After much deliberation and some guidance from my professors, I’ve decided to change my final project/thesis. The motivation for the change is largely driven by three factors:
- The programme’s readings to date led me to the realization that for many companies, the answer to creating more ethical technology is based on a combination of new statistical tools and a deontological set of values set by the company.
- The exposure to Professor Shannon Vallor’s work on the topic of Virtue Ethics taught me that moral capacity is developed over time, not an untapped resource we’re born with waiting to be mined.
- I have a natural disposition towards creating practical tools and solutions to problems rather than relying solely on abstract theoretical ideas and frameworks that are difficult to operationalize.
Overview of the new topic idea
We’ve all seen companies espouse their enduring ethical principles of responsible AI. These principles often include themes such as safety, security, transparency, and fairness. On the surface, these principles seem reasonable, albeit somewhat anodyne. Who, after all, wouldn’t want safety? Many responsible AI frameworks today are based on a ‘rules and tools’ approach to ethics, consisting of an idealized set of principles coupled with statistical tools that attempt to quantify these principles and minimize unethical outcomes. This ‘rules and tools’ approach is based on a deontological assumption that ethical decisions will somehow conform to the predefined ethical norms set by the company. What is often missing is an articulation of exactly how the employees will learn to make ethical decisions or improve their ethical decision-making abilities over time. This crucial variable highlights the need for ‘ethics as a skill set,’ a central tenet of virtue ethics and the missing component of responsible AI practices.
Ethics as a skill set fills the gap left by the rules and tools approach to responsible AI by training employees throughout all levels of the organization how to actively think about responsibly developing, managing, auditing, and measuring the technology they build rather than deferring to some moral authority. It is created through a constantly evolving body of practical know-how requiring regular training for increasingly complex situations. The training can take many forms, including complex case study analysis, explanatory videos on specific topics, and analytical scrutiny of training and test datasets. Perhaps most importantly, ethics as a skillset isn’t immutable, whereby the concepts and skills constantly evolve as the ethical, legal, and regulatory landscape changes over time. The result is an organization that can adapt to new and different challenges by expanding its collective ethical capacity.