There are so many things that I learned about the MSc degree and program. One of which is that we have an assignment to design an “intervention.” There aren’t too many details about this yet, but the idea seems to be that you choose an area, topic, or domain struggling or dealing with an ethical AI issue and then design an intervention, be it legal, ethical, etc., to help mediate or solve it. The possibilities here are endless, but two thoughts immediately surface for me. First, choosing something where I can apply domain expertise makes sense. This could be sector-specific, like advertising or education, or topic-specific, such as generative models, enterprise-grade productivity tools, or nefarious data collection methods. Secondly, it probably also makes sense to survey the faculty, Shannon Vallor specifically, and ask which types of companies or industries contact EFI the most to understand where the biggest pain points lie. But there are some other things I learned worth noting:
- Ethics, by itself, may not be a robust solution to problems. Ethics must graft itself to policy, law, governance, intervention, and planning for it to be successful. In other words, ethics needs alliances to succeed. One could argue that to label something as “ethical” or “unethical” isn’t really that helpful because it is not an authentic action since it does not move towards a resolution. Perhaps that is why the creators of this program purposefully didn’t create a standalone ethics course. Moreover, ethics must be viewed through these multiple lenses and contexts for it to be robust, somewhat like blending different metals with the goal of ultimately creating a stronger alloy.
- Virtue ethics. A big takeaway was not to focus on individual actions but instead think of them as the derivatives of a larger ethical entity, e.g., person, cause, or organization. This helps to give the actions some cohesion and may help to mitigate the downside risk of creating a loose confederation of ad hoc decisions.
- There is global demand for AI Ethicists, or at least early-stage demand. Shannon Vallor stated that outside companies contact EFI weekly to ask for help with these issues. This cohort and this program are meant to serve as the “pipeline” to meet those needs.