TEAMING.AI and auditable ethics

Within TEAMING.AI, a central concern is that all project results are designed, created, and tested with European values in mind. Therefore, we aim to create a human AI teaming framework that is aligned to the EU’s framework for Responsible Research and Innovation (RRI) and provides a methodology for mapping the identified requirements into verifiable policies. By doing that, TEAMING.AI addresses the core problem of auditable ethics conformity in the context of human-AI system interactions.

“Our ambition is to provide methods to monitor at runtime whether high-level ethical guidelines are being violated or not.”

Our approach relies on two key advances in hybrid AI, which combines data-driven machine learning with knowledge-based methods such as knowledge graphs. The main idea is to leverage complex event processing in combination with fast reasoning techniques by means of lightweight knowledge graph embeddings. This way, TEAMING.AI is contributing to the necessary regulatory insight and oversight for AI-based technologies to enable sustainable development in the frame of the upcoming AI regulations.

Our first results towards auditable ethics will be presented by Hans Graux in the next AI4Manufacturing workshop of the ICT-38 Cluster on “Ethical and Legal Issues of AI in Manufacturing” (see https://ai4manufacturing.com/category/news/events).

Hope to see you there.

Source: https://grc2020.com/2018/02/22/technology-priorities-for-compliance-ethics/