The Responsible AI in Healthcare consortium focused on ethics and safety in their first meeting in London after calls to better regulate the use of generative AI in high risk settings.

The consortium is aimed at accelerating the responsible development and use of generative AI technologies through collective learning, experimentation, and policy advocacy and is built around a unique hands-on responsible generative AI testbed to enable members to actively experiment with and refine responsible generative technologies in a real-world healthcare context.

A diverse portfolio of experts across the NHS, Harvard Business School, Turing Institute, St. Edmund’s College at University of Cambridge and industry partners including Trustwise have allowed knowledge-sharing across the AI value chain across academia, policy makers, investors, and healthcare providers.

“We are in the middle of rapid advancements and adoption of generative AI and navigating the responsible AI landscape is proving to be a formidable challenge for all,” said Manoj Saxena, Founder and Chairman at RAI Institute.

“Now, more than ever, we need to work together to make AI safe and aligned with human values. The creation of our Responsible Generative AI Consortium, with its practical testbeds and GenAI Safety Ratings, is a vital step towards our mission of helping AI practitioners to build, buy and supply safe and trusted AI systems.”

The quality of AI systems has been under fire for some time as they can exacerbate issues of bias in data and, in the case of generative AI, their solutions are not auditable or repeatable.

A recent paper in Nature found that in healthcare settings, the use of AI to inform care options resulted in biased care results and options.

The World Health Organisation also called for the safe and ethical use of AI in healthcare settings as companies and governments pushed ahead with technological advancements.

In the UK, AI technology received a £16 million funding boost to fast-track applications into the NHS to speed up diagnoses.

Currently, the Scottish Health Board is testing AI algorithms to see if they improve earlier diagnoses of lung cancer.

Recommended

• How MadeBrave is Fostering Creativity in a Post-Pandemic Workforce
• Scotland’s 10-year Plan For “World-class” Entrepreneurial Campuses
• EHRC: UK’s Proposed AI Regulations “Do Not Go Far Enough”

While AI can offer many benefits to healthcare and potentially speed up diagnosis, the consortium plans to investigate its use to ensure it is safe, ethical, and effective.

The Responsible GenAI Consortium in Healthcare is the first in a series of industry-specific consortia and GenAI testbeds to be launched by The Responsible AI Institute, with others being planned for later this year and next.

The testbeds will incorporate the new Generative AI Safety Rating – a scoring system that grades the safety and reliability of generative AI systems, providing organisations, AI developers, policymakers, investors and other stakeholders a clear measure of system performance, fairness and compliance with rules and regulations.

The rating is based on the evaluation of metrics including but not limited to bias detection, model hallucination, IP and privacy protection, transparency and accountability.

Similar to the FICO credit score model, the numerical representation of the Generative AI Safety Rating will help guide improvements, enable progress tracking and foster a culture of responsible AI development and deployment.