Organizations face a difficult challenge when it comes to ethically-informed data collection, sharing and use. On the one hand, there is increased sensitivity to ethical issues and desire for responsible stewardship of people’s information, both internal and external to organizations. On the other hand, outside of mere legal compliance, there is very little guidance for organizations about what being ethically responsible involves, let alone how to incorporate ethical consideration into product and service design, particularly at scale.
Even among organizations that have adopted institutional value statements for data/AI ethics platforms, there has been limited success translating them to organizational practices, decision-making and products. This situation is problematic for stakeholders from within and outside the organization. For individuals, privacy, data control, security and fairness are at stake. For organizations, there are new risks concerning data misuse and insecurities, with potential for loss of trust from users and consumers.
Some organizations are exploring the use of ethics committees to help build capacity for data and AI ethics. However, up to now there has been almost no resources from which to draw to guide development of effective and well-functioning data and AI ethics committees.
This report is intended to help fill that gap by:
Why use an ethics committee?
A committee-based model has been used effectively in several contexts, such as protecting human research subjects, supplying guidance in health care contexts, and providing oversight for embryonic stem cell research. It has several features that make it well-suited for building organizational data and AI ethics capacity.
A well-designed committee will:
Ethics committees can take a wide variety of forms and roles. Crucial to beginning the building process is putting together the right team of people and engaging with organizational stakeholders to begin to think through the key questions about function, values, principles, location, composition and process. Once the basic outlines of the committee are established and the initial committee is formed, quite a lot of the operational details will be developed in the context of the committee’s work.
There is no existing data and AI ethics committee template and the field of data and AI ethics is still maturing. Creating meaningful and effective ethics committee oversight models not only offers benefits and protection to the organization, it is critical to the broader data and AI ethics development process.
Northeastern researchers team up with Accenture to offer a road map for ethics oversight for artificial intelligence.