Turkish Scientists Develop Groundbreaking AI for Behavior Analysis

Obaid Ahsan

0 Comment


Ankara, Turkey – In a breakthrough, scientists at Bilkent University have developed an artificial intelligence (AI) technology that can detect signs of depression and analyze personality based on human behavior.

The researchers, led by Assistant Professor Hamdi Dibeklioğlu, used machine learning techniques to create algorithms that can discern depression levels and personality traits by analyzing speech content, voice, facial expressions, and body language. This AI system processes the same visual and auditory cues psychiatrists use to make clinical observations.

Technology Termed “Affective Computing”

Dibeklioğlu terms this field “affective computing” and notes the recent surge of interest due to the popularity of AI chatbots like ChatGPT. His team has collaborated with psychiatrists to develop ethical protocols, secure patient consent, and protect sensitive data.

The system aligns with clinical theories on social withdrawal and other behavioral patterns in depression. It autonomously makes diagnoses, sometimes detecting nuances overlooked during human analysis.

Lie Detection and Personality Assessment

In separate research, Dibeklioğlu’s team also analyzed deception levels by processing visual cues, vocal tones, and sentences. Though error rates mean this cannot be definitively used in legal settings, it may assist in interviews.

The researchers are also working on AI for personality assessment across various dimensions and pain level detection, which is particularly beneficial for children. Dibeklioğlu stresses the need for meticulous ethical approvals in applying such sensitive AI systems.

Developers Emphasize Ethical Use and Human Oversight

The innovations at Bilkent University demonstrate AI’s potential in behavioral analysis and mental healthcare. However, the developers emphasize the importance of human oversight and accountability in deploying such technologies.


Post Comments:

Leave a comment

Your email address will not be published. Required fields are marked *