Computational personality assessment

Stachl, Clemens; Boyd, Ryan L.; Horstmann, Kai T.; Khambatta, Poruz; Matz, Sandra; Harari, Gabriella M.

doi:10.5964/ps.6115

Cited by 21 publications

(11 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, self-reports are subject to a myriad of methodological influences (Vaerenbergh & Thomas, 2013) and intentional faking (Goerigk, Hilbert, Jobst, Falkai, Bühner, Stachl, Bischl, Coors, Ehring, Padberg, & Sarubin, 2020). New approaches have been developed to assess psychological phenomena and psychopathology combining objective data on people's digital behavioural and online footprints with machine learning (Insel, 2017;Stachl, Boyd, Horstmann, Khambatta, Matz, & Harari, 2021). Text data in particular has long been considered as a valid source of information on psychological processes (Allport, 1942).…”

Section: Digital Behavioural Data For Assessing Psychological Constructsmentioning

confidence: 99%

Against Algorithmic Exploitation of Human Vulnerabilities

Strümke¹,

Slavkovik²,

Stachl³

2023

Preprint

View full text Add to dashboard Cite

Decisions such as which movie to watch next, which song to listen to, or which product to buy online, are increasingly influenced by recommender systems and user models that incorporate information on users' past behaviours, preferences, and digitally created content. Machine learning models that enable recommendations and that are trained on user data may unintentionally leverage information on human characteristics that are considered vulnerabilities, such as depression, young age, or gambling addiction. The use of algorithmic decisions based on latent vulnerable state representations could be considered manipulative and could have a deteriorating impact on the condition of vulnerable individuals. In this paper, we are concerned with the problem of machine learning models inadvertently modelling vulnerabilities, and want to raise awareness for this issue to be considered in legislation and AI ethics. Hence, we define and describe common vulnerabilities, and illustrate cases where they are likely to play a role in algorithmic decision-making. We propose a set of requirements for methods to detect the potential for vulnerability modelling, detect whether vulnerable groups are treated differently by a model, and detect whether a model has created an internal representation of vulnerability. We conclude that explainable artificial intelligence methods may be necessary for detecting vulnerability exploitation by machine learning-based recommendation systems.

show abstract

Section: Digital Behavioural Data For Assessing Psychological Constructsmentioning

confidence: 99%

Against Algorithmic Exploitation of Human Vulnerabilities

Strümke¹,

Slavkovik²,

Stachl³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Instead of using self-report measures to assess people's psychological traits, predictive algorithms can estimate these traits directly from digital footprints (i.e. the prediction of a target's psychological traits, [14][15][16]. For example, people's personality traits have been successfully predicted from a wide range of data sources, including Facebook Likes (15,17), language (18)(19)(20)(21), profile pictures (22), credit card spending (23,24), and smartphone sensing (25).…”

Section: Introductionmentioning

confidence: 99%

The Potential of Generative AI for Personalized Persuasion at Scale

Matz¹,

Teeny²,

Vaid³

et al. 2023

Preprint

View full text Add to dashboard Cite

Matching the language or content of a message to the psychological profile of its recipient (i.e., personalized persuasion) is widely considered as one of the most effective messaging strategies. We demonstrate that the rapid advances in large language models (LLMs), like ChatGPT, could accelerate this influence significantly by making personalized persuasion scalable. Across five studies (N=1,171) with different topics (i.e., consumer products, social issues, and health advice) and psychological profiles (i.e., personality traits, political ideology, and moral foundations), we show that personalized messages designed by LLMs exhibit significantly more influence than non-personalized controls. Our findings are among the first to demonstrate the potential for LLMs to automate, and thereby scale, the use of personalized persuasion in ways that enhance its effectiveness and efficiency. We discuss the implications of the work for researchers, practitioners, and the general public.

show abstract

“…In addition, future studies on cue processes and the accuracy of personality judgments based on Instagram could be complemented by advances in Big Data, machine learning, and personality computing approaches (i.e., technologies dealing with human personality; Phan & Rauthmann, 2021; Stachl et al, 2021). Cannata et al (2022) proposed how such an integrative approach, combining accuracy and personality computing research, can overcome current limitations in answering open key questions and addressing important challenges of each of the disciplines.…”

Section: Discussionmentioning

confidence: 99%

#Insta personality: Personality expression in Instagram accounts, impression formation, and accuracy of personality judgments at zero acquaintance

Osterholz

Mosel

Egloff

2022

Journal of Personality

View full text Add to dashboard Cite

Objective This study examined personality expression, impression formation, and the consensus and accuracy of zero‐acquaintance personality judgments that were based on people's Instagram accounts. Method Self‐ and informant reports of the Big Five personality traits, self‐esteem, and narcissism were collected for 102 Instagram users. Screenshots were taken of Instagram users' profiles, including up to the 102 latest available Instagram posts. A number of Instagram cues were objectively retrieved, counted, and rated by independent trained cue coders from the screenshots. 100 unacquainted observers then judged the Big Five traits, self‐esteem, and narcissism on the basis of Instagram screenshots only. Results We identified Instagram account characteristics that were associated with users' personality traits (measured with self‐reports, informant reports, and self‐informant composites) and observers' zero‐acquaintance personality judgments. Personality judgments that were based on Instagram accounts demonstrated consensus and significantly converged with Instagram users' Big Five traits, self‐esteem, and narcissism across all three personality criteria. Averaged‐observer accuracy correlations for self‐informant composite scores ranged from r = .44 (p < .001) for extraversion to r = .25 (p = .013) for conscientiousness. Conclusions Our findings provide insight into cue processes of online self‐portrayal and impression formation on Instagram and the level of zero‐acquaintance accuracy.

show abstract

Computational personality assessment

Cited by 21 publications

References 64 publications

Against Algorithmic Exploitation of Human Vulnerabilities

Against Algorithmic Exploitation of Human Vulnerabilities

The Potential of Generative AI for Personalized Persuasion at Scale

#Insta personality: Personality expression in Instagram accounts, impression formation, and accuracy of personality judgments at zero acquaintance

Contact Info

Product

Resources

About