Frequently Asked Questions
The process of understanding challenges created by the rapidly increasing applicability of AI, and then establishing the principles to follow has involved people and organizations around the world. At this point, hundreds of events have taken place and dozens of documents on AI principles are already published. Based on discussions with policymakers, now is the time to take the next step by building on those broader principles to help them in making practical decisions about AI.
The AI Policy Forum (AIPF) begins with a number of preparatory task forces that are exploring some of the most pressing issues of AI policy. Each task force, chaired by an MIT researcher and including a diverse range of technical and policy experts from around the world, will be active for several months, convening one or more workshops for the participants during that time. The yearlong process culminates in a series of events to gather high-level decision makers and to provide a focal point for the work of the AIPF, including an online symposium in May 2021 and followed by a summit later this year.
We have identified three broad sectors that are relevant across countries, regions, and cities to focus on — mobility, finance, and health care. Specific issues within each of the topic areas will be determined by the task forces.
Our goal is to help policymakers in making practical decisions about AI policy by providing context and guidelines specific to a field of use of AI that can aid with implementation. The research from our task forces will also feed into the development of the AI Policy Framework that comprises a broad analytical toolkit and sector-specific information on emerging issues, as gleaned from expert task forces, as well as identifies possible policy levers and fundamental trade-offs that the decision makers are likely to face.
AI and the computing systems that underlie it are more than just matters of technology. The applicability of AI is increasing at such a rapid pace today that there are fundamental economic, political, societal, and ethical issues we need to consider as a result, such as work of the future, privacy, fairness, accountability, and even our sense of reality. In order to address such issues, governments and companies need frameworks tools to consider trade-offs in developing and adopting concrete policies.
Some of the most pressing issues facing the world are of direct relevance to AI policy, including mobility, finance, and health care. Implementing policies requires not only principles but also trade-offs, which is what makes concrete policies hard to develop. For instance, privacy is an important AI principle, but protecting privacy may affect accuracy. If this is in health care and the cost is lost lives, how does one make these trade-offs? Other trade-offs may go against other important collective concerns like climate change, for example when new computing techniques imply a growing carbon footprint. Also, two AI principles may conflict against each other, for example when in some specific contexts more explainable models may result in more unfair algorithmic classification. Who decides what is the right choice here, and how? Our task forces will zero in on exactly these kinds of questions.
Moving beyond principles means understanding the trade-offs, and identifying the technical tools and the policy levers to address them. Therefore, we aim to create policy frameworks that provide methods of analysis for policymakers. The focus should be on explaining the trade-offs, specifying options, and proposing criteria for making practical decisions about AI policy. This process needs to be highly specific, including specific to a field of use of AI. This is why we need a year of work. A rigorous process is required for investigating and specifying mechanisms that allow us to get to the requisite level of specificity for concrete policy action, driven not by policy analysis and the understanding of technology and its complexities.
The AI Policy Forum is convened by the MIT Schwarzman College of Computing. However, there are many individuals from across MIT and beyond who are dedicating their time to this effort. See our people page to learn more about who is on our steering committee, leadership group, planning group, and task forces.