Society will need complete trust that ethical principles underpin Artificial Intelligence (AI) before the industry can use it in everyday operations, according to European aviation safety regulators who have drafted an early roadmap.
The European Union Aviation Safety Agency (EASA) aims to use the document, which it published in February, to start consulting the industry on the implications of AI and to identify high-level objectives and actions to address the many issues surrounding the technology. EASA believes that AI’s potential to deliver autonomous flight, preventive maintenance, and ATM optimization raises significant and complex issues. “Among innovations, AI is probably the most disruptive,” said EASA executive director Patrick Ky. “It raises a number of questions that EASA, as a leading aviation safety agency, will have to answer.”
Those questions include how to establish public trust in AI-based systems; how to ensure transparency, non-discrimination, and fairness in safety certification processes; and how to certify AI systems and what standards, protocols, and methods will ensure that AI only improves on the current level of safety in air transport.
EASA insists that AI can only be considered trustworthy if developed and used in a way that respects widely shared ethical values and so future guidelines will need to build on the existing regulatory framework.
It bases its approach on the European Commission-led High-Level Expert Group on Artificial Intelligence (AI HLEG), which examined future policy development and ethical, legal, and societal issues related to AI, including socio-economic challenges.
In April 2019, the AI HLEG proposed the following seven key requirements for trustworthy AI: accountability; technical robustness and safety; oversight; privacy and data governance; non-discrimination and fairness; transparency; and societal and environmental wellbeing.
While the guidelines developed by AI HLEG are not binding, the Commission invited regulators such as EASA to assess their practical implementation.
EASA noted that it will face several challenges, including securing sufficient resources to guarantee staff competency to support the industry, research institutions, and the overall EU strategy and initiatives on AI. “Certainly, this work cannot be done by the agency in isolation,” said Ky. "EU member states and EU industry shall be involved in the further elaboration and implementation of the roadmap.”
Ky said the launch of AI Roadmap 1.0, therefore, marks a starting point, intended to serve as a basis for discussion with the agency’s stakeholders. “It is intended to be a living document, which will be amended once a year and augmented, deepened, improved through discussions and exchanges of views, but also, practical work on AI development in which the agency is already engaged,” he added.