Scaling Artificial Intelligence in Health

4. Scaling fast while doing no harm

AI is not a panacea for health. It is a powerful tool for addressing some of the most complex and urgent challenges facing health systems today. AI alone cannot be expected to revolutionise healthcare. It is the policies, governance structures, and collective societal choices that will determine whether AI will become a driver of equitable, safe, and sustainable health impacts.

In this report, four key pillars and nine key policy areas were explored encompassing 42 policy questions at the intersection of AI and health. Several OECD countries are adopting legislation, strategy, and action plans to help address the tensions between market forces, the Hippocratic oath, and reaching every person – scaling fast and doing no harm.

Chapter 3 investigated OECD Member’s progress across six of the nine policy areas (summarised in Table 4.1). This finds that there is a strong focus on the foundations of data and AI, but less attention is paid to the people behind and impacted by these technologies.

Table 4.1. Readiness for the responsible scale of AI in health by select policy areas

Policy Checklist Pillars	Policy Areas	Yes	Partly	No	N/A
Enabling Practices	Better use of health data	72%	10%	17%	2%
	Enabling the use of AI	20%	32%	38%	10%
	Capacity and capability among the health workforce	29%	47%	16%	8%
Guardrails	Agreed objectives	58%	16%	25%	1%
	Oversight, measurement and monitoring	18%	18%	55%	8%
Engagement	Public engagement	41%	13%	42%	4%

Note: Cells are coloured green where > 66%; orange when between 65‑34%; grey when < 33%

Source: Based on OECD Survey on Health Data Governance, the Impact of Artificial Intelligence on the Health Workforce, and Interoperability. This table summarises the analyses from Table 3.9.

This found that 72% of OECD countries have enacted policies supporting the better use of health data. In contrast, only 29% have acted on workforce capacity and capabilities, meaning that as AI solutions are approved, they will be faced with challenges achieving their optimal impact.

Oversight of AI solutions also lags, with only 18% of countries demonstrating policy actions in this area. Without effective oversight, it is likely that AI in health solutions will continue to be developed in siloes, be challenged to scale, and fail to achieve their potential.

Further progress is needed in strengthening the enabling conditions for AI adoption, with only 20% showing activity. OECD Members have shown active progress in establishing strategies that provide agreed upon objectives (58%) and advancing public engagement (41%); however, without plans for adoption and scale the value of AI will be limited.

A collaborative and focussed effort to strengthen policies would meet future challenges for the responsible scale of AI in healthcare and optimises human and economic value.

Beyond the application of the AI in Health Policy Checklist, there are also opportunities for collaboration. As seen with the strategies at the intersection of AI in health, countries have taken different approaches based on their local contexts, but compatibility of policies across borders will help accelerate the responsible scale of AI.

Hence, there is urgency to act together for health systems and align on the responsible scaling and use of AI in healthcare. There are opportunities to learn from each other to recognise leading practices and to develop approaches to policy that are compatible across borders. Guided by the OECD AI in Health Policy Checklist, countries could design their solutions in a way that enables local autonomy while simplifying the ability to collaborate. It will take time, leadership, will, effort, and investment to achieve and sustain benefits from AI in health. Policymakers can proactively shape the evolution of AI in health policies so that advances can scale fast while doing no harm.