ChatGPT Surpasses Average Human Score in Emotional Intelligence Testing

OpenAI’s ChatGPT-4 exceeds average human EQ in new study by Psychometrica, underscoring potential to mimic emotional response and raising concerns over use.

Source: news release

July 5, 2023

GEORGE, SOUTH AFRICA, July 5, 2023 — OpenAI’s advanced language model, ChatGPT-4, has scored higher than the average human adult in a standardized Emotional Intelligence (EQ) test, according to a study conducted by the non-profit organization, Psychometrica.

ChatGPT-4 scored an overall 117 on the EQ test, surpassing the normalized average of 100 for the adult population and achieving “Gold Heart” status on the test’s descriptive result scale. Individual scores across the different dimensions of EQ were: Self-awareness: 105, Social awareness: 123, Self-management: 122, Self-motivation: 117, and Relationship management: 116. This performance suggests a particularly strong ability in social awareness and self-management.

The study was a preliminary investigation to evaluate the capability of modern AI in appropriately responding to users on sensitive topics involving emotional understanding. The authors note, “As AI models like ChatGPT-4 continue to gain widespread usage in advising on personal life, career, and academic decisions, evaluation of their applied Emotional Intelligence becomes more important. The increasing prevalence of AI in consumer applications underscores the importance of this research, with potential implications for users seeking advice on mental health issues.”

This test was completed on June 23 to assess the apparent EQ of modern artificial intelligence (AI). The test questions were prompted to ChatGPT-4 and evaluated using the Psychometrica free EQ test web application at https://psychometrica.org/tests/eq-test, which is identical to the Android and Apple iOS mobile versions.

The researchers emphasize that while ChatGPT-4 was capable of interpreting and responding to emotional prompts, this does not suggest that the AI experiences genuine emotional states. And while ChatGPT-4 expresses a high degree of apparent Emotional Intelligence, Psychometrica urges caution when interpreting or acting based on the model’s output, as the appearance of emotional understanding or connection may entice users into inappropriately trusting its output.

The full ChatGPT EQ study summary is available at: https://psychometrica.org/research-emotional-intelligence-eq-test-chatgpt/

About OpenAI:

OpenAI (https://openai.com), founded in December 2015, is an artificial intelligence research lab comprised of both for-profit OpenAI LP and its parent company, the non-profit OpenAI Inc. The organization’s mission is to ensure that artificial general intelligence (AGI) benefits all of humanity. OpenAI is the developer of the GPT AI model and the ChatGPT product.

About Psychometrica:

Psychometrica (https://psychometrica.org) is a non-profit organization specializing in psychometric testing and peer-to-peer support groups. Founded on the principle of using psychometric evaluations to increase self-awareness and facilitate self-improvement, the organization provides free standardized tests (including IQ tests, EQ tests, and personality tests) that are employed globally for educational guidance, professional development, relationship improvement, and research. For inquiries related to this study please contact: research@psychometrica.org

This article is not professional healthcare advice. Please consult your physician for personalized medical advice.

New Health Journal (NHJ) reports developments in health research, health trends, and wellness options. Our nonprofit mission is to advance public health and wellness knowledge by publishing concise digests of useful and interesting health topics.