Who defines what is and what is not responsible? In this video, learn how to define the term in a strict sense rather than be philosophical and learn where the responsibility comes from and how to interpret data use.
- [Instructor] What is responsible? There is not one single definition. It's usually used as a blanket term next to good AI and comprises steps to understand applied AI. At its core, it means being able to justify automated decisions to customers, managements, and within an organization. And also, not to sound too grandiose, but to justify these decisions to society at large. It's a way to accept responsibility for the automated decision-making process. These automated decisions can be controlled by policy, process, and technical applications. So, policy means a formal policy that ideally is created, curated, and driven by a steering group within an organization. This allows for everyone to know how the policy is being applied and what the guidelines are. Also, this enables questions and alterations to be made when the environment changes. Process involves the actual steps needed to be responsible. These should be a subset of the policy, and also include technical guidelines, which focus on the implementation. These steps allow the outcome to be aligned with the initial goals of the project. Technical applications extend the process and policy and show the manner that the technical solution can be designed and the internal tools available to manage the requirements. So, although there is a definition, what you'll find is different companies use their own terminology.