ARTIFICIAL INTELLIGENCE AI in aviation Should we be wary or could it help the industry? By First Officer Chris Belfield, Log Board member A rtificial Intelligence (AI). Its the big buzzword of the 2020s and its prevalence is only going to increase in the coming years. From McDonalds using it for order prediction at drive-throughs to ChatGPT and its ilk redesigning the arts, AI is our next stepping stone through the digital age but what is it, and how does it apply to aviation? When we use the term AI in current parlance, what we are generally referring to is known within the tech industry as machine learning which, in simple terms, gives computers the ability to learn without explicitly programming them. Traditional models require detailed instructions for a computer to follow, which can be time-consuming and resource-intensive for the developers when done manually. For some tasks, the length of the required programming makes the task impossible. With AI, machine learning feeds large datasets into a computer so that it can learn to see patterns and, in doing so, make future predictions based on the information provided. Have you ever had to fill out captchas online, identifying a bicycle, series of buses or builders bums for a security check? While that helps to identify you as human, it also helps to train these massive AI projects. Behind the scenes Aviation, as always, is slow to change. With such an overwhelming focus on safety (and cost) companies rarely implement new technology without lengthy safety testing and checking processes. Many of us fly with an EFIS, designed and built in the 1980s. These technologies are tried and tested, and changes to them tend to be evolutionary, not revolutionary. It will take a few years of hard testing before we see Star Trek-esque cockpit designs in passenger aircraft (although those of you who have the privilege to fly Gulfstreams may already be familiar with some of the recent innovations in their design). AI is already having an impact behind the scenes, however. Major companies and government bodies are working hard to bring these tools to the aviation sector. Take easyJets innovation with FOMAX and Skywise, for example. Since 2018, FOMAX (Flight Operations and Maintenance Exchanger) has been collecting all aircraft system and performance data across the entire easyJet fleet. It collaborates with Skywise, an Airbus service that parses the data and aims to predict impending failures within individual aircraft. Old sensors onboard an easyJet airliner collected around 390 parameters, whereas FOMAX gathers upwards of 10,000 and feeds them all into Skywise for data prediction. Skywise, in turn, produces specific reports gathered from a combination of aircraft data, Airbus feeds, and other information points. Reports cover a range of factors, including fleet reliability, individual component reliability, and logistics optimisation, plus many other specific sectors of its business. For those of us in the cockpit, predictive maintenance is probably the most important result of all this. In the example given, aircraft G-EZAJ was identified with a faulty engine fuel LP twin motor actuator. Notably, this was before any faults were being displayed on the aircraft. If left to fail this would mean, for management, an aircraft out of service; for us as crew, an unexpected and potentially very serious event. With corrective service performed ahead of an impending fault, pilots can have more confidence in their aircrafts ability to perform consistently. That means fewer surprises and less potential mistakes in the cockpit. There is much to be enthusiastic about in the aviation sector with the emergence of machine learning 24 THE LOG Sum 24 pp24-25 AI in Aviation.indd 24 13/06/2024 12:44