Digital fitness
Explainable AI (XAI)
Also known as: XAI, Explainable and responsible AI
AI whose decisions can be explained to users, lawyers and the public, not just to the people who built it.
From Dancing with Change & Fluidity: discussing the ethical and legal readiness of new technology, the book describes explainable AI (XAI) as artificial intelligence whose workings and decisions can be explained, not only to the people who built it, but to users, lawyers, judges, the media and the public. Many machine learning models are opaque: they work well, but no one can precisely explain why a particular outcome came about.
For the author this is not a technical footnote but a condition for trust, especially in government, which must act reliably and in ways that can be checked. He illustrates the point with a real Dutch police investigation, where a purpose-built teaching program gave every party involved equal insight into how AI-assisted evidence had come about. Alongside explainability, the book also argues for an ethical review by a broadly composed ethics committee, and for weighing legal frameworks before an AI application is actually rolled out.
Alongside digital fitness and cobots, explainable AI shows that the book never treats digital fitness apart from responsibility: an organization that moves faster with AI must also be able to explain how that AI reaches its outcomes.
Discussions about this
Have a question about this topic, or an experience to share? Start a discussion; it stays linked to this page.
Comments on this concept
No comments yet
Be the first to comment.
Want to comment? Log in or join for free.