Professional codes of ethics tend to have certain common features.
These features grow out of the general rationales for having such codes of ethics or codes of practice in the first place. Professionals have expertise and resources which the ordinary public, and their clients, generally lack. There is thus a power gradient between professionals and clients, and the larger public. This then gives general responsibilities of professionals towards others. The professions are also generally held in good regard; there is a general assumption that the products or services they offer are of broad benefit to society or at least to certain individuals within society. But there is also a recognition that, conducted badly, these same services and products could lead to harm, even disaster, for individuals and for society in general. Drugs can cure, and they can kill. Bridges can link communities, and they can collapse.
Hence, codes of professional ethics tend to have elements which help to mitigate the power gradient between professionals and others. These include:
Undertakings to assure levels of professional competence, and to update skills and knowledge appropriately.
Undertakings of honesty and transparency in dealings with the public and with clients.
Undertakings to operate within the law of the appropriate jurisdiction and any relevant local or regional government regulations (often simply implied).
Undertakings not to harm, and to provide benefits to clients in the operation of their products or services, and not to harm the public (for example, through consideration of the environmental effects of a profession’s activities). The benefits involved are generally understood in terms of the particular area of expertise of the profession; avoidance of harms may, of practical and legal necessity, be understood more broadly.
In some cases, undertakings not to bring the profession into disrepute.
Procedures for reporting problems and violations of codes, which may include protection for whistleblowers and accounts of penalties for proven misconduct.
The broad implication of all of this is that the professional has control over the products or services and can hence ensure that they are used appropriately and for a good outcome.
The codes of ethics concern two main elements: the behaviour of the professionals, and the impact of their products or services.
Enforcement of codes of ethics may involve professional sanctions, restrictions on membership of professional bodies, which for some professions may make it impossible to practice, and in worse cases, legal ramifications. Enforcement also occurs through the soft power of the authoritative weight and respect with which the relevant professional body and its codes of ethics are held.
In the case of research funding councils, for example, adherence to ethics protocols in the stages of developing projects, and throughout the life of a project, including dissemination of results, is generally a condition of receipt of funding. (See our Resources page for some examples of this.)
There can be cooperation between different bodies for the enforcement of codes of conduct. For example, concern over the bias in findings of research by pharmaceutical companies by the suppression of negative results has led to moves whereby clinical trials must be openly registered before their start, and academic journals will not publish any trials which are not compliant with this.
The very existence of a professional body which produces codes of ethics or conduct also itself helps to contribute to the relatively high status of the professions. This alerts us to one danger of publicly visible codes of ethics – that they may lapse into mere window dressing.
Context of codes of ethics: To understand many codes of ethics fully, it’s helpful to look at the wider institutional background and rationales for their production.
For example, codes of medical ethics cannot be fully understood without at least some awareness of the history behind such codes, for example, the history of the development of medical ethics in the 20th century since the Nuremberg trials, the development of the Nuremberg code, and including the Helsinki declaration and its many revisions.
In medicine, and elsewhere, codes of ethics have developed in response to catastrophes or scandals, and understanding this can help to understand how codes have grown up as they have. For example, response to the Tuskegee Syphilis trial have had a big impact upon medical ethics, to name just one of many such instances.
Codes and regulations may be developed in anticipation of possible problems, with attempts to avert these. For example, the EPSRC Principles of Robotics may be seen as an attempt to avert the kind of public backlash that was seen in the UK over GM crops. Hence, such background issues are again important to understand in considering the purpose and final shape of any codes or regulations.
Codes of ethics of professional bodies also often have a wider national or international context. For example, codes of medical ethics for different countries have their own integrity, yet also exist in the wider context of the policies of the World Health Organisation. Codes of medical ethics are closely linked to the development of medical law in the relevant jurisdiction; and the development of medical law in separate jurisdictions is itself often influenced by developments and cases in other jurisdictions. For instance, the case of Tarasoff v Regents of the University of California (1974) has influenced the development of law regarding medical confidentiality and the duty to warn in jurisdictions outside of the USA, including the UK, and there are many other such examples.
This wider international context also has great impetus from the way in which many products or services, and much research, also takes place in a global context. For example, much research in genomics of necessity needs to study different population groups of humans in order to conduct scientifically robust research which can be used in application to those different groups. Complex ethical considerations of how to marry global standards with local sensitivities may be needed.
Underlying values behind professional codes of ethics: Looking closely at different professional codes of ethics, it’s possible to discern broad values that lie behind these. As debate and thinking about ethics continues, and as society changes, there may be changes in how these underlying values are articulated and promoted. There is also no particular reason to assume that all these underlying values are necessarily going to be entirely consistent. Differences of interpretation and emphasis may mask or reveal deeper differences of opinion, or commonalities, between individuals, groups and communities, and geographical regions, towards these broad underlying values.
Take the value of autonomy. It’s useful to consider this, not just because it’s such a core value in contemporary society, it’s especially key of course to AI which is developing autonomous systems and machines. Respect for the autonomy of the individual is a key value in codes of medical ethics, expressed in various ways and articulated via concern for issues such as confidentiality and free and informed patient consent. The history of medical ethics over the last century or so can be read in no small way as the history of how patient autonomy has been granted greater and greater emphasis, as opposed to the ‘doctor knows best’ model. This demonstrates how our values are changing.
But there are philosophical and practical questions and differences in how exactly this value should be understood. Here’s one challenge: respect for individual autonomy in clinical medicine may sit in some tension with principles of public health. There are also large cultural differences in how, and to what extent, individual autonomy in medicine is to be valued. Reading journal articles on medical ethics, it’s often fairly easy to guess if the authors originated in the USA, or in Northern Europe, by the ways in which autonomy is discussed and ranked alongside other more communitarian or social-oriented values. The facts of the world can also present us with challenges. How do we carry on valuing autonomy, for example, in patients with advanced dementia? Often, it’s science in combination with advances in technology which are presenting us with new, or newly acute challenges. How do we value individual privacy of genetic data, when it’s now constantly apparent that by revealing genetic information about one person, we may be ipso facto revealing information about others at the same time?
Some of the underlying values may easily be assumed, but often it is during difficulties with applying codes that complexity emerges. For example: the practice of medicine seems naturally to be one that benefits people, because it aims to improve health. But what is health? How do we value quality of life versus length of life, when there is a clash? Such a clash can occur visibly in individual health care decisions, or less visibly at policy and research priority levels. And providing a definition of health is surprisingly hard. The World Health Organisation gives one definition here; but some argue that this is too broad, including as it does social wellbeing and rejecting the idea that health is simply absence of disease or infirmity; others argue that it’s too narrow, for example, a well-used definition of Aboriginal health from Australia includes references to the cultural wellbeing of the community and includes reference to a cyclical life cycle.
As codes of ethics are formulated and revised, it’s vital to be aware of the underlying values which are motivating and shaping these codes. It’s noteworthy that it is often through the development of new practices and new technologies that important questions about the underlying values become apparent. For example, it was only with the advent of reproductive technologies, and coupled with changing societal attitudes to same-sex couples, that questions about the precise definition of what it is to be a parent arose in their present form.
How does AI fit into all this? (See also the parent page, Is AI Ethics Special? for further discussions.)
AI and professional bodies: AI covers such a wide remit that developments in AI research may not necessarily be carried out by people operating with links to accredited professional bodies.
However, it’s noteworthy that a major initiative by the Standards Association of the IEEE (the Institute of Electronic and Electrical Engineers), the Global Initiative for the Ethical Design of Autonomous Systems, is underway currently. The IEEE has a worldwide membership of over 400 000 so this is of large significance.
Companies whose work encompasses AI are also involved in considerations of ethics. Google’s DeepMind has undertaken to do work on ethical issues, although to date, it’s not possible to comment on this initiative, as there isn’t any information on their website at the time of writing this page (08/11/2016). Lucid Holdings plc does have an Ethics Advisory Committee, with some details given here.
For research with public funding, e.g. from funding councils, these may have or develop specific codes of ethics for research involving AI. Note the EPSRC’s Principles of Robotics; this however is at present in the form of general principles rather than something more specific.
AI and the gradient of power between professional and others:
A major issue in AI is that even the professionals may not understand fully how some aspects of AI works. For example, machine learning may produce outcomes which are not fully accessible. Autonomous machines and systems may lead to decisions and actions which their makers did not, and perhaps could not, predict.
This is of great importance, for it means that the presumed power gradient between professionals and others is eroded in certain respects.
AI and machine behaviour – added layers of complexity:
As we saw above, codes of professional ethics typically deal with two elements – the behaviour of the professionals themselves, and the impact of their products or services. It’s also assumed that the professionals have (adequate) control over the latter.
However, in dealing with autonomous machines and autonomous systems, a third layer of complexity is introduced – the behaviour of those machines or systems themselves.
There are various approaches towards trying to deal with this added complexity.
For further discussions, see this post, The Distinctiveness of AI Ethics, and Implications for Ethical Codes.
The values lying behind codes of ethics and AI:
This raises fascinating and complex questions. AI, considered generally, has such a broad remit that it potentially affects a very wide area of human life and endeavour. This is a significant issue, and point of divergence from most codes of professional ethics which deal only with specific areas of life.
Moreover, AI raises questions that reach to the heart of what it means to be human, what it means to have agency in the world, and what activities individual humans, and society in general, wish to value. For some further discussion of this complex issue, see here and here.
We would like to thank the Future of Life Institute for generously supporting our research project.