global civil aviation authorities wander through the landscape of AI

EASA released its Artificial Intelligence Roadmap 1.0 this past week on 7 Feb 2020. Aside from the observation that “roadmap” seems like an ill-fitting analogy for an industry based in the cloud, I wonder what took them so long. If you decide to read it, I recommend that you skip the opening pages, one of which declares that the last great revolution in aerospace was the jet engine, overlooking the many changes since then, including avionics and ground based computer systems integrated into the aerospace ecosystem. I hope to share more thoughts on the roadmap in a later column.

While perusing this report, I found some other interesting reports:
Application of AI in the NAS – the Rationale for AI-Enhanced Airspace Management. There are two particular things about this paper that I found unhelpful. First, one might observe that a common refrain in discussions about aerospace and technology (as it is in this paper) is “we need more bandwidth.” I think that’s a foolish claim. Personally, I think we don’t need to send so much useless and unused data across the data links we have. That is a waste of the available bandwidth. Second, many section headings are completely irrelevant to the material both before and after the heading, but in those cases where there is a semblance of relevance, the authors descend in abstract discussion or trivial definition making, instead of applying the ideas to the problem presented in the paper. I give a practical example addressing the shallow nature of their bandwidth argument in a tweet here.

During my meandering stroll, I came across several articles discussing Heathrow’s plans to use AI to study traffic flow. These are almost all a year old. I’m curious if anyone has fresh news on the topic. I’m curious if Heathrow has actually learned anything.

An October article on the topic seemed like a press release from the company running a new study in Amerstdam, a release that did more to brand the company’s intellectual property than inform the reader about substantive uses of AI.

The FAA has done a better job of wandering down the path of AI, at least in the sense that they have done it earlier than EASA. One finds it mentioned here, in a strategic plan issued for FY2019-2022. They also studied it several years ago, documented in a poorly titled report here: Identification of Artificial Intelligence. In their defense, the report is not poorly titled, but whoever created the web page to host the report did an incomplete job.

My thoughts have wandered here and there, but I think this is an appropriate response to the apparent behavior of the civil aviation authorities.

There is good news to end this mostly pessimistic collection of rambling prose: The FAA if *finally* hiring someone with AI credentials to work in the Aircraft Certification Office:

Copyright © 2018