This article is part of a Special Section of On Board with Professional Psychology that focuses on the intersection of professional psychology and Artificial Intelligence (AI). Learn more about ABPP’s Artificial Intelligence Taskforce.
Personality assessment is a cornerstone of clinical practice. It evolves as we continue striving to capture and articulate the complexity of human individuality. Personality assessment has been dominated by nomothetic methods based on normative data because they offer empirical validity and reliability. However, these approaches fail to capture the unique and nuanced experiences of the individual. This distinction fuels the ongoing debate between generalizability and specificity in personality assessment.
Artificial Intelligence (AI) can reshape the integration of idiographic and nomothetic personality assessments by addressing the challenges associated with narrative-based measures. AI programs like ChatGPT-4o can be used for coding, summarizing, interpreting, and integrating data, allowing for the efficient and objective extraction of thematic information from narrative data. This capability provides deeper insights into an individual’s lived experience while mitigating biases in traditional scoring and interpretation methods. Moreover, AI’s critical analysis can enhance the validity and reliability of these assessments, aligning them with scientific standards and paving the way for new possibilities in idiographic assessment.
AI can also address the utility problem often associated with idiographic methods by making them more practical for widespread use. By streamlining complex coding systems, AI makes narrative assessments more accessible to office-based psychologists and reduces the need for extensive training and scoring. Our initial explorations used ChatGPT-4o by Open AI, an artificial intelligence chatbot that uses natural language processing to create human-like conversational dialog. Our initial research suggests that ChatGPT-4o can code ratings scales as accurately as humans.
An AI-assisted personality assessment protocol involves preparing psychological test data, including computerized test reports, interview transcripts, self-descriptions, and narrative-based psychological tests. Teaching guides for the AI, such as coding systems like SCORS-G or EMQ (Acklin et al., 1991), are uploaded and analyzed with case data. This process is guided by queries that instruct the AI to summarize, interpret, and integrate assessment data. This approach is particularly effective for analyzing narrative-based measures like the Thematic Apperception Test (TAT), interview data (e.g., McAdams Life Story Interview, 2008, 2013), early memories, and free speech samples. Follow-up questions or prompts allow for hypothesis testing and linking multi-method idiographic findings with nomothetic assessment data. The protocol adheres to the scientific method of discovery and validation, examining the subject’s self-understanding, internal conflicts, relationships with self and others, and related concepts.
While no specific theory of personality is necessary, the methodology is highly amendable to applying psychodynamic theory, structural developmental theory, McAdams’s narrative identity theory and McClelland’s needs theory, object relation theory, thematic analyses, and SCORS-G to test data.
In the case example below, narrative-based test data was examined by analysis of themes, structural organization, human relationships, and signature events or experiences.
Case Example: MR is a 20-year-old male referred for a pre-seminary mental health evaluation.
Tests used for ChatGPT-4o Analysis: MR was administered the Early Memories Questionnaire (EMQ) and Washington University Sentence Completion Test (WUSCT). Ethical precautions included using a paid subscription for ChatGPT-4o, research and consultation about the security of the subscription plan, omitting names or compromising demographic information from the test data, and fact-checking outputs. The following case analysis does not include clinical interviews or the results of self-report personality measures (e.g., adjective checklists, MMPI-3, MCMI-IV) given the limited space available.
ChatGPT-4o Prompts: The application requires an iterative conversational process using primary and follow-up prompts. The prompts for this sample application follow.
- Primary Prompt: Attached is MR’s earliest memories of himself, his mom, and his dad. Conduct an analysis of his memories using psychodynamic theory.
- Follow-up Prompt #1: Identify the protective factors and, if any, psychological weaknesses indicated in these early memories.
- Follow-up Prompt #2: Here is more information about MR from the Washington University Sentence Completion Test. Provide an integrated interpretive summary of MR’s memory analysis, protective factors, and psychological weaknesses with his ego development levels. Use paragraph form. Do not use bullet points.
ChatGPT-4o Output – Integration and Formulation (limited to 250 words): MR’s personality reveals a complex interplay of altruism, mistrust, and high self-imposed standards, coupled with ongoing ego development. His most prominent characteristic is his self-sacrificing nature, where he prioritizes the needs of others, often at the expense of his own well-being. This behavior is driven by a deep sensitivity to others’ pain and a desire to avoid feelings of guilt or selfishness. However, this selflessness may also lead to inner conflicts as he struggles to balance his own needs with those of others.
In contrast, MR also harbors significant mistrust towards others, stemming from past experiences where he feels he has been mistreated or exploited. This mistrust manifests in his belief that others may have ulterior motives, leading him to be cautious and sometimes overly suspicious in his interactions. This defensive stance may serve as a psychological barrier, protecting him from potential harm but also limiting his ability to form close, trusting relationships.
MR’s internal drive is further complicated by his unrelenting standards, where he feels a constant pressure to achieve perfection in his endeavors. This perfectionism, while motivating him to excel, can also lead to feelings of inadequacy and stress, as he often perceives his efforts as falling short of his own expectations.
MR’s ego development is in transition between conformist and self-aware stages. While he adheres to societal rules and seeks approval, he is also beginning to explore his own identity and personal values. This ongoing development suggests that MR is in the process of integrating his self-concept with his broader understanding of the world around him.
Ethical Considerations
The application of AI methods to personality assessment presents complex ethical considerations. Issues of privacy, data security, and the potential for algorithmic bias must be addressed. Use of generative AI in practice must adhere to ethical standards and laws. Subject data must be properly de-identified. Additionally, there is a need for ongoing research to validate the reliability and validity of AI-assisted personality assessments across diverse populations. Large Language Models (LLM) are shaped by the preexisting knowledge they have been trained on, which can hold historical and societal biases. For this reason, generated output of LLM’s need to be checked for bias and de-biased when possible (Banker et al., 2024).
The application of AI methods is not a substitute for the human clinician. Instead, it is to be used as an assistive tool with awareness of its strengths and limitations. Inexperienced or remiss users are susceptible to declaring AI’s output as factual without critical review and fact-checking. All users must be wary of confirmation and anchoring biases that may skew clinical judgment.
Conclusions
AI methods will lead to a revival of the relevance and effectiveness of narrative approaches to personology and idiographic personality assessments, like the TAT, by capturing intricacies of individual personalities where nomothetic measures fall short. Such a development would have a widespread impact on personality assessments in clinical settings such as private practices, research, and forensic evaluations.
With the advent of AI, the field of personality assessment stands at a pivotal juncture. The immediacy and power of these applications bring the individual personality to life. AI presents the possibility of bridging the nomothetic-idiographic divide, breathing new life into narrative-based personality methods, restoring personology to clinical assessment practice, and strengthening the science-practitioner foundation for personality assessment. The approach promises to revolutionize 1st person natural language personality assessment and research. Although we have focused on ChatGPT-4o, there are a plethora of other applications (e.g., www.poe.com). As we continue exploring these possibilities, we may find renewed vitality and excitement for integrative narrative-based idiographic personality assessment.
References
Acklin, M. W., Bibb, J. L., Boyer, P., & Jain, V. (1991). Early memories as expressions of relationship paradigms: A preliminary investigation. Journal of Personality Assessment, 57(1), 177–192. https://doi-org.eres.library.manoa.hawaii.edu/10.1207/s15327752jpa5701_19
Banker, S., Chatterjee, P., Mishra, H., & Mishra, A. (2024). Machine-assisted social psychology hypothesis generation. American Psychologist, 79(6), 789–797. https://doi.org/10.1037/amp0001222
McAdams, D. P. (2008). The life story interview. The Foley Center for the Study of Lives: Northwestern University.
McAdams, D. P. (2013). The psychological self as actor, agent, and author. Perspectives on Psychological Science, 8(3), 272–295.
https://doi-org.eres.library.manoa.hawaii.edu/10.1177/1745691612464657
SCORS-G Social Cognition and Object Relations Scale-Global Rating Method Supplementary materials. https://www.scors-g.com/supplemental-scorsg-materials
Marvin W. Acklin, PhD, ABPP
Board Certified in Clinical Psychology and Forensic Psychology
Correspondence: acklin@hawaii.edu
Erica Vargas, MA
Correspondence: Erica.vargas@student.chaminade.edu
Kailey Topping, BA
Correspondence: kailey.topping@student.chaminade.edu