In an analysis published in BMJ Global Health, a coalition of medical experts raises pressing concerns about AI development, urging an immediate halt and stringent regulations. Here we explore the risks of patient harm, compromised health data security, and amplified social inequalities. We’ll look at broader societal threats, including misinformation dissemination, trust erosion, potential conflicts, and the impact on employment. Stay informed as public health experts and doctors navigate the profound implications of AI in healthcare.
As artificial intelligence (AI) continues to advance and integrate into various aspects of our lives, concerns about its potential dangers to public health and human existence have been raised. Recently, a coalition of UK doctors and public health experts called for a halt in AI development and robust regulation and policy-making—before a “window of opportunity to avoid serious and potentially existential harms” closes. This article examines the risks associated with AI and discusses potential solutions.
Beyond the immediate risks to healthcare, AI poses more extensive threats that could destabilise society and undermine democracy. These include misinformation and manipulation, the use of lethal autonomous weapons, and mass unemployment with subsequent psychological fallout.
As AI becomes more integrated into healthcare, it is essential to ensure that AI technologies are used to complement human decision-making. Just as X-rays and MRIs are used to inform diagnoses alongside other health information, AI should support, rather than replace, patient and doctor collaboration in treatment decisions. The importance of patient consent and input in embarking on treatments must be built into the systems.
AI can revolutionise healthcare by improving diagnoses, finding better treatments, and extending care to more people by automating routine tasks, making remote monitoring more effective, and providing health insights from vast data. However, the development of AI also has the potential to produce negative health impacts, as outlined in the analysis published in BMJ Global Health. These include:
The reliability of medical advice generated by artificial intelligence (AI) systems has become the subject of intense debate. As headlines boast of the diagnostic capabilities of AI, questions arise regarding their superiority compared to human medical professionals. These technologies face significant challenges. Limited or biassed datasets can lead to inaccuracies and disparities, particularly affecting women and individuals from minority communities. For instance, according to a Guardian report, an AI-driven pulse oximeter that overestimated blood oxygen levels in patients with darker skin resulted in the undertreatment of their hypoxia, causing harm to these patients.
The integrity of the training data itself is a concern, as it may contain flawed or incorrect information. An insightful analysis from MIT Review delves into these complexities, highlighting the need for a critical assessment of the limitations and risks associated with relying solely on AI for medical advice in the healthcare field.
AI systems may displace many knowledge workers. While there are benefits to eliminating repetitive, dangerous, and unpleasant work, unemployment is strongly associated with adverse health outcomes and behaviour. The psychological and emotional effects of a world where more and more jobs are unavailable or unnecessary are unknown, and the policies and strategies needed to address these issues remain largely unexplored.
The most significant concern the coalition of doctors and public health experts raises is the potential existential threat posed by self-improving artificial general intelligence (AGI). AGI, theoretically capable of learning and performing the full range of human tasks, could apply its intelligence and power in ways that could harm or subjugate humans, whether deliberately or not. As AI research and development grows exponentially, the window of opportunity to avoid these serious and potentially existential harms is closing.
Amid these developments, another important concern has caught the attention of health experts, independent fact-checkers, and medical charities in the UK. They have called for the government’s forthcoming online safety bill to be amended to address health misinformation. They advocate internet companies must have clear policies for identifying and addressing harmful health misinformation on their platforms, which would protect users and improve trust in public institutions.
The coalition of experts calls for effective regulation of AI development and use to avoid harm. They argue that a moratorium on AGI development should be instituted until robust regulations are implemented. Additionally, the group emphasises the need for international cooperation and collaboration in establishing comprehensive AI policies and guidelines.
As algorithms analyze mammograms and smartphones capture lived experiences, researchers are debating the use of AI in public health. https://t.co/IWge9gwooU
— Harvard Public Health magazine (@PublicHealthMag) October 17, 2022
To ensure that AI is developed and used responsibly, designers and developers must prioritise the needs of the people affected by AI technologies. This includes collecting diverse biological data and information about the beliefs and wishes of various communities, even though this may be an expensive endeavour. It is vital to build systems that are designed to eliminate the threats above, by access to carefully vetted data and the use of robust protocols created with harm elimination as a priority.
AI poses "an existential threat to humanity" akin to nuclear weapons in the 1980s and should be reined in until it can be properly regulated, doctors and public health experts warned.https://t.co/eXYftAxgH0
— Axios (@axios) May 10, 2023
While AI has the potential to revolutionise healthcare and improve the lives of millions, the risks it poses to public health, social stability, and human existence cannot be ignored. AI has the potential to exacerbate existing social and health inequalities with inherent biases towards certain demographic groups, leading to inaccurate diagnoses or treatments for minority populations. Additionally, access to AI-powered healthcare services may be limited for those in low-income or rural areas, further widening the healthcare quality and availability gap. By addressing these concerns through robust regulation, policy-making, and responsible AI design, we can harness the benefits of AI without reaping negative impacts.