AI in Aerospace - a successful panel

With a huge registration of 444 people, the IET's Aerospace Technical Network recently hosted a thought-provoking panel discussion exploring how artificial intelligence (AI) is transforming the aerospace industry. Hosted by Patrick Capaldo, and thoroughly organised by Victoria Rooke and Harry Hancock, the session brought together five experts working at the cutting edge of AI and aerospace:

• Alex Spiri – Digital Engineering Manager, Rolls-Royce
• Alistair Nottle – AI Research Project Leader, Airbus
• Alex Brooker – Founder & CEO, Airside Labs
• Ian Henderson – Head of AI Research, Thales UK
• Dr Mark Hall – AI Research Project Leader, Airbus

The primary motivation for the event was to discuss how we reconcile these two industries: the AI industry that "moves fast and breaks things" and the aerospace industry that is safety-first, methodical, and cautious. The combinations of experience from our guests allowed for diverse perspectives from large organisations to startups, defence applications, frontier research and applying new technology to the real-world.

We covered five main topics in the discussion being:

• Predictive vs. Generative AI
• Industrialising AI Prototypes
• Ethics, Compliance & Regulation
• Human–AI Teaming
• Research Horizons & the Next Decade

We began by examining the shift from predictive to generative AI. Traditionally, AI was about solving well-defined problems with in-house models, such as predictive maintenance. Today, with the advent of large foundation models, AI is more often a service you rent rather than a tool you own. We discussed how this transition is changing the roles of engineers, freeing them from repetitive tasks like writing boiler-plate code and instead allowing more focus on creative problem-solving.

The discussion then moved on to the difficulty of transitioning from prototype to production, that is, the challenge of industrialisation. Here we considered why it's so easy to think of prototype ideas but so hard to scale them. Additionally, however, we also discussed that failure is a necessary part of any problem-solving process, if you don't have this, then you're not really doing anything novel.

We also touched on the topic of ethics, compliance, and regulation and the challenges that comes with abiding by these important principles. One key struggle is for these principles to just keep pace with the rate of technological progress, let alone trying to implement their requirements in a world where regulation is so fragmented. Guests shared their best practices in this space which included keeping the ethics experts and users in-the-loop throughout the entire development process.

In the aerospace industry, we also recognised that whilst there is a baseline level of compliance in all systems, there are some that require more scrutiny than others (for example, an assistant for an engineering tool compared to an assistant for a pilot in the plane).  Echoing this, we also discussed the importance of keeping two-way communication with regulators as this is the likely best option we have for keeping pace with technology.

These themes then lead nicely into the next topic which focused on the human-centric aspect of developing AI for a safety critical industry. The key discussion points for human-AI teaming revolved around trust and how we establish this from a technological and psychological perspective. Within this conversation, we looked at what it takes for a human to trust any form of autonomous system and how the user interface/experience is paramount to getting the outcome you desire. Our guests shared that too little trust is just as bad as too much trust.

Finally, we looked to the future to discuss where we think the next innovations will be at the intersection of these industries. Naturally, this was an impossible question to answer given the rapid pace of change of the AI industry - yet it did provide valuable insights into where the experts see the most value moving forward with the information we currently have. 

From there, we moved into a rapid-fire of questions submitted by the audience covering topics like advice for young professionals, profits and safety, and preventing sensitive data leakage through AI assistants, amongst many others.

A huge thank you to our expert panel, hosts, organisers, and the audience who joined the discussion. This webinar underscored both the opportunities and challenges in applying AI to one of the world’s most safety-critical industries.

If you’d like to continue the conversation, you can connect with the IET Aerospace Technical Network via EngX (https://engx.theiet.org/technical-networks/aero).

Watch the event below: