The Growing Use of AI in Aviation: Supporting Pilots and Protecting Trust

If you rewind a few years, most of the AI conversation was about replacement. How many jobs could it eliminate? How far could autonomy go?

That’s not the conversation in aerospace.

What I’ve seen over the last few years is a shift away from hype and toward something much more pragmatic. In aerospace, the question isn’t what AI can theoretically do. It’s whether you can certify it. Trust it. Whether you know exactly what you’re going to get. You can’t introduce a system that produces unpredictable outputs. If you can’t explain it and verify it, it doesn’t go on an aircraft.

From Hype to Practical Application

In the broader world, most people’s reference point for AI is a chatbot - large language models, consumer tools. I use them, too. I tried using one to calculate a fantasy football playoff scenario. It was wrong. I had to double-check it.

That’s fine for fantasy football. It’s not fine for aerospace.

So internally, our mindset has shifted from “what’s possible?” to “what’s certifiable?” And that changes the tone of the entire conversation.

Augmentation Is the Undercurrent

One thing I’ve been clear about is that AI in aerospace is about augmentation support, not replacement.

Honeywell Aerospace supports pilots. That’s foundational to who we are. Our integrated flight decks, avionics, APUs, everything we put on an aircraft, is there to help the operator do their job safely.

So when we talk about AI, we’re talking about improving situational awareness and helping pilots manage workload during demanding phases of flight . We’re talking about synthesizing terabytes of data generated during a flight and surfacing insights a human would otherwise have to hunt down manually.

If something happens in flight, pilots still have manuals, sometimes massive ones. Paging through a paper manual or a PDF during a high-workload moment adds friction. AI can help synthesize what’s happening on the avionics side, combine it with how the aircraft normally behaves and offer informed suggestions, which is meaningful in keeping pilots central, supporting them.

The other area where AI can enable new use cases is in autonomy in defense.  There are missions that fall into the 3Ds: dirty, dangerous, and dull where robotics are used in other market segments and I believe we’ll see more of that used in aerospace in defense

So along with AI, we really see a spectrum of pilot augmentation and a shift to highly autonomous operations over the longer term.

Where We’re Using AI Today

There are really two tracks of where we’re using AI. The first is to make our internal  development processes more efficient. I’m genuinely most excited, near term, about what AI can do for our 10,000-plus     member aerospace engineering community.

If we can accelerate development cycles, we can generate draft requirements faster, simulate earlier in the design process and compress time to market. That’s real value. A human still reviews everything. That doesn’t change. But the end-to-end  process can be accelerated and scaled.


The second track is product-focused. We’re researching GPS-denied navigation approaches, using AI to help understand when GPS might be at risk and how to augment position with other signals.  

We’re looking at digital assistants that help during high workload phases of flight. For example, suggesting weather-aware routing so the pilot isn’t trying to manually process multiple competing inputs at once. We’re also using AI to analyze field data by identifying trends in service issues earlier, diagnosing equipment problems more quickly and reducing aircraft downtime. These are contained, targeted applications, solving real pain points.

Why Aerospace Moves Deliberately

Aerospace is one of the safest and most trusted industries in the world. That trust was earned. Millions of people get on airplanes every day assuming they will arrive safely. Cargo moves around the world efficiently. If you introduce AI in a way that jeopardizes that trust, the cost outweighs the benefit.

Certification is a major gating factor. For commercial systems, you cannot show up with a black box that’s right 95 percent of the time and vague the other 5 percent. Requirements must be traceable and behavior must be verifiable. That’s hard with AI today, which means timing matters as much as capability.

Even outside aerospace, trust is still forming. Cities are experimenting with robo-taxis. And as those experiences scale safely, public trust grows.

With aerospace, there’s a confluence that must happen. The technology and regulatory guidance has to mature. Public and operator trust must rise. When those intersect, that’s when you’ll see things start to get adopted. Until then, deliberate progress is the path.

Infrastructure Sets the Pace

Infrastructure helps set the pace and timeline for AI. Right now, airspace modernization is a real priority — and it needs to be. A lot of core air traffic communication systems are still analog. When that’s your baseline, it naturally limits how far and how fast you can scale more advanced capabilities.

If we’re going to introduce smarter decision-support tools, the foundation has to evolve alongside them. The system needs to be more digital, more connected and built for the level of complexity we know is coming.

In Europe, where airspace is already more fragmented, we’re working on modernization efforts with that future in mind, especially as we prepare for the possibility of thousands of new vehicle types entering the system.

AI can help manage that scale. But it can’t compensate for infrastructure that isn’t ready. The two have to move together.

How We Stay Pragmatic

Internally, I think about AI in aerospace as living in two worlds.

One foot is in the research community with AI scientists, conferences and cutting-edge work. The other foot is with OEM customers — hearing directly about their pain points, their operators, their market realities.

Some AI concepts sound impressive in a lab. When you put them in front of a customer, they ask a simple question: why should I care? If you can’t answer that clearly, you probably shouldn’t be pursuing it.

We don’t “throw AI on everything.” If it doesn’t add value, it doesn’t move forward. In some cases, poorly applied AI subtracts value, frustrating people. That’s not what we’re here to build.

Looking Ahead

Longer term, AI and autonomy will influence supply chains, middle-mile cargo and broader accessibility of air travel. Some of that will mature first in defense applications and migrate into commercial uses over time, which is a pattern aerospace has followed for decades. But even then, the underlying principle doesn’t change.

We are not trying to remove people from aviation. We are trying to support them. If we do this right, it should mean safer operations, more predictable travel and better reliability in the field.

And we’ll introduce AI the same way aviation introduces any technology: deliberately, transparently and with the bar for safety exactly where it’s always been.