AI Likely to Improve Health Care, Research Shows—But Not for Blacks and Ethnic Minorities

Key Takeaways

• AI scheduling algorithms cause 33% longer wait times for Black patients
• Risk models underestimate illness severity for Black patients using cost proxies
• Dermatology AI underperforms on darker skin due to biased training data
• Mental‑health chatbots show lower accuracy for Black users
• Inclusive design and diverse data can mitigate AI‑driven health inequities

Pulse Analysis

Health‑care providers are embracing AI at unprecedented speed, with 84% of insurers reporting use for fraud detection, utilization management and prior‑authorization automation. Consumer adoption mirrors this trend: over 40 million global users turn to chatbots like ChatGPT for health advice daily, and roughly one‑third of U.S. adults consult AI tools for medical information. These efficiencies promise lower costs and faster decision‑making, but the rapid rollout often outpaces rigorous validation, especially for vulnerable populations.

A growing body of research highlights how AI can amplify racial and ethnic disparities. A 2024 systematic review of 30 studies found AI‑driven scheduling systems extending Black patients’ wait times by 33%, while cost‑based risk algorithms misclassify Black patients as less ill despite higher disease burden. Diagnostic models trained predominantly on lighter‑skinned images miss conditions on darker skin, and language‑focused mental‑health bots underperform for Black users, detecting only 10% of suicides compared with 62% for White patients. These biases arise from historical data gaps, proxy variables that encode systemic inequities, and the continued reliance on race‑adjusted clinical formulas.

Mitigating bias requires a multi‑layered approach: diversifying training datasets, embedding equity checkpoints into model development, and fostering a heterogeneous data‑science workforce. Policy initiatives such as the Coalition for Health AI’s equity framework and the Encoding Equity alliance aim to standardize transparent, bias‑audit processes. However, recent federal actions under the Trump administration have rolled back equity mandates, challenging state‑level AI regulations. For AI to fulfill its promise in health‑care, stakeholders must prioritize inclusive design, continuous monitoring, and robust governance that aligns technological innovation with the goal of health equity.