Toolkit produced by Open Roboethics Institute.
Unlocking the power of data using algorithms and intelligent systems has the potential to help tackle some of the world’s biggest challenges. Most of us set out to launch AI projects because we want to make a positive impact in the world. However, regardless of our intent, if we are not careful in making the incremental design and deployment decisions, our well-intentioned technology can fail or have serious negative societal and ethical consequences. Thinking through what those consequences could be can give you the foresight you need to avoid such failures and maximize the benefits of the technology for everyone. We designed Foresight in AI Ethics (FAIE) as a toolkit to help you build an ethics roadmap that is tailored to your particular project. The process highlighted in toolkit is born out of our work in providing AI ethics consulting in 2017. It is a systematic way you can follow to uncover what key ethics issues are relevant to your project and strategize how to better anticipate, manage, and act to mitigate the ethics issues. This toolkit can introduce AI ethics foresight early in the design and deployment process, rather than serve as an auditing or evaluation tool.
This toolkit for anyone who is actively involved in the development and deployment of data-driven technologies for healthcare. This includes data scientists, engineers, product managers, business leaders, or entrepreneurs looking to incorporate ethics into their AI project. It also includes people who have recently been assigned to assess the ethical integrity of a new AI system being developed.
There are three main phases to FAIE, and a total of eleven steps to follow. Throughout the process, it forces us to think about the following three things in a systematic manner:
- People: How the stakeholders are related to each other and the technology,
- Values: What values are important for each stakeholder groups and the society, and
- Technology: How the new technology is related to the people and their values, and what impact it will have on them.
At the end of this process, you will have a strategy that enables key stakeholders of the technology you are designing to become knowledgeable stewards of the technology. You will have a mapping of foreseeable ethics issues specific to your project in one of three categories of things (make design, business, or communication decisions) you can do to manage or address the identified set of issues. Due to the many different techniques and perspectives that are used to enrich the final outcome, we recommend an interdisciplinary team (e.g., data scientist, business leader, product manager, communications manager) to take on this project together. Unlike checklists or technical evaluation tools that can be used in a few minutes, this toolkit requires you and your team to take the time to explore, investigate, and think. In order to better demonstrate the process, we walk you through a fictional example of a hospital that is creating a new AI division to innovate their operations. [ . . . ]
Related information: https://openroboethics.org/ai-in-healthcare-is-your-solution-ethics-proof/