities: mechanisms, ontogeny, functions and phylogeny'.