Ideas to balance AI safety and innovation

Editor's Note
While artificial intelligence (AI) tools can inspire innovation and efficiency, they also pose safety risks, especially in a healthcare setting. Therefore, balancing centralization versus decentralization for AI oversight requires the right approach, according to research findings in a December 10 article in Harvard Business Review.

The investigators sought to uncover the dualities and nuances of using AI in any workplace, including in healthcare. For example, they noted that when a team of Polish endoscopists began using AI to detect cancer, their accuracy improved. However, their performance on non-AI procedures got worse.

The researchers gathered studies and lessons from more than 100 people working in healthcare and other industries bringing AI into the workplace. They uncovered AI challenges and offered solutions that might help leaders.

One major example relevant to healthcare today is how to control AI. One school of thought noted is to centralize AI oversight to enforce standards, manage risk, and control access. However, this can delay the application of innovation. Meanwhile, when AI development becomes too decentralized, innovation can outpace integration, resulting in fragmented tools and “digital exhaustion,” the investigators noted.

To build in enough governance around AI without “grinding every good idea to a procedural halt,” the researchers suggested centralizing AI oversight specific to scale and safety such as for data security and compliance, but decentralizing AI oversight for learning and speed, such as for workflow automations.