Written by Kim Macaulay, IATA Senior Vice President Information and Data, Chief Information and Data Officer
We are often asked how much AI is revolutionizing the aviation industry. The simple answer is “a lot,” but the more complex answer is that the degree of penetration of AI into airline operations will vary widely depending on what part of the business we are talking about.
Consider the aviation industry’s top priority: safety. Scaling AI for aviation safety is less about technological maturity and more about organizations evolving to absorb AI as their ability to analyze vast amounts of data for prediction, text classification, and efficient image and speech recognition increases. Here, AI offers the advantage of processing large amounts of information to support faster and more accurate decision-making in identifying and managing safety risks.
New use case
There are many other areas where AI is already being implemented by airlines. Airlines are investing in training their own AI systems to help calculate things like route optimization, predictive maintenance scheduling, sales efficiency, revenue maximization, and forecasting. AI-powered chatbots also offer opportunities to improve customer service. Of course, not everyone wants to talk to a robot, but chatbots are likely an upgrade to standard information-searching capabilities on websites, and once they can handle basic queries, human customer service agents can help with more complex issues. Overall, I believe that chatbots are more than just a cost-saving measure, they can improve the customer experience.
AI agent
Perhaps the most interesting application of AI today is in the area of “agent AI.” There, AI “agents” work autonomously on our behalf, performing increasingly complex tasks. For example, IATA publishes dangerous goods regulations every year. This is a huge document covering thousands of rules for the safe handling of goods. Agentic AI allows freight forwarder agents to chat with shipper agents and freight forwarder agents to arrange shipments and ensure that they are correctly labeled and packaged according to DGR rules with minimal human oversight.
For those who wonder if there is a safety risk, remember that as AI improves, mistakes will gradually decrease, but most humans in the chain are probably already at “peak” performance. Human oversight and accountability are essential, but humans are not infallible. In fact, your machine will end up making fewer errors.
How quickly will these AI applications be implemented? It’s happening now. Airlines are investing heavily in data and large-scale language models (LLMs). The first step is to ensure you develop a reliable framework for sharing data. It is clearly important that commercially sensitive information cannot be hacked and that critical operational data can be shared confidentially between industry partners.
cost reduction
Cost remains a barrier. Airline profit margins are much thinner than many other large business sectors, reducing potential investment in expensive, cutting-edge AI. However, “Hypermoore’s Law” will dramatically increase computing power and reduce costs, making AI applications accessible to more airlines. Currently, around 20-40% of airline coding is done by AI, but developments in this area are expected to increase in the coming years, resulting in significant productivity gains. Airlines see a potential opportunity in this. For example, IAG and Emirates announced a partnership with OpenAI.
Ultimately, we can expect shared AI models to work between airlines, airports, ANSPs, and ground handlers to optimize efficiency across the aviation chain. Baggage and cargo, air traffic management, slot allocation, and passenger biometrics are just some of the obvious applications that will further improve operations and the passenger journey experience.
Bad news for jobs?
Perhaps the biggest question facing companies and employees on the eve of this AI revolution is: “Will this lead to large-scale layoffs?” My honest belief is that, at least in the aviation industry, this is not the case. Some traditional “back office” roles and support functions will be taken over by AI. But new jobs could also be created. And with passenger numbers expected to double by 2050, more pilots, flight attendants and ground staff will be needed. We still seek and expect that human touch, whether it’s the purser’s smile on the gangway or the pilot’s reassuring voice relaying information to us at 30,000 feet. It’s not going to be replaced by AI anytime soon.
