Thursday Edition: AP-SAS 2026

Boeing has established a set of principles to guide its use of artificial intelligence (AI)-enabled systems that, among other things, establish safety as the priority and set ground rules for human oversight, control, and responsibility.

“The idea was to develop a set of principles to help guide us,” Ben Ivers, director, emerging technologies and regulatory strategy, The Boeing Company, said in opening Day 3 of AP-SAS 2026 in Osaka on Thursday. “Everything really starts with safety,” he said, adding that “whatever we do needs to enhance or maintain safety.”

The principles also say that Boeing AI-enabled systems incorporate human oversight and control, ensuring that humans can safely override or modify system behavior during operation; and that each Boeing AI-enabled system is under the control and oversight of an accountable human, as required. Ivers encouraged other companies to establish their own principles.

During his presentation, Ivers outlined the AI innovation partnership contract (IPC) that Boeing has established with the European Union Aviation Safety Agency (EASA). The idea behind the IPC is to establish regulatory requirements, means of compliance (MOC), and the verification and validation strategy for a machine learning-based system; and to use an EASA concept paper on AI as the basis for the requirements and MOC.

The project uses Boeing’s experimental automated taxi and safe runway systems as surrogates for the certification process, he said. The purpose of the project is to take work EASA has done and evaluate it by running it through a real-world system. A paper on the project is expected later this year, he said.

The future of aviation safety is not about choosing between people and technology, but about combining them, according to Ingrid Lagarrigue, North Asia Representative, EASA.

Addressing the topic “From Data Chaos to Smarter Insights for Safer Operations” as part of an expert panel, she said that predictive safety intelligence is built on cooperation, data sharing, and mutual trust. As many other speakers have noted this week, Lagarrigue said that “AI should support humans, not replace them”

AI enables organizations to process larger volumes of information, identify patterns, and provide faster, more accurate, and more predictive safety intelligence that supports better decision-making, she said. But developing safety intelligence also involves the expert judgment of humans, benchmarking, and prioritizing.

During his presentation, Kohei Funabiki, senior research engineer, Aeronautical Facility Research Unit, Aviation Technology Directorate, Japan Aerospace Exploration Agency, said he was shocked by the fatal runway incursion accident at Haneda Airport in January 2024. Despite more than 50 serious incursions in the prior 25 years, there had never been an accident.

He said that historical data alone cannot predict accidents of types that have never occurred and suggested that model-based simulation can explore future accident scenarios beyond observed data. He said these models should reflect underlying physical or human-performance mechanisms rather than relying solely on statistical relationships.

Transportation is undergoing a transformational change akin to the Wright Brothers’ first powered flight or the dawn of the jet age, according to Akbar Sultan, director, Office of Research and Engineering, U.S. National Transportation Safety Board. With that as context, he said that maintaining the status quo in terms of data analytics or taking a measured iterative approach is insufficient to address future safety requirements. “We need to fundamentally change the way we do our safety analytics,” he said.

The empathy gap is real, and AI, if developed correctly, can be a tool to fill that gap, according to Dr. Fabio Mattioli, associate professor, The University of Melbourne.

Mattioli and his colleagues have built an aviation chatbot, and research based on use of the chatbot indicates that 71 percent of the situations that crews bring to the AI chatbot are interpersonal, yet only 1.5 percent of users frame their questions empathetically, he said, adding that crews treat human factors issues as threats to manage and not people to understand.

AI shows unexpected strength in training, he said. Instructors’ emotional intelligence can predict student progression, and AI can help instructors understand and adapt to trainees.

Advanced air mobility (AAM) companies are not creating a new aviation system but rather want to extend the existing system with new technology, according to Earl Lawrence, aviation policy lead for advanced projects at Joby Aviation.

The greatest safety challenge for AAM is building confidence across the stakeholder ecosystem. For piloted powered-lift aircraft, the challenge is less about creating a new safety system and more about demonstrating how these aircraft integrate into the proven aviation safety framework — through certification, pilot training, maintenance, air traffic management, and emergency response, according to Lawrence. The aircraft may be new, but they are operated in a traditional manner.

“What we’re trying to do is integrate a new ecosystem into the well-established system we have today,” said Simon Whalley, chief regulatory and safety office, Skyports Infrastructure. He said AAM integration is fundamentally a systems challenge and that the near future will involve hybrid operations.

Eguchi Makoto, director, Unmanned Aircraft Systems (UAS) Division, Japan Civil Aviation Bureau, said Japan released its AAM roadmap in March and that the map envisions commercial operations starting in some areas in 2027 or 2028. The early 2030s will see the introduction of new traffic management and passenger transport by remotely piloted aircraft, and the late 2030s will see the partial implementation of automated and autonomous operations. The roadmap envisions a society where air mobility is part of everyday life in the 2040s and beyond.

Richard Stocker, national manager, Airworthiness and Engineering Branch, Civil Aviation Safety Authority (CASA), Australia, said among the questions that need to be answered is how AAM validation will occur in the State of design and how that validation will be transferred to the State of registration. He said there will need to be collaborative partnerships between regulators, between regulators and AAM manufacturers, and between the State of design and the State of registration.

Dr. Nobuo Kishi, board director, executive vice president, chief aviation safety officer, SkyDrive, said his company’s first use case is for sightseeing, but that an emergency response use case will be important in gaining public acceptance of AAM technology. His overall message was that global harmonization that transcends technical, infrastructure, and regulatory barriers will be necessary for the widespread adoption of AAM.