AI Is More Caring Than Your Clinician? Data Suggests So

⚡ KEY TAKEAWAYS

A large meta-analysis found AI chatbots often rate higher in empathy than clinicians.
AI scored better across multiple specialties on warmth and emotional tone.
The results challenge assumptions about emotional intelligence in healthcare.

Amid its speedy proliferation, artificial intelligence (AI) may already be outperforming human clinicians in empathy, with data suggesting that patients see AI chatbots as warmer, more validating, and more empathic than doctors, nurses, and patient-relations staff.

Indeed, a wide-ranging meta-analysis of 15 studies, covering more than 2,100 live patient interactions and additional specialty-specific trials, has found that patients consistently rated AI as more caring and having better bedside manner than medical staff.

Specifically, across 10-point empathy scales, tools such as ChatGPT regularly outscored licensed healthcare professionals by roughly two points, and in head-to-head comparisons, AI responses had a 73% probability of being judged more empathic.

As the researchers noted:

“In text-only scenarios, AI chatbots are frequently perceived as more empathic than human HCPs.”

Even more strikingly, the findings appear to undermine a confident 2019 UK government claim that emotional intelligence was one core skill that AI could never replicate.

Where AI Outperformed Human Clinicians Most Strongly

As it happens, across nine studies covering mental health, oncology, thyroid disease, autism, email-based clinical communication, and hospital response systems, ChatGPT-4 consistently ranked higher on warmth, emotional sensitivity, and supportive tone.

The empathy gap was particularly large in thyroid surgery Q&A, with 1.42 standard deviations above surgeons, in mental health prompts with 0.97 standard deviations above clinicians, and hospital complaint responses with 2.08 standard deviations above human staff.

Notably, it wasn’t only patient perception that drove the difference. In lupus-related queries, physicians themselves judged AI responses as more empathic than those of their peers. Patient-representative scorers for multiple sclerosis also gave AI higher empathy scores than neurologists.

Dermatology was the only category where human clinicians outperformed chatbots, which might be a clue that empathy perception may vary with problem type rather than being universal across medicine.