The proliferation of artificial intelligence (AI) technologies has led to the rise of conversational decision-making support systems, such as dialogue persona systems that provide conversational access to various user segments. For example, product managers can ask personas about features before implementing them, politicians can learn about the needs of local communities through personas, and so on. Nascent research has looked at challenges when users interact with AI personas, but has not framed it as a principal–agent problem, in which the AI represents a persona that itself represents real people in the data. This setting exposes unique interaction challenges that decision makers face when engaging with AI-generated conversational personas, which we examine through a user study with 56 participants using AI-generated conversational personas. Our results indicate seven interaction challenges: (1) Hidden Information, (2) Hidden Personas, (3) Hidden UI, (4) Lack of AI Agency, (5) AI’s Selective Attention, (6) Confusing Distributional Information, and (7) Conversational Cold Start that we conceptually link with agency theory. We discuss how the interaction challenges could be alleviated and suggest directions for future work. • This article explores how decision makers interact with AI-generated conversational personas derived from real survey data. • It conducts a comparative study between conversational personas and traditional profile personas in decision-support contexts. • The study employs a think-aloud user study with 56 participants to capture interaction experiences and challenges. • It identifies seven specific interaction challenges unique to conversational personas that may hinder effective decision making. • The article provides insights and recommendations for designing conversational decision support systems using AI-generated personas.

The development of persona systems provides a possibility for end users to interact with different persona modalities. In a 54-participant randomized controlled experiment, we compare two persona interaction modalities, document and dialogue personas, both generated using AI approaches from survey data. Overall, dialogue personas appear to be perceived more favorably than document personas. However, document personas exhibit a wider range of perceptions, suggesting that experiences with document personas are more polarizing among users. The document personas had higher transparency and were perceived as more complete, but the task completion was perceived as more difficult, although the task success rate was higher. The dialogue personas were perceived as more usable, with a higher System Usability Scale score, and more enjoyable. Our findings provide critical insights into the increasingly important area of persona interaction modalities and the broad paradigm of human-persona interaction.

We investigated the presence and acceptance of hallucinations (i.e., accidental misinformation) of an AI-generated persona system that leverages large language models for persona creation from survey data in a 54-user within-subjects experiment. After interacting with the personas, users were given a task to ask the personas a series of questions, including an unanswerable question, meaning the personas lacked the data to answer the question. The AI-generated persona system provided a plausible but incorrect answer half (52%) of the time, and more than half of the time (57%), the users accepted the incorrect answer, and the rest of the time, users answered the unanswerable question correctly (no answer). We found that when the AI-generated persona hallucinated, the user was significantly more likely to answer the unanswerable question incorrectly. Also, for genders separately, when the AI-generated persona hallucinated, it was significantly more likely for the female user and the male users to answer the unanswerable question incorrectly. We identified four themes in the AI-generated persona's answers and found that users perceive AI-generated persona's answers as long and unclear for the unanswerable question. Findings imply that personas leveraging LLMs require guardrails to ensure that personas clearly state the possibility of data restrictions and hallucinations when asked unanswerable questions.

Large Language Models (LLMs) are emerging as a powerful tool for AI-generated personas. This study evaluates the usability of AI-generated personas, comparing chat and profile formats. The findings indicate chat personas tend to be perceived more favourably, and profile personas exhibit greater variability in user perception. The increased difficulty and longer dwell time experienced by users with the profile persona, despite negative usability metrics, paradoxically resulted in better task performance. Usability issues indicate that many current limitations of AI, including verbosity, hallucinations, and empty rhetoric which was described as the persona having ‘no soul’, are inherited in AI-generated chat personas. However, there are also new issues. For one, the risk of information overload in an AI-generated profile persona implies that the AI does not consider human users’ cognitive limitations when designing the persona (but usability scores for profile personas increase with dwell time, implying that users get used to the longer format the more time they spend). Another is the ‘fourth wall’ effect of AI-generated chat personas in which the user feels they are talking to someone describing the persona rather than the persona itself. Future work could address the usability paradox and the fourth wall effect of using personas.CCS CONCEPTS Human-centered computing Human computer interaction (HCI)

Deepfake technology is increasingly integrated into global mobile and web services when human representation is not feasible or cost-effective. Our analysis of 202 deepfake avatars from three deepfake providers reveals significant demographic disparities with 18 out of 48 possible demographic groups unrepresented. Deepfake avatars' gender distribution was nearly balanced (49.01% male, 50.99% female), but older age groups (Baby Boomers and Silent Generation) were substantially underrepresented by 64.36% and 76.24%, respectively, relative to the average number of all deepfake avatars. Differences in language representation were present in deepfake avatar providers with only 1.06% of global languages covered. The findings indicate that current deepfake technology lacks diversity, primarily favoring young white individuals, neglecting older demographics, Asians, and Middle Eastern populations, with underrepresentation of 40.59% and 52.48%, respectively, relative to the average number of all deepfake avatars. Only 15.27% of deepfake avatars portray any occupational characteristics. Addressing these diversity gaps is crucial for better serving varied user groups and warrants attention from deepfake providers and caution from those using deepfakes.

Demographics are considered foundational information in most persona profiles. However, the effect of persona ethnicity and gender on designers’ identification with the persona has limited evaluation in the human-computer interaction literature. We conducted a study with 64 professional designers from the United States, Indian, Korean, and Mexican nationalities to investigate the effects of AI-generated persona ethnicity and gender on persona identification. The personas were created using Generative AI in the persona narratives and the persona video creation. The contribution of this work is that, against assumptions, neither persona ethnicity nor gender play a major role in persona identification among designers with different ethnic backgrounds. While there were some insinuations of ethnicity and gender in the open-ended feedback from the designers, the emergent qualitative themes describing persona identification were overwhelmingly universal and applicable regardless of ethnicity or gender. This implies that professional designers can effectively use personas with different demographic backgrounds, and effects of demographic attributes in personas leading to stereotyping are less impactful than presumed.

Deepfakes, realistic portrayals of people that do not exist, have garnered interest in research and industry. Yet, the contributions of deepfake technology to human-computer interaction remain unclear. One possible value of deepfake technology is to create more immersive user personas. To test this premise, we use a commercial-grade service to generate three deepfake personas (DFs). We also create counterparts of the same persona in two traditional modalities: classic and narrative personas. We then investigate how persona modality affects the perceptions and task performance of the persona user. Our findings show that the DFs were perceived as less empathetic, credible, complete, clear, and immersive than other modalities. Participants also indicated less willingness to use the DFs and less sense of control, but there were no differences in task performance. We also found a strong correlation between the uncanny valley effect and other user perceptions, implying that the tested deepfake technology might lack maturity for personas, negatively affecting user experience. Designers might also be accustomed to using traditional persona profiles. Further research is needed to investigate the potential and downsides of DFs.

This research explores the value of personas for supporting professional advertisers to design adverts for social media. We test if a personified user group (PUG), when provided to online ad designers, results in better ad performance than when using a non-personified user group (NUG) that had no face picture or name. Our experiment has 30 participants that created Facebook ads using both PUG and NUG. We found that using PUG did increase advertising click performance of ads created by people who are more experienced with ads and personas. Moreover, an analysis of the ad texts showed that the use of PUG increased the empathy of the created ads, supporting the foundational empathy benefit cited in HCI literature. However, the use of PUG did not significantly increase purchase intent. The results imply that using PUG for online ad design evokes more empathy and improves click-through performance. More empathetic ads can have a positive impact on social media users, given that they appear to increase relevance.