Airlines turns to artificial intelligence to reach ultra-long flight destinations

Marathon commercial flights that test the limits of long-range jets are getting help from an unlikely source to avoid falling short of their destinations: machine-learning algorithms. At times, Air New Zealand has struggled to get all the way to Auckland from New York in a single hop using its Boeing 787s. Qantas, meanwhile, is adding extra fuel tanks to its Airbus A350s before attempting non-stop trips from Sydney to New York and London in late 2025. The 20-hour flights are set to be the world’s longest regular passenger services. Both airlines are relying on data-hungry software to plot fuel-efficient flight paths and avoid unplanned and embarrassing stops to refuel. The route-planning programs can help pilots avoid heavy weather and catch a tailwind, or even tell them to fly slower to burn less kerosene — anything to squeeze extra miles from the tanks. And rather like an internet search engine that learns on the go, the mapping software is designed to get better the more it’s used. Artificial intelligence is rippling through aviation’s decades-old manual systems, impacting everything from ticket sales to cockpit procedures. While route planning is hardly new, AI offers carriers new ways to save money and fuel as ultra-long voyages proliferate and the mammoth task of reaching net zero emissions by 2050 looms. Flightkeys GmbH calculates about 300,000 flight plans each day for customers including Southwest., American Airlines and Air New Zealand. Story is an edited interview with Raimund Zopp, co-founder and head of innovation at Vienna-based Flightkeys. Zopp, 67, is a former Austrian Airlines AG pilot.<br/>