It seems we have found the reason why you are not getting a call back after the interview.
Researchers tested whether it's possible to automatically assess personality traits and link them to career outcomes using a simple facial photograph. They used a large sample of MBA graduates' profiles from LinkedIn and ran the images through a machine learning model that produces scores on the Big Five personality scale. These include openness to experience, conscientiousness, extroversion, agreeableness, and neuroticism. These parameters have long been used in applied psychology and in hiring.
The authors specifically highlight the limitations of this approach and do not propose using it as a practical selection tool. Attempting to infer personality traits from appearance, in their view, inevitably leads to discrimination , because decisions begin to depend on biological traits rather than actual skills and experience. However, the researchers note existing practices. Admissions committees and HR departments have long used personality questionnaires and behavioral tests, and services that automatically analyze video interviews and questionnaires are rapidly spreading. Testing such technologies in an academic setting, they argue, is necessary to ensure that discussions of potential restrictions and rules are based on measurable data rather than guesswork.
The technical basis was an algorithm from a 2020 scientific publication . In that study, the researchers trained the model using a combination of two types of data. On one side, they used standard portrait photographs, while on the other, they used self-assessment results from psychological questionnaires. The neural network looked for statistical correspondence between facial features and how the participants described their personalities. The new paper specifically clarifies the logic of the method. The system attempts to reconstruct a person's self-reported responses, not the impressions of outside observers. This means that it essentially predicts self-reports based on the Big Five personality scale, not a subjective visual assessment from an outside observer.
The approach itself has already sparked considerable controversy in the scientific community. A 2024 paper categorized similar studies as a trend where machine learning methods are applied to weak initial hypotheses to create a semblance of robustness. Critics point out that even a sophisticated statistical model fails to correct underlying assumptions if they are inherently questionable. The authors of the new study acknowledge this point but are still testing it on a large sample of real-world profiles to quantitatively assess whether such an algorithm exhibits measurable predictive power in tasks related to education and careers.
Researchers compiled a database of over 96,000 photographs of business school graduates from LinkedIn. Each photo was run through a model that assesses personality traits using the Big Five scale, and these assessments were then linked to profile data, education, initial salary, subsequent income dynamics, and job transitions. These calculated characteristics were found to provide additional guidance when predicting career outcomes. Using these parameters, the model more accurately predicted the educational level, starting salary, and subsequent income changes than when analyzing without this parameter.
The authors specifically examine how such tools could be used in managerial selection. If HR integrates such image analysis, the result would be a formal prediction of a candidate's likely career performance. The researchers warn that such assessments will inevitably be biased and will impact different groups unevenly. However, elements of such solutions are already beginning to appear in automated recruitment services.
Researchers tested whether it's possible to automatically assess personality traits and link them to career outcomes using a simple facial photograph. They used a large sample of MBA graduates' profiles from LinkedIn and ran the images through a machine learning model that produces scores on the Big Five personality scale. These include openness to experience, conscientiousness, extroversion, agreeableness, and neuroticism. These parameters have long been used in applied psychology and in hiring.
The authors specifically highlight the limitations of this approach and do not propose using it as a practical selection tool. Attempting to infer personality traits from appearance, in their view, inevitably leads to discrimination , because decisions begin to depend on biological traits rather than actual skills and experience. However, the researchers note existing practices. Admissions committees and HR departments have long used personality questionnaires and behavioral tests, and services that automatically analyze video interviews and questionnaires are rapidly spreading. Testing such technologies in an academic setting, they argue, is necessary to ensure that discussions of potential restrictions and rules are based on measurable data rather than guesswork.
The technical basis was an algorithm from a 2020 scientific publication . In that study, the researchers trained the model using a combination of two types of data. On one side, they used standard portrait photographs, while on the other, they used self-assessment results from psychological questionnaires. The neural network looked for statistical correspondence between facial features and how the participants described their personalities. The new paper specifically clarifies the logic of the method. The system attempts to reconstruct a person's self-reported responses, not the impressions of outside observers. This means that it essentially predicts self-reports based on the Big Five personality scale, not a subjective visual assessment from an outside observer.
The approach itself has already sparked considerable controversy in the scientific community. A 2024 paper categorized similar studies as a trend where machine learning methods are applied to weak initial hypotheses to create a semblance of robustness. Critics point out that even a sophisticated statistical model fails to correct underlying assumptions if they are inherently questionable. The authors of the new study acknowledge this point but are still testing it on a large sample of real-world profiles to quantitatively assess whether such an algorithm exhibits measurable predictive power in tasks related to education and careers.
Researchers compiled a database of over 96,000 photographs of business school graduates from LinkedIn. Each photo was run through a model that assesses personality traits using the Big Five scale, and these assessments were then linked to profile data, education, initial salary, subsequent income dynamics, and job transitions. These calculated characteristics were found to provide additional guidance when predicting career outcomes. Using these parameters, the model more accurately predicted the educational level, starting salary, and subsequent income changes than when analyzing without this parameter.
The authors specifically examine how such tools could be used in managerial selection. If HR integrates such image analysis, the result would be a formal prediction of a candidate's likely career performance. The researchers warn that such assessments will inevitably be biased and will impact different groups unevenly. However, elements of such solutions are already beginning to appear in automated recruitment services.
