🔬 Research summary by Angshuman Kaushik, Researcher in AI Policy, Governance and Ethics.
[Original paper by Inga Strumke, Marija Slavkovik and Vince Istvan Madai]
Overview: In spite of the fact that there is no dearth of Artificial Intelligence (AI) ethics guidelines, the number of unethical use cases of AI has proliferated over the years. This paper argues that a possible underlying cause for this is that AI developers face a social dilemma in incorporating ethics into the AI systems developed by them. It then goes on to define what social dilemma is, and describes why the current crisis in ethical development of AI cannot be solved without relieving AI developers of their social dilemma.
The paper begins by referring to the precarious position an AI developer finds herself in while discharging her duties. It takes the example of a company that develops an AI tool to be used to guide hiring decisions. If after deployment, the developer notices ethical challenges, and endeavors to mitigate it, she does so at the risk of losing her job. The paper calls such situations as falling within the category of a social dilemma and that the goal of the paper is to highlight the impediment to ethical AI development due to the social dilemma faced by AI developers. The researchers argue that the current approaches to ethical practices in AI development fail to account for the existence of the challenge for developers to choose between doing the right thing and keeping their jobs. According to the paper, a social dilemma exists when the best outcome for society would be achieved if everyone behaved in a certain way, but actually implementing this behavior would lead to such disadvantages for an individual that they abstain from it. Further, the problem that the researchers identify is that the current arrangements often put the burden to refuse unethical development on the shoulders of the developers who cannot possibly do it, due to their social dilemma. The paper also discusses how structures as contained in the domain of medicine can serve as a blueprint for ethical AI development.
AI’s Social Dilemma
A social dilemma, also referred to as a ‘collective action problem’, is a decision-making problem faced when the interests of the collective conflict with the interests of the individual making a decision. Instances of social dilemmas include the prisoner’s dilemma, the tragedy of the commons, the bystander-problem, fishing rights, etc. The researchers have put forward their own definition of social dilemma in the context of AI development. Further, the paper alludes to the fact that in the social dilemma in AI development, three agents are encountered, each with their possibly conflicting interests. They are the society, a business corporation, and an AI developer who is a member of the society and an employee of the business corporation. The interest of the society is ethical AI development, the interest of the business corporation is profit and surviving in the market and the interest of the developer is primarily, maintaining her employment, and secondarily, ethical AI development, because she is also a part of the society. The developer is thus put in a situation where she has to decide between her interest as a member of the society and her interest as an employee of the corporation. This is the social dilemma which the researchers do not want the AI developer to face.
The case for professional codes
In the opinion of the researchers, legislation cannot solely solve the challenge of ethical AI development for the following reasons namely, legislation develops at a much slower pace than current technology and legislating against anything that could potentially be unethical would disproportionately impede progress. In fact, the paper stresses on the point that if the corporations do not shoulder the ethical responsibility, individual developers will be hindered in pursuing ethical development due to their social dilemma. Therefore, they provide a possible solution to this problem in the form of professional codes as they have been used to promote the agreed upon professional values in areas where legislative solutions are inadequate. Further, according to the researchers, in the context of AI, professional codes are a natural remedy against social dilemmas encountered by AI developers in professional settings. The reason being, the AI developer or for that matter any other employee would be relieved from the potential consequences of refusing to work against the diktat of the management, as long as they perform their duties in conformity with the professional code of conduct. Besides, the public would have insight into what is the standard ethical conduct for the entire profession.
Drawing inspiration from the field of medicine
The paper suggests using the professional code of conduct contained in the domain of medicine as a framework for developing the AI ethics code. Although other fields have also developed professional codes, the researchers chose medicine because of its societal impact. In fact, they consider AI having the potential of a similar or even more substantial impact, depending on its future development. Further, the paper elaborates on the development of the professional code of conduct in medicine spanning from the first written records dating back to Sumeria in 2000 BC to the recent ones. According to the researchers, medical professionals are guided in their work by these ethical codes, and are protected from the social dilemma as the publicly known ethos enables them to refuse unethical behavior without any fear of ramifications.
Professional Codes in AI
The researchers suggest that the topic of professional codes for AI has raised considerable interest during recent years, and several works have pointed out the fluid nature of this field and its complexities. Their primary proposition is that AI development must become a unified profession, taking medicine as an example and, as in medicine, it must become licensed. Further, the license must be mandatory for all developers of medium to high risk AI systems, in line with the EU Proposal for a Regulation laying down harmonised rules on artificial intelligence. This would protect the individual developer in an exceptional manner. The chances of being replaced by another professional would be almost negligible, since employers would know that all AI developers abide by the same code. Thus, an AI developer could refuse to act in an unethical way without the fear of the social dilemma consequences. Secondly, national AI developer organizations shall maintain registers of employed AI developers, in the same manner as maintained by the national medical societies. The paper further suggests that lastly, unethical behavior may lead to the loss of one’s license, which is a strong incentive not to take part in unethical development, even if required by an employer. The researchers further argue that establishing ethical AI development will otherwise fail, for instance, as long as professionals are uncertain as to whether they are AI developers or not, when AI developers are left alone with their social dilemmas etc. According to the researchers, although overcoming all the obstacles to establish a unified AI developer profession will be a tiresome endeavor, it will at least remove the social dilemma faced by the developers. They further argue that this is the only realistic way to ensure that ethical AI development follows goals in alignment with societal benefit. Once a professional code of conduct is brought into existence, it will serve as a safeguard against unethical corporate and government practices. This is vital as the role of corporations can be manifold, and a unified profession will help to steer their decisions in a direction aligned with society’s ethical expectations. The paper sums up by mentioning that the idea of a professional ethos for AI professionals and its role in achieving AI ethics has been argued for in other related literature, and it gives a summary of some of these arguments. The researchers also acknowledge the fact that the impact of professional codes of conduct might be less effective than they believe. However, they are convinced that the current debate will profit from the inclusion of the social dilemma aspect and a discussion of the potential solution that they suggest.
Between the lines
Raising dissent – primarily, on ethical issues – against the management or the board for that matter, when it comes to development/deployment of an AI product that could garner a healthy profit, can prove disastrous for the professional career of an AI developer. Being an individual with duties towards the family and also being a member of the society, the position of an AI developer becomes very precarious in such situations. In fact, there are real-life examples of AI developers being shown the door for flagging issues that have deleterious effects upon society. Although, the companies have started appointing AI ethics officers – who are trained in the relevant field – this paper suggests plausible propositions for relieving the burden of taking critical decisions from the shoulders of AI developers.