Related Experiment Video
Updated: May 29, 2025
06:53
Creating Virtual-hand and Virtual-face Illusions to Investigate Self-representation
Published on: March 1, 2017
13.2K
The Efficacy of Conversational AI in Rectifying the Theory-of-Mind and Autonomy Biases: Comparative Analysis
Marcin Rządeczka1,2, Anna Sterna3, Julia Stolińska1
1IDEAS NCBR, Warsaw, Poland.
JMIR Mental Health
|February 7, 2025
Summary
General-purpose AI models demonstrated superior performance over therapeutic chatbots in identifying and rectifying cognitive biases in mental health interactions. These advanced models show greater adaptability and accuracy in recognizing user affect and biases.
Area of Science:
- Human-Computer Interaction
- Artificial Intelligence in Mental Health
- Cognitive Science
Background:
- Conversational AI is increasingly used in mental health interventions, requiring evaluation of its efficacy in addressing cognitive biases and affect recognition.
- Cognitive biases can worsen mental health conditions like depression and anxiety by reinforcing maladaptive thought patterns in human-AI interactions.
Purpose of the Study:
- To compare the effectiveness of therapeutic chatbots (Wysa, Youper) against general-purpose language models (GPT-3.5, GPT-4, Gemini Pro).
- To assess their capabilities in identifying and rectifying cognitive biases and recognizing affect within user interactions.
Main Methods:
- Utilized constructed case scenarios simulating user-bot interactions to evaluate bias rectification.
- Assessed specific cognitive biases: theory-of-mind (anthropomorphism, overtrust, attribution) and autonomy (illusion of control, fundamental attribution error, just-world hypothesis).
- Responses were scored for accuracy, therapeutic quality, and adherence to cognitive behavioral therapy principles by cognitive scientists and a clinical psychologist.
Main Results:
- General-purpose models outperformed therapeutic chatbots in rectifying cognitive biases, notably overtrust, fundamental attribution error, and just-world hypothesis.
- GPT-4 achieved the highest scores, while Wysa scored the lowest; general-purpose bots demonstrated superior accuracy and adaptability.
- General-purpose chatbots also excelled in affect recognition, adapting quicker to emotional nuances and outperforming therapeutic bots in 67% of tested biases.
Conclusions:
- Current therapeutic chatbots have limitations in addressing cognitive biases and simulating empathy effectively.
- Enhanced AI systems need to analyze and rectify biases as integral to human cognition for precision and simulated empathy.
- Future research must improve simulated emotional intelligence, address ethical concerns, and ensure safe, effective AI mental health support.
Keywords:
AIaffect recognitionartificial intelligencebias rectificationchatbotscognitive biasconversational artificial intelligencedigital mental healthMore Related Videos
Related Concept Videos
01:37Reason and Intuition
6.3K
The human brain processes information for decision-making using one of two routes: an intuitive system and a rational system (Epstein, 1994; popularized by Kahneman, 2011 as System 1 and System 2, respectively). The intuitive system is quick, impulsive, and operates with minimal effort, relying on emotions or habits to provide cues for what to do next, while the rational system is logical, analytical, deliberate, and methodical. Research in neuropsychology suggests that the...
6.3K
01:20Introduction to Cognitive Psychology
407
Cognitive psychology is the field of psychology dedicated to examining how people think. It attempts to explain how and why we think the way we do by studying the interactions among human thinking, emotion, creativity, language, and problem-solving, as well as other cognitive processes. Cognitive psychology studies how information is processed and manipulated in remembering, thinking, and knowing.
This field emerged in the mid-20th century, following a period dominated by behaviorism, which...
This field emerged in the mid-20th century, following a period dominated by behaviorism, which...
407
02:16Stereotype Content Model
13.9K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
13.9K
02:34Empathy
9.5K
Some researchers suggest that altruism operates on empathy. Empathy is the capacity to understand another person’s perspective, to feel what he or she feels. An empathetic person makes an emotional connection with others and feels compelled to help (Batson, 1991). Empathy can be expressed in several ways, including cognitive, affective, and motor.
9.5K
01:18Cognitive Development During Adolescence
40
During adolescence, individuals experience significant cognitive development that enhances their understanding of others' emotions and thoughts, known as cognitive empathy. This period is marked by an increased ability to adapt to others' perspectives and a more nuanced understanding of others' mental states, a skill that is foundational for social problem-solving and conflict avoidance. The development of cognitive empathy relies heavily on the theory of mind — the...
40
01:24Revisionist Views of Adolescent and Adult Cognition
23
A revisionist approach to Jean Piaget's theory of cognitive development has brought new insights that challenge and reinterpret his established ideas. Piaget proposed that the formal operational stage, emerging in adolescence, represents the culmination of cognitive maturity. During this stage, individuals are said to develop abstract thinking, engage in systematic problem-solving, and show a form of egocentrism, believing others are as preoccupied with their behavior as they are...
23