How Developer Personality Shapes the Use of AI Coding Assistants

Annemarie Wittig, Alina Mailach, Norbert SiegmundarXiv 2026 · arXiv

TL;DR

Personality isn't just for HR—it fundamentally changes how developers interact with AI. This paper provides the first empirical evidence that Big Five traits like Extraversion and Agreeableness shape whether a developer finds an AI coding assistant helpful or a hindrance, suggesting that the future of AI tools lies in personality-aware personalization.

What Problem Does This Paper Address?

AI coding assistants like GitHub Copilot and ChatGPT have seen meteoric adoption, with some reports suggesting up to 90% of tech professionals use them. However, these tools are currently "one-size-fits-all." While some developers see massive productivity gains, others find the tools disruptive or untrustworthy.

Current research often focuses on the AI’s personality (e.g., making a chatbot sound "friendly"). This paper flips the script to look at the user’s personality. The authors argue that we are currently building tools without an underlying theory of who needs what. By understanding how a developer's innate traits (like how organized or outgoing they are) affect their experience, we can move away from "trial and error" personalization toward tools that adapt to the individual.

Methods

The researchers tackled this in two stages:

  1. Systematic Scoping Review: They surveyed existing literature across general AI and Software Engineering (SE) venues to see if personality-perception links had been studied. They identified 12 relevant papers (though none were specific to the SE context).
  2. Two-Wave Empirical Survey: They conducted a longitudinal study with professional developers across three companies.
  3. Wave 1 (Feb-Apr 2025): 29 developers.
  4. Wave 2 (Apr 2026): 11 developers (a follow-up one year later).
  5. Measurement: They used the Big Five Model (BFI-2-S) to measure Extraversion, Agreeableness, Conscientiousness, Negative Emotionality (Neuroticism), and Open-Mindedness. They then used multiple linear regression to link these traits to developers' expectations and actual experiences with AI tools.

The study structure is summarized in Figure 1: Fig.

  1. Study Structure

![Fig.

  1. Study Structure](/figures/art_4c6a0d96a5ad4334a07922e267673606)

Datasets

  • Literature Corpus: 12 peer-reviewed papers identified from an initial pool of over 9,300 candidates (see Figure 2 for the selection process).
  • Survey Data: Primary data from 29 professional developers (Wave 1) and 11 returning participants (Wave 2). The participants had between six months and 20 years of experience and primarily used GitHub Copilot and ChatGPT.
  • Replication Package: The authors provided a comprehensive package on Zenodo including survey data and analysis scripts.

![Fig.

  1. Scoping review structure and intermediate results](/figures/art_bc64d610ae9f46808f11560a3cd0717a)

Key Insights and Ideas

  • Experience Trumps Expectations: Interestingly, personality traits did not predict what developers expected from AI tools. However, traits were significant predictors of their actual lived experience. This suggests that personality effects manifest through interaction rather than preconceived notions.
  • The Extraversion Edge: Higher Extraversion was consistently associated with more positive experiences, particularly in private programming projects. This aligns with broader AI research suggesting extraverts are more open to new technology.
  • The Agreeableness Evolution: In the first wave, Agreeable developers reported more negative professional experiences (specifically feeling "insufficiently prepared"). However, by the second wave a year later, Agreeableness was linked to positive experiences, particularly regarding code quality and teamwork. This suggests that personality-based reactions to AI might evolve as the developer gains experience with the tool.
  • Context is King: Associations were stronger for "private use" than "professional use." The authors hypothesize that organizational constraints and office culture might "mask" personality effects at work, whereas developers' true preferences shine through when they use AI for personal projects.

Key Results

  • Scoping Review: 11 out of 12 reviewed papers found that at least one Big Five trait was associated with AI perception, reinforcing that personality is a relevant (but context-dependent) predictor.
  • Quantitative Findings (Survey 1):
    • Extraversion was a significant positive predictor for private AI experience ($\beta = 0.34, p = .04$).
  • Agreeableness was associated with higher negative-experience scores in professional settings ($\beta = 0.58, p = .03$).
  • Quantitative Findings (Survey 2):
    • The model for private AI experience showed an extremely high fit ($R^2 = 0.93, \text{adjusted } R^2 = 0.87$).
  • Extraversion remained a strong positive predictor ($\beta = 0.46, p = .01$).
  • Open-Mindedness emerged as a significant positive predictor ($\beta = 0.26, p = .04$).

Limitations and Future Work

  • Sample Size: The study is exploratory. With only 29 (and later 11) participants, the statistical power is limited. The authors emphasize these are "tentative propositions" rather than definitive proof.
  • Self-Selection Bias: Participation was voluntary, which might attract developers with stronger opinions (positive or negative) about AI.
  • Future Directions: The authors call for larger-scale longitudinal studies to build a formal "grounded theory" of developer personality. They suggest that tool builders should explore "personality-aware designs," such as adaptive interfaces that offer different levels of initiative based on the user's profile. For organizations, the takeaway is clear: uniform, one-time rollouts of AI tools may fail because different developers need different types of support.

Summary updated

Abstract

Developers differ in personality traits that shape how they approach software development tasks, yet AI coding assistants either behave uniformly or offer personalization features whose fit with individual developers must be found through trial and error. This is because it remains unclear how personality shapes developers' experiences with AI assistants, and thus which traits personalization should account for. In other words, we provide tools with no underlying theory on what works when and why. We take a first step towards such a theory with a systematic scoping review on how users' personality relates to their perception of LLM-based tools. We identified twelve studies through general search engines, major Software Engineering (SE) venues, AI tooling, and snowballing. They all reported associations between personality and perceptions, particularly perceived usefulness, yet none addressed this in a SE context. Building on these findings, we conducted an exploratory two-wave survey with 29 and 11 developers, collecting first empirical evidence of how developers' personality (measured using the Big Five Model) relates to the perception of AI coding assistants. Higher Extraversion was associated with more positive private-use experiences in both waves, while Agreeableness was associated with negative professional experiences in the first wave and positive ones in the second. These results provide first indications that the personality effects found in the literature extend to developers. Although further studies with bigger samples are needed to build a theory, our work takes the first step towards an empirical grounded theory of developer personality and AI personalization traits.
BibTeX
@misc{wittig2026how,
  author = {Annemarie Wittig and Alina Mailach and Norbert Siegmund},
  title = {{How Developer Personality Shapes the Use of AI Coding Assistants}},
  year = {2026},
  eprint = {2611.21240},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.21240},
}
Figure