For decades, the public perception of AI came down to one of two choices. One was the friendly robots who work and live alongside humans, like WALL-E or Data from Star Trek. The other was a dystopian vision embodied by Skynet in The Terminator or HAL 9000 in 2001: A Space Odyssey.

While both concepts still lie firmly in the world of fiction, the reality is that—even if we don’t yet realize it—AI is already impacting the lives of millions around the globe. Driven in part by the acceleration of AI capabilities in recent years, a growing number of companies are exploring the use of AI as a tool to help reach their sustainability goals.

Today, AI systems can monitor and analyze emissions to identify areas for improvement, increase energy efficiency and optimize the integration of renewable energy resources into the grid. As the energy transition continues, the role of AI in sustainability is poised to grow dramatically in the coming years.

While the impact of AI depends on each organization’s strengths and capabilities, there are five major areas where it will have the most impact going forward: identifying opportunities, efficiency, flexibility, reliability and optimization.

Identifying Opportunities

Among the most exciting applications of AI is identifying new sustainability opportunities.

As an example, companies might use generative AI to automatically suggest a range of options for converting CO2 into methanol, and then use optimization to identify the best option based on the specific challenges they face.

In other cases, companies can use AI to interpret geological and topological information to identify the best areas for renewable energy generation or to identify subsurface reservoirs for storing CO2. The technology could even be used by mining companies to target specific areas where critical resources—like rare earth minerals for use in electric motors or lithium for energy storage—might be found.

Generative AI can even be used to propose new materials that are optimized around a particular set of properties. This approach could allow companies to develop new materials that create more efficient batteries, improved solvents for carbon capture and more.

Efficiency

Though they have been a useful approach to modeling industrial processes for years, first principles models suffer from what is known as the simulation-to-reality gap. Simply put, industrial processes are so complex that it’s impossible to model every aspect.

Instead, many companies are now turning to AI-driven hybrid models. Because they are built using data collected from real assets, such hybrid models can close that gap. The result is more accurate models, which companies can use to drive increased efficiencies in a host of processes.

Once implemented, those hybrid models can be accelerated through relatively inexpensive GPUs or tensor processing units. That increased computing power allows companies to run “what if?” simulations and complex, probabilistic models to explore different use cases. In addition, companies can create plant-wide simulations or use generative design to explore and identify ways to increase efficiency.

AI-driven models could also allow companies to easily create robust schedules and optimize supply chains. Those efforts can minimize the emissions that arise from transporting inventory across the globe, as well as limiting the need to warehouse unnecessary inventory.

Flexibility

Hybrid models driven by AI also allow companies far greater flexibility, particularly when it comes to the deployment of new sustainability applications like direct air carbon capture.

With traditional first-principles models, those systems are optimized only for a particular design and a particular location. That means that as the technology improves and newer designs are built in other locations, engineers would need to create a new model for every installation.

Hybrid models, however, can be adapted and optimized each time based on data from their unique conditions, enabling faster development and deployment.

AI is also poised to play a key role in managing electrical grids going forward. From helping to predict renewable energy production based on weather and other factors to managing distributed energy resources like smart thermostats, AI technology will be critical to balancing increasingly complex grids.

Reliability

A growing number of companies are employing AI-enabled asset performance management (APM) software to develop predictive and prescriptive maintenance programs that can predict failures even before they happen.

By minimizing or even eliminating unexpected failures, these tools can help companies save millions from unplanned downtime, as well as reduce environmental impacts. In some cases, unplanned shutdowns can release more carbon in a single incident than is released in a year of normal operations. By stopping those incidents before they happen, AI can help significantly reduce emissions.

Reducing or eliminating unplanned downtime also allows companies to create managed, orderly and safe maintenance plans that fit their operations. Those plans help to reduce the need for spare parts and allow for better planning of personnel and costly repair equipment.

Optimization

Lastly, AI can play a key role in helping to simplify the global challenges that come with the energy transition.

As an example of that simplification, consider the challenges of incorporating EV chargers into the layout of traditional gas stations. Using AI tools, companies can create customized layouts that optimize the space for both EVs and internal combustion engine vehicles, speeding up the design of facilities in different locations.

Similar technology could also be applied to the design of carbon capture facilities, ensuring each facility can capture the maximum amount of carbon, helping to reduce emissions and meet net-zero goals.

Conclusion

Significant challenges lie ahead as the world works to navigate the energy transition and create a sustainable future for everyone around the globe. While there’s little doubt that technology will play a critical part in those efforts, AI—if properly applied and responsibly managed—will be an invaluable tool in making that future a reality in the decades to come.

This article originally appeared on Forbes.com