Introduction
The pursuit of healthy longevity—defined as extending the period of life spent in good health—is rapidly evolving from an aspirational concept to a booming biomedical industry. Central to this pursuit is artificial intelligence, which is being used by many longevity start-ups, research labs and medical clinics. AI applications include analysis of biomarkers, building predictive models, and personalisation of treatments and interventions. Indeed, given the quantity of data now produced by wearable devices, smartphones and healthcare systems more broadly, achieving healthy longevity goals and objectives would be practically impossible without artificial intelligence (AI). At the same time, however, using AI for longevity purposes carries distinct ethical risks and concerns for individuals and society at large. One way to examine these risks and concerns is through the lens of the four principles of biomedical ethics—beneficence, non-maleficence, justice and respect for autonomy. These four principles underpin the ethics of modern medicine; they are also found within many AI ethics frameworks. This article discusses some of the ethical issues with the use of AI in the field of healthy longevity.
Beneficence: achieving better health for longer
The principle of beneficence requires stakeholders (e.g. clinicians, biotech companies, regulators and individuals) to promote health and well-being in a trustworthy and responsible manner. Recent advancements in longevity treatments and interventions like molecular biomarkers, epigenetic clocks and multi-omic analytics allow for more precise risk stratification and personalised interventions. AI algorithms are required for interpreting the complex datasets required for these interventions, and can also be used for designing protocols tailored to an individual’s health profile such as their biological age. Risks abound with using AI for such purposes, however.
For starters, AI-powered algorithms that perform health analysis are often trained on datasets that have been obtained from patients in particular regions or countries and therefore possess a limited set of demographic information. If this data is shared via public-private partnerships and merged with additional data, or if the algorithms are used outside of their contextual setting, future applications of the algorithms can lead to biased outputs which can lead to social discrimination, overgeneralisations or other inaccuracies. The sharing of data and inappropriate use of algorithmic systems must be carefully scrutinised and managed to ensure their application minimizes the inherent risks and dangers with AI systems.
Large language models (LLMs) further complicate matters due to the way in which they can produce seemingly authoritative health advice that might be inaccurate. This is especially concerning for individuals who use platforms like ChatGPT instead of doctors when seeking health advice, diagnostic answers to their health worries, or for designing longevity protocols. One major problem with this is that, because each person’s biology is unique, the LLMs, which are trained on massive quantities of generalized data, have little or no understanding of the person’s biological and environmental conditions. LLM outputs are therefore unlikely to capture and reflect the person’s lifestyle and unique circumstances that are crucial for determining the appropriate medical advice and correct courses of action.
Other concerns exist in relation to the healthy longevity industry. One major concern is how advertising and marketing claims are prone to hyping or exaggerating the benefits of certain products and services without also noting the associated risks or uncertainties. Such claims can lead people to buy and use unproven treatments and interventions that fail to produce the expected outcomes. This carries a financial risk for consumers: many longevity products and services are expensive and it’s often not clear whether the desired results are in fact being realized. This can, in turn, inflict negative psychological impacts on people’s health and well-being; the constant monitoring of devices and tracking of data can produce “longevity anxiety” and general discontent if results are unsatisfactory.
For longevity products and services to be beneficial, they should inform consumers of the potential risks and harms associated with their use, similar to how most food packaging presents the ‘nutritional information’ thereby empowering the consumer to make more informed decisions about their health choices.
Non-maleficence: mitigating risks and minimizing harms
As with all AI algorithms, the primary ethical concern centers on the need to minimise bias within datasets and prevent discrimination from occurring in algorithmic outputs. In the context of healthcare, one of the most pressing practical challenges for medical professionals lies in understanding how opaque AI systems like LLMs—often referred to as “black box” AI—produce their outputs. Since many medical professionals lack technical training on the AI systems they are using, it is practically impossible for them to explain to their patients how an AI system produces a given output.
Further still, medical professionals can find themselves disagreeing with AI generated medical advice. This is an especially difficult problem for doctors because such disagreements can lead to greater uncertainty and self-doubt about what the correct course of action ought to be for their patients. If doctors do not follow an AI recommendation that turns out to be correct then questions arise as to why they didn’t follow that recommendation, and whether they should have? As AI systems become more sophisticated and powerful in their analytic capacity, serious ethical questions arise over how doctors ought to use AI systems when treating their patients, the level of trust they should have in AI systems, and who is liable when AI systems make incorrect decisions that inflict harm.
Moreover, given the quantity of health data used in healthcare and healthy longevity pursuits, there is still significant risk surrounding the capacity for AI to exacerbate pre-existing social inequalities as embedded within large datasets. Eliminating such bias and ensuring that AI systems do not perpetuate social inequalities is crucial for achieving moral progress at both local and global scales. Since the principle of non-maleficence—otherwise stated as “do no harm”—is one of the oldest ethical principles in medicine, mitigating risks and minimizing harm must stand at the forefront of AI development going forwards into the future.
Justice: ensuring fair access and equitable opportunities
The principle of justice emphasises fair distribution of benefits and burdens across society. Concerns with justice are beginning to arise in relation to the fact that many longevity treatments and interventions are expensive, e.g. advanced therapies like CAR-T and CRISPR can cost hundreds of thousands of dollars per treatment, and are therefore available only to the wealthy. This risks creating a new form of “longevity privilege” where only the wealthy can benefit from new medical advancements. This can have a compounding effect whereby the wealthy become healthier and thus more capable and productive for more years than they otherwise would, thus providing them with additional opportunities throughout life. This could also result in a reduction or loss of certain opportunities for those unable to afford health longevity treatments and therapies.
At the societal level, the scientific and technological advancements driving the healthy longevity movement may disadvantage the younger generations and create intergenerational divides. One way this might occur is if older populations (e.g. 40 years or older) were to start living healthier for longer, then this could have a limiting effect on the economic opportunities and political progress for the younger generations. For instance, if the older generations remain in the workforce for longer then the number of job promotions and career opportunities might decrease, particularly given the risk that AI poses on future employment opportunities already. One political implication of this is that older “worldviews”, values and perspectives are likely to remain active within society for longer, potentially hindering social justice causes. While these implications and consequences may seem speculative, the mere possibility of them occurring forces us to consider their future impact on society and mitigate the social risks and potential harms sooner rather than later.
Respect for personal autonomy
Respect for autonomy involves respecting individuals’ rights to make informed decisions that align with their personal values and life preferences. With AI-powered algorithms gaining more agency and interactive capability, the ability to respect personal autonomy becomes complicated. One area of concern relates to the doctor-patient relationship. Traditionally, doctors have relied on their own expertise combined with that of their colleagues for analysing and diagnosing medical conditions and illness. However, the adoption of AI into medicine has complicated the doctor-patient relationship. In some cases, AI-powered algorithms perform better than doctors do with diagnosing illness. Furthermore, some AI platforms are proving to be more empathetic and relatable than some doctors when conveying medical information to patients. This presents a major concern for medical professionals, for not only must they manage the risks inherent to AI-powered algorithms but their bedside manners are now being challenged by AI agents that are becoming more sensitive to the emotional well-being of patients and in some cases, are doing a better job at conveying difficult and sometime traumatic information. Navigating the use of AI agents in healthcare is one of the most pressing topics in medical ethics today, for people’s lives are literally at stake, as are the careers of medical professionals.
Conclusion
Healthy longevity and modern medicine more broadly promises unprecedented gains in both personal and public health, but the pursuit of a longer, healthier life also raises important ethical questions, especially in an era where AI is deeply embedded in biomedical research and clinical practice. It’s worth noting that the four bioethical principles covered in this article are reflected in the World Health Organization’s six core principles for AI in health: protect autonomy, promote human well-being, ensure transparency and explainability, foster responsibility and accountability, ensure inclusiveness and equity and promote sustainability. To ensure that longevity research, science and medicine achieves the objective of extending human healthspan, all stakeholders should commit to these basic principles, for doing so will not only help with avoiding certain harms but also bring the best out of AI systems. This can be achieved by using evidence-based treatments and interventions, with a focus on providing equitable access, transparency of information, robust data governance and inclusive engagement with diverse communities. AI will be essential for achieving these objectives, but only if the technology is managed in an ethically responsible way.
