Advanced Artificial Intelligence

Lord Freyberg Excerpts

Monday 24th July 2023 Lords Chamber

[Lord Freyberg (CB)]

My Lords, I too add my thanks to the noble Lord, Lord Ravensdale, for securing today’s timely debate. With rapid advancements in artificial intelligence, the possibilities seem boundless, but they also come with potentially significant risks. Like the noble Lord, Lord Kakkar, I will speak to how the opportunities and risks of the development of AI pertain to healthcare.

Machine learning, and more recently deep learning—commonly referred to as AI—have already shown remarkable potential in various fields, and both harbour opportunities to transform healthcare in ways that were previously unimaginable. AI can be used to process vast amounts of medical data, including patient records, genomic information and imaging scans, and to assist doctors in more accurate and timely diagnosis and prognosis. The early detection of diseases and personalised treatment plans can dramatically improve people’s quality of life and help save countless lives. AI can be used to analyse the genetic make-up of patients, and, in time, will better predict how individuals will respond to specific treatments, leading to more targeted and effective therapies, reducing adverse reactions and improving overall treatment success rates.

AI-assisted automation can streamline administrative tasks, freeing up healthcare professionals to focus on direct patient care. That has the potential to improve productivity dramatically in healthcare, as well as patient satisfaction, at a time when waiting lists and workforce shortages are, rightly, giving rise to concerns about their impact on our well-being and the UK economy. AI-powered algorithms can significantly accelerate, and thereby derisk, drug discovery and development, potentially leading to new breakthrough medications for diseases that have remained incurable.

While the promises of AI in healthcare are alluring, we must acknowledge its limitations and the potential risks associated with its development. The use of vast amounts of data to train AI models is bound to raise concerns about data privacy and security. Unauthorised access or data breaches could lead to severe consequences for public trust in new uses of this potentially game-changing technology. The models which underpin AI are only as good as the datasets they are trained on. Bias in the data underpinning AI in healthcare could lead to discriminatory decisions and exacerbate healthcare inequalities. Complex algorithms can be challenging to interpret, leading to a lack of transparency in decision-making processes. This opacity is liable to raise questions about accountability and give rise to new ethical considerations. We must ensure that we do not enter trading arrangements which might prevent our being able to assess the efficiency and risks associated with AI development elsewhere for its use in healthcare settings.

Crucially, where risks have the potential to be matters of life or death, we must resist the temptation to underresource pertinent regulators, and we should be mindful of hyperbole in our pursuit of innovation. To harness fully the potential of AI in healthcare while mitigating its risks, comprehensive and adaptive regulatory frameworks are imperative, both at national and international levels. The UK Government, in collaboration with international organisations, should commit to developing common standards and guardrails, by making the most of the global summit on AI safety that they will host in the autumn and contributing to the Hiroshima AI Process established by the G7. Any guardrails should be guided by the precautionary principle and prioritise patient safety, both now and in the future.

AI used in healthcare must undergo rigorous testing and validation to ensure its accuracy, safety and effectiveness. Independent bodies such as the MHRA can oversee this process if they are appropriately resourced, instilling confidence in both healthcare providers and patients. As the noble Lords, Lord Browne, Lord Bilimoria and Lord Holmes, and others said, the public should be involved in shaping AI regulations. While the Government’s AI task force is to be welcomed, it is imperative that civil society be engaged in the development of standards and guardrails applicable to AI in healthcare from the outset. The ongoing development of AI in healthcare harbours immense promise and potential. However, it is crucial that we approach this transformative technology with a careful understanding of its risks and a clear commitment to robust regulation and maintaining public trust. By fostering collaboration, we must usher in a new era of healthcare that is safer and more efficient and delivers improved patient outcomes for all.