Artificial intelligence is developing at a rapid pace and is increasingly being used by scientists to better understand climate change. While AI is perhaps best-known for its prominent role in large language models such as ChatGPT, Claude, and Gemini, the broader scientific field is vast and growing exponentially. From advanced weather forecasting to analyzing massive quantities of satellite imagery, climate scientists are actively leveraging various types of artificial intelligence to develop faster and more efficient ways to study a rapidly warming planet.

However, this technological revolution comes with no shortage of risks and challenges. Researchers warn of an overreliance on unverified results, intensive energy and water consumption by data-heavy systems, and the potential to exacerbate existing inequalities in climate science globally. To better understand these complex dynamics, Carbon Brief spoke with ten experts working across a range of climate-related fields—spanning everything from developing long-term climate forecasts to analyzing structural changes in modern energy grids—about how AI is fundamentally reshaping their work.

What are the Benefits and Challenges of Using AI in Weather and Climate Forecasting?

One of the most significant recent developments in the discipline has been the striking success of AI-based weather-prediction models. These systems learn from vast volumes of historical and real-time weather data to predict conditions a few hours ahead, then repeatedly feed those predictions back into themselves to generate reliable forecasts up to two weeks into the future.

Dr. Doug McNeall, science lead in data-driven climate modelling at the UK Met Office, notes that these models have demonstrated that artificial intelligence can provide accurate predictions while requiring far less computing power than traditional numerical models, which rely heavily on complex mathematical equations to simulate the atmosphere.

Yet, significant challenges remain. AI weather-prediction models have primarily been developed for relatively short timescales, where large amounts of observational data regarding temperature, rainfall, and sea ice extent are readily available, allowing model performance to be evaluated quickly and rigorously. Climate prediction, however, is fundamentally different. To provide useful climate information, scientists need to make projections for future climates that lie entirely outside the historical observational record.

One promising route forward, according to Dr. McNeall, is using AI to emulate or accelerate existing numerical models rather than relying solely on purely data-driven approaches. This blended methodology could make climate risk information far more accessible and affordable to produce, while retaining the physical understanding based on fundamental scientific principles that underpins confidence in long-term climate projections.

Prof. William Collins, a professor of AI for climate science at the University of East Anglia, highlights a renaissance currently underway across weather and climate science. Machine learning, a subset of AI that uses algorithms to let computers learn from data and make predictions without being explicitly programmed, has had an enormous impact on numerical weather prediction. For instance, the European Center for Mid-Range Weather Forecasts is now deploying an operational AI forecast system alongside its traditional forecasting framework for the European community, proving that machine learning systems have become fully competitive with the world’s leading traditional forecasting platforms.

Nevertheless, the central challenge remains daunting. Much of artificial intelligence involves learning from historical data, yet the climate represents an unchartered future for which direct observational data does not exist. Researchers are literally trying to predict a future for which no observations are yet available, prompting the global scientific community to search for innovative ways to inform and constrain these predictive models.

Dr. Amirpasha Mozaffari, a researcher in Earth artificial intelligence at the Barcelona Supercomputing Centre, points out that while AI forecast models perform well on typical weather patterns, they consistently underperform physics-based models when applied to record-breaking and unprecedented extremes. These extreme events are precisely what matter most for societal adaptation to climate change. Solving this limitation—whether through hybrid models, physical constraints, or new evaluation methods—remains the central open problem for the field.

What are Some Other Current Uses of AI in Climate Science Research?

Beyond core meteorological modelling, the proliferation of Earth-observation satellites has created a massive influx of geospatial data, including high-resolution images mapping deforestation and melting sea ice. Dr. Stefano Materia, a researcher in climate variability and change at the Barcelona Supercomputing Centre, explains that AI is increasingly used to translate raw pixel data into actionable climate intelligence.

Methane tracking and ecosystem monitoring serve as prime examples of this transformation. Methane is a potent greenhouse gas, but leaks are largely invisible and difficult to track manually. AI algorithms now automatically scan hyperspectral imagery from satellites to pinpoint specific industrial methane leaks globally, alerting operators in near real-time. Similarly, machine learning models process thousands of satellite images daily to detect illegal logging, track shifting land-use patterns, and monitor glacier retreat long before human researchers could manually review the footage.

Dr. Clare Eayrs, a polar research scientist at the Environmental Fluid Dynamics Lab at New York University, notes that machine learning has integrated into nearly every aspect of the sea ice modelling workflow. The most mature applications involve extracting finer details from observations, such as mapping sea ice deformation from ship radar and classifying snow layers in micropenetrometer profiles. In forecasting, deep-learning systems like Ice Net have demonstrated an ability to outperform leading physical models in seasonal Arctic forecasts, predicting critical variables such as sea ice concentration, thickness, and drift.

Institutional frameworks are also adapting. Prof. Cristiana Stan, vice-chair of the World Climate Research Programme, points out that AI now appears across all major agency initiatives. The weather prediction model intercomparison project includes physics-based numerical models, hybrid configurations, and AI weather systems side-by-side, accelerating the deployment of next-generation operational models. Downscaling is another critical area harnessing AI power to refine coarse-resolution model outputs into high-resolution regional data.

Furthermore, the Intergovernmental Panel on Climate Change (IPCC) has incorporated an AI model intercomparison project specifically for its seventh assessment report (AR7). Dr. Mozaffari notes that machine learning architectures—such as diffusion models, foundation models, and generative ensembles—are now migrating into climate science within months of their appearance in the broader tech sector, drastically dropping the cost of running global forecasts.

How is AI Being Used to Advance Climate Policy and Action?

As power grids integrate higher volumes of solar, wind, and other renewable energy sources, managing complex supply and demand dynamics requires sophisticated oversight. Prof. Priya Donti, assistant professor at the Massachusetts Institute of Technology and co-founder of Climate Change AI, explains that artificial intelligence is being used to generate superior predictions of solar and wind power production, thereby reducing uncertainty and allowing grid operators to plan with greater foresight. AI also helps optimize and control power generators, batteries, industrial loads, and flexible endpoints like electric vehicle chargers and smart heating and cooling systems.

Dr. Carlos Rodriguez-Pardo, a postdoctoral researcher at Politecnico di Milano and affiliated scientist at the European Institute on Economics and the Environment and the Euro-Mediterranean Center on Climate Change, adds that AI supports every stage of the climate policy cycle. In the measurement phase, machine learning helps process satellite imagery to estimate emissions, improve air quality monitoring, harmonize fragmented policy databases, and map an overwhelming body of scientific literature.

During the futures-modelling phase, AI-based emulators reproduce the behavior of computationally expensive climate and economic scenarios at a fraction of the cost, allowing researchers to explore thousands of scenarios and quantify previously unreachable uncertainties. Reinforcement learning also permits the study of how multiple actors, such as sovereign nations or energy producers, compete and cooperate under varied policy frameworks. Finally, in the evaluation phase, causal machine learning extends the toolkit used to rigorously determine whether a specific implemented policy actually achieved its intended outcomes.

How Could AI be Used in Climate Research Over the Coming Years?

Looking ahead, experts anticipate profound structural shifts in how climate research is conducted. Prof. William Collins suggests that future assessment reports, such as the upcoming IPCC evaluations, may increasingly rely on large language models to synthesize sprawling bodies of scientific literature, helping researchers identify critical "known unknowns" and direct future research priorities.

Dr. Stefano Materia emphasizes that hybrid modeling—merging the physical consistency of traditional Earth system models with the pattern-recognition speed of machine learning—represents a true paradigm shift. This approach overcomes computational bottlenecks, addresses unresolved subgrid processes, and improves understanding of the terrestrial biosphere, helping to narrow uncertainties surrounding the terrestrial carbon sink and identify non-linear tipping points.

Additionally, researchers like Dr. Clare Eayrs emphasize that machine learning is uniquely suited to extract historical climate data from centuries-old ship logs, whaling records, expedition photographs, and pre-1979 aerial archives, potentially lengthening the observational record of systems like Antarctic sea ice and improving the training of future AI algorithms.

Are There Misconceptions About AI in Climate Science?

A recurring concern among researchers is the public conflation of climate-focused AI with massive commercial generative models. Dr. David Rolnick, associate professor at McGill University and co-founder of Climate Change AI, points out that while the heavy energy and water footprints of large-scale generative models and AI agents are genuine societal worries, those resource-intensive systems are largely distinct from the specialized, smaller-scale AI applications deployed to fight climate change. Prof. Priya Donti echoes this concern, warning that mismatched public and funding perceptions can result in misallocated research dollars.

Could AI Improve or Exacerbate Inequalities in Climate Research?

Questions of equity remain central to the discourse. Dr. Amirpasha Mozaffari cautions that frontier AI models are built almost exclusively in the Global North, trained on data densest in observation-rich regions, and validated against matching benchmarks. Mere access to "cheap inference"—the ability to run a pre-trained model—does not grant vulnerable regions a say in how models are trained or evaluated. Without deliberate efforts to share data infrastructure and center local wellbeing, AI risks automating and amplifying existing global divides.

Conversely, Dr. Carlos Rodriguez-Pardo argues that well-implemented AI-based Earth system models could eventually lower barriers to entry, enabling small research groups or agencies in data-poor regions to adapt pre-trained models on modest hardware. However, Dr. Rolnick stresses that technological consolidation can easily run counter to the democratization of climate action if institutional developers fail to understand or prioritize the needs of the majority of the world’s population.

What are the Remaining Risks and Challenges?

Ultimately, trust remains the paramount challenge. Dr. Clare Eayrs warns that the apparent success of AI systems must not be used as a justification to disinvest in vital physical observations, noting that AI is never a substitute for empirical data and physical understanding. Dr. William Collins cautions against "truthy" outputs—answers that look plausible but lack underlying physical validity.

As the scientific community navigates the intersection of academia, public institutions, and private tech industry data centers, researchers emphasize that artificial intelligence must remain a tool to augment human judgement in climate decisions rather than replace it, ensuring that technological innovation serves as a reliable force for global progress.