Reshaping the world with responsible AI for health

Consider the elevator. Perhaps you used one today. For most of us, elevators are a convenience that doesn’t require a second thought. For others, they are a critical form of accessibility – whether because of individual mobility needs or because our buildings have grown too tall to climb.

In the modern day, elevators are deeply integrated into our lives, but in the 1800s, elevators were a source of anxiety for many. When Elisha Otis introduced the first safety elevator at the 1856 World’s Fair, the general public feared the potential for injury or death if the elevator fell. Standing atop a model of his design, Otis demonstrated its safety features by cutting the cable. The elevator did not fall.

This demonstration and the subsequent improvements in elevator technology not only made elevators safer but also shifted the public perception of the potential risks. As elevators became more commonplace, they began to shape our world and unlocked our ability to construct buildings that scrape the sky. New York, Dubai, Singapore, Johannesburg: these cities would look remarkably different without Elisha Otis.

Today, artificial intelligence (AI) is facing this same hurdle. Anxiety about the safety and efficacy of AI is prominent in news stories and societal dialogue. At the same time, technologists, ethicists, human rights experts and many others are actively designing safety features to protect us from potential harms. As policymakers, decision-makers and global citizens, it is our job to shepherd this technology towards social good as a means to build trust.

The future of health

AI has numerous applications that range across human activity. Natural language processing, learning models and many other AI tools will change the way we learn, the way we work and the way we promote health and well-being. The digitalisation of health systems created a foundation of data and digital platforms perfect for developing AI solutions.

We are already seeing the impact of AI on drug development, radiology and imaging, and outbreak monitoring, to name a few. Researchers can rapidly analyse vast amounts of data with the support of AI to learn more about diseases and treatments alike. We will gain a new understanding of consciousness and the human body. We will also create new services – from AI-powered chatbots to AI-assisted surgery – that could revolutionise the care available to us.

These advancements must come with balance and protections. We need to understand the trade-offs and the downsides. How data are applied to training models and even the data themselves must be thoughtfully considered to account for bias and discrimination. The risk of ‘AI hallucinations’ becomes more concerning when AI is diagnosing patients with cancer. The road to building trust in AI-enabled solutions for health must come with transparency and alignment with global standards of ethics.

A question of equity

As we grapple with AI technologies’ ethical and technical implications, we must also ensure equity remains central in our work. Addressing biased data is a start, but questions of equity are far-reaching in both the fields of technology and medicine.

We have an opportunity to apply AI to some of the most entrenched health challenges facing communities everywhere. AI-powered imaging can help detect tuberculosis or sequence the genomes of malaria parasites. AI-powered chatbots can serve as a vital lifeline to communities with too few healthcare workers or clinics. Research supported by AI tools can discover possible treatments for neglected diseases. There are many more areas where AI can be a changemaker in our pursuit of the Sustainable Development Goals – but we must do this intentionally.

As we design AI-powered health solutions, we must think beyond ourselves. We must account for low-connectivity settings and communities without regular access to smartphones or even electricity. How these communities benefit from AI is just as important as how any other community will benefit, so we must work with them when designing our AI-powered future.

Regulation as innovation

Health is necessarily a highly regulated industry. The impact of mistakes, malpractice, malicious intent or ethical breaches cannot be taken lightly. We must hold AI to this same standard when applying it to health.

The World Health Organization, the International Telecommunication Union and the World Intellectual Property Organization have taken a leading role in developing global standards that will guide the development, assessment and use of AI technologies for health – and these standards will be a keystone for responsible AI.

The responsibility of applying these and other global standards falls on all of us. Governments will manage the regulation of these technologies, but we all must have a voice in determining what is and what is not acceptable and safe. We will also require new types of regulations that are designed to be more flexible and responsive to accommodate the rapid pace of innovation. But if done right, new methods of regulation can support this technological innovation and bring us closer to health and well-being for all.

As with the elevator, there is no telling how AI technologies will reshape our world.

The views expressed in this #CriticalThinking article reflect those of the author(s) and not of Friends of Europe.