A groundbreaking trial conducted by Google and American Airlines demonstrates that artificial intelligence can effectively reduce the climate impact of air travel by modifying flight paths to avoid the formation of contrails. The study, involving over 2,400 real flights between the US and Europe, revealed an overall reduction of 11.6% in visible contrail formation when pilots were given AI-optimized routing options.
The Problem with Contrails: More Warming Than CO₂?
For years, aviation’s environmental impact has been largely framed around carbon dioxide emissions. However, recent research suggests that contrails – the condensation streaks left by aircraft engines – may contribute more to warming than CO₂ itself, especially in certain atmospheric conditions. These ice crystals trap heat and can have a significant cumulative effect on global temperatures.
The key is predictability: AI can now forecast where these problematic ice-rich regions of the upper atmosphere will form, enabling flights to be rerouted. While small-scale tests showed promise, this trial marks the first large-scale, randomized, controlled implementation in commercial operations.
How the Trial Worked
From January to May 2025, flights heading eastward were split into two groups. One group’s dispatchers had access to AI-suggested routes designed to minimize contrail formation; the other did not. The trial focused on nighttime flights because contrails cause more warming at night (when they don’t reflect sunlight back into space).
Despite the option being available, only 112 out of 1,232 flights in the AI-assisted group actually took the alternate route, citing operational concerns like cost and safety. Even with limited adoption, the results were significant: flights following the AI-optimized paths showed a 62% reduction in visible contrails.
Implications and Scaling Challenges
The team estimates that the warming effect of the entire group with suggested routes was reduced by 13.7%, while flights that took the optimized route saw a staggering 69.3% decrease. Importantly, fuel consumption remained statistically unchanged, suggesting that contrail avoidance doesn’t necessarily add operational costs.
Edward Gryspeerdt of Imperial College London notes this is “probably the best you can do with current tools.” However, scaling up to a 60% reduction across all flights is unlikely due to real-world complexities in air traffic management. Even a 10% reduction, however, would represent a meaningful step toward greener aviation.
“This validated the thesis: if we can integrate this safely into flight planning, it’s a scalable way to address contrail avoidance.” – Dinesh Sanekommu, Google.
The trial underscores the potential of AI to mitigate climate impact beyond simply reducing carbon emissions. As predictive modeling improves and operational integration becomes smoother, expect this strategy to become a standard practice in the aviation industry.
