Julian De Freitas’s research while affiliated with Harvard University and other places

Humanoid robots are a form of embodied artificial intelligence (AI) that looks and acts more and more like humans. Powered by generative AI and advances in robotics, humanoid robots can speak and interact with humans rather naturally but are still easily recognizable as robots. But how will we treat humanoids when they seem indistinguishable from humans in appearance and mind? We find a tendency (called "anti-robot" speciesism) to deny such robots humanlike capabilities, driven by motivations to accord members of the human species preferential treatment. Six experiments show that robots are denied humanlike attributes, simply because they are not biological beings and because humans want to avoid feelings of cognitive dissonance when utilizing such robots for unsavory tasks. Thus, people do not rationally attribute capabilities to perfectly humanlike robots but deny them capabilities as it suits them.

The deployment of autonomous vehicles (AVs) and the accompanying societal and economic benefits will greatly depend on how much liability AV firms will have to carry for accidents involving these vehicles, which in turn impacts their insurability and associated insurance premiums. Across three experiments ( N = 2677), we investigate whether accidents where the AV was not at fault could become an unexpected liability risk for AV firms, by exploring consumer perceptions of AV liability. We find that when such accidents occur, the not‐at‐fault vehicle becomes more salient to consumers when it is an AV. As a result, consumers are more likely to view as relevant counterfactuals in which the not‐at‐fault vehicle might have behaved differently to avoid or minimize damage from, the accident. This leads them to judge AV firms as more liable than both firms that make human‐driven vehicles and human drivers for damages when not at fault.

A preregistered audit of 600 images by generative AI across 150 different prompts explores the link between humor and discrimination in consumer-facing AI solutions. When ChatGPT updates images to make them “funnier”, the prevalence of stereotyped groups changes. While stereotyped groups for politically sensitive traits (i.e., race and gender) are less likely to be represented after making an image funnier, stereotyped groups for less politically sensitive traits (i.e., older, visually impaired, and people with high body weight groups) are more likely to be represented.

Based on a review of relevant literature, we propose that the proliferation of AI with human-like and social features presents an unprecedented opportunity to address the underlying cognitive and affective drivers of prejudice. An approach informed by the psychology of intergroup contact and prejudice reduction is necessary because current AI systems often reinforce or avoid prejudices. Against this backdrop, we outline unique opportunities for prejudice reduction through 'synthetic' intergroup contact, wherein consumers interact with AI products and services that counter stereotypes and serve as a 'proxy' members of the outgroup (i.e., counter-stereotypical AI). In contrast to human-human contact, humanizing and socializing AI can reduce prejudice through more repeated, direct, unavoidable, private, non-judgmental, col-laborative, and need-satisfying contact. We illustrate the potential of synthetic inter-group contact with counter-stereotypical AI using examples of gender stereotypes and hate speech and discuss practical considerations for implementing counter-stereotypical AI without inadvertently perpetuating or reinforcing prejudice. K E Y W O R D S artificial intelligence, intergroup psychology, prejudice, stereotypes, synthetic contact

There has been a call for a potentially revolutionary change to our existing understanding of the psychological concept of personal identity. Apparently, people can psychologically represent people, including themselves, as multiple individuals at the same time. Here, we ask whether the intransitive judgments found in these studies truly reflect the operation of an intransitive concept of personal identity. We manipulate several factors that arbitrate between transitivity and intransitivity and find most support for transitivity: In contrast to the prior work, most participants do not make intransitive judgments when there is any reason to favor one individual over another. People change which single individual they personally identify with, depending on which individual competes more strongly or weakly for identity, rather than identifying with both individuals. Even when two individuals are identical and therefore both entitled to be the same person, we find that people make more transitive judgments once they understand the practical commitments of their responses (Experiment 4) and report not being able to actually imagine two perspectives simultaneously when reasoning about the scenario (Experiment 5). In short, we suggest that while people may make intransitive judgments, these do not reflect that they psychologically represent identity in an intransitive manner.

Can consumers form especially deep emotional bonds with AI and be vested in AI identities over time? We leverage a natural app-update event at Replika AI, a popular US-based AI companion, to shed light on these questions. We find that, after the app removed its erotic role play (ERP) feature, preventing intimate interactions between consumers and chatbots that were previously possible, this event triggered perceptions in customers that their AI companion's identity had discontinued. This in turn predicted negative consumer welfare and marketing outcomes related to loss, including mourning the loss, and devaluing the "new" AI relative to the "original". Experimental evidence confirms these findings. Further experiments find that AI companions users feel closer to their AI companion than even their best human friend, and mourn a loss of their AI companion more than a loss of various other inanimate products. In short, consumers are forming human-level relationships with AI companions; disruptions to these relationships trigger real patterns of mourning as well as devaluation of the offering; and the degree of mourning and devaluation are explained by perceived discontinuity in the AIs identity. Our results illustrate that relationships with AI are truly personal, creating unique benefits and risks for consumers and firms alike.

Chatbots are now able to engage in sophisticated conversations with consumers in the domain of relationships, providing a potential coping solution to widescale societal loneliness. Behavioral research provides little insight into whether these applications are effective at alleviating loneliness. We address this question by focusing on AI companions applications designed to provide consumers with synthetic interaction partners. Studies 1 and 2 find suggestive evidence that consumers use AI companions to alleviate loneliness, by employing a novel methodology for fine tuning large language models to detect loneliness in conversations and reviews. Study 3 finds that AI companions successfully alleviate loneliness on par only with interacting with another person, and more than other activities such watching YouTube videos. Moreover, consumers underestimate the degree to which AI companions improve their loneliness. Study 4 uses a longitudinal design and finds that an AI companion consistently reduces loneliness over the course of a week. Study 5 provides evidence that both the chatbots' performance and, especially, whether it makes users feel heard, explain reductions in loneliness. Study 6 provides an additional robustness check for the loneliness alleviating benefits of AI companions.