Google tests AI to reduce aircraft contrails

Google and American Airlines are testing an artificial intelligence tool that helps commercial flights avoid routes where they could form contrails, the white lines aircraft leave in the sky. In a trial involving 2,400 transatlantic flights, flights that followed the recommended plans reduced the contrail formation rate by 62% compared with a control group.

Contrails appear when water vapor from engines freezes under certain atmospheric conditions. Some disappear quickly, but others remain and can trap heat in the atmosphere. So while avoiding a contrail does not eliminate a flight’s CO₂ emissions, it can reduce part of its climate impact.

From testing to incorporating it into flights

The research builds on a test conducted in 2023. At the time, Google and American Airlines showed that pilots could use AI-generated forecasts to avoid areas with a high risk of contrail formation. In 70 test flights, they achieved a 54% reduction.

The problem was that identifying the right flights and coordinating each decision required hours of manual work. The new trial aims to solve that bottleneck: Google’s forecasts were integrated directly into the software American Airlines already uses to plan its flights.

That means the AI does not operate as a separate app that pilots have to check on their own. The system analyzes atmospheric conditions and flags possible alternative routes within the usual planning process.

What it means for passengers

If this technology is adopted widely, passengers would not have to do anything differently. The aircraft could make a slight detour to avoid an area favorable to contrail formation, as long as the alternative is safe, operationally feasible, and reasonable in terms of fuel and flight time.

The result also does not mean that all flights reduce their climate impact by 62%. That figure applies to flights that managed to carry out the avoidance plans during the trial and measures the reduction in the contrail formation rate compared with the control group. It is not an equivalent reduction in all of the flight’s emissions.

The significance of the trial lies elsewhere: it shows that contrail avoidance can be integrated into an airline’s normal operations, rather than being limited to isolated experiments. With more testing, airlines will need to assess how the technique affects fuel consumption, schedules, safety, and air traffic controller capacity.

The next step will be to automate and expand the system without turning every detour into an operational problem. AI can already identify opportunities to avoid contrails. Now airlines need to show that they can act on them consistently, safely, and at a reasonable cost.