Artificial intelligence and handwriting analysis: A study on gender and sexual orientation identification through machine learning algorithms

Introduction: The integration of artificial intelligence (AI) into handwriting analysis introduces significant neuroethical concerns, particularly when it is used to identify sensitive personal traits. Handwriting, a motor skill linked to brain function, can serve as a biometric indicator, but its use in this context raises ethical questions around privacy, bias, and consent. This study addresses the ethical implications of using AI, specifically convolutional neural networks (CNNs), for handwriting analysis, with a focus on gender and sexual orientation identification. Aim: The study aims to replicate previous findings on gender identification through handwriting and to introduce novel evidence regarding the identification of sexual orientation using AI. Additionally, it seeks to compare the accuracy of AI models with human evaluators and explore the broader neuroethical concerns related to privacy and the ethical use of biometric data. Methods: Handwriting samples were collected from 107 males, 201 females, and subsets of individuals based on their sexual orientation. A CNN model was trained on 80% of the data and tested on the remaining 20%. The results were compared to evaluations made by human judges. The accuracy of gender and sexual orientation identification was measured. Results: The CNN achieved an accuracy of 81% in identifying gender, surpassing human evaluators. For sexual orientation identification (heterosexual vs homosexual orientation), the results were more variable, with accuracy rates of 63% for handwriting samples from males and 69% for those from females. These findings demonstrate the potential of AI in biometric analysis but also highlight the limitations, particularly in more complex tasks like sexual orientation identification. Conclusions: This research contributes to the growing field of neuroethics by exploring the intersection of AI and sensitive personal data. While AI can offer improved accuracy in tasks such as personal identification, its use in areas like gender and sexual orientation raises ethical concerns about privacy, data protection, and the risk of reinforcing societal biases. The study underscores the need for robust ethical frameworks to govern AI applications in biometric analysis, ensuring that advancements in technology are balanced with the protection of individual rights.