USC Study: AI Therapists Show Strong Communication Skills but Raise Safety Concerns
A USC Viterbi School of Engineering study found that AI large language models (LLMs) show strong performance in general mental health communication, empathy, and specificity. However, the researchers flagged significant safety concerns, as the LLMs provided varying levels of unauthorized or potentially harmful responses in critical situations where professional judgment is essential.
Key quotes
The new USC study suggests that all the large language models (LLMs) they tested showed 'strong performance in general communication, receiving high ratings for overall quality, empathy and specificity from the mental health professionals who evaluated the responses.'
However, the LLMs tested by the USC researchers were flagged for 'providing varying levels of unautho
From the article
You might also wanna read
Risks of Using Large Language Models as Therapists
The article discusses the use of large language models (LLMs) as therapists and highlights the limitations and risks associated with replaci
The Cognitive Risks of Over-Reliance on Large Language Models
The article argues that over-reliance on Large Language Models (LLMs) reduces cognitive load to dangerous levels, potentially causing humans
desunit.com·10mo agoThe Case for Personality in AI Systems: Why Human-Like LLMs Are Good Engineering
The article argues against the view that AI systems should remain purely as tools without human-like personalities. It counters the position
AI Chatbots May Standardize Human Expression and Reduce Cognitive Diversity, Study Warns
A study led by USC researchers warns that AI chatbots and large language models are standardizing human expression, potentially reducing cog
LLMs in Gastroenterology: Balancing Clinical Benefits Against Privacy and Accuracy Risks
This article examines the use of large language models (LLMs) in gastroenterology and hepatology practices. It highlights benefits such as d


Comments
Sign in to join the conversation.
No comments yet. Be the first.