Activity: Community Talks, Presentations, Exhibitions and Events › Public lecture/debate/seminar
When it comes to machine learning results, we need to enable stakeholders to critically examine the evidence for accepting the risk. This will enable them to assess it's trustworthiness.
The deployment of an algorithm in production should be limited by its context of use. A condition of deployment should be to mandate that the chosen method, data collection, biases, model limitations and any other discriminating information is made available for scrutiny and is contestable.