A new way to look at climate change: using geoengineering

article A new approach to modelling the effects of climate change is using artificial intelligence and cloud computing, to understand how human behaviour changes over time, according to a team from MIT and the University of British Columbia.

Their research is published in Nature Climate Change.

It was inspired by a number of previous work on the issue, said Dr Stephen Meyer, a senior research fellow at MIT and a co-author of the study.

In general, artificial intelligence models the behaviour of people and groups of people over time by using artificial neural networks, which are computers with the capability to simulate the actions of other computers.

“Our work focuses on the ability to create an artificial intelligence that is very, very intelligent,” said Meyer.

The researchers wanted to find out how well they could do that, to see if their results could be used to better understand climate change.

“This kind of work can inform us about what the best way to reduce greenhouse gas emissions is,” he said.

To get to the answer, they used a new method to analyse climate models for the past two decades, and then compared their results to observations.

The method involves analysing a set of historical datasets, and modelling the way they have changed over time.

The models are fed with data on the past several decades, but the models also use a range of different scenarios, from a world without human influence to one where humans are a major driver of climate changes.

“We are trying to understand why the climate has changed in the way it has over time,” said Professor Christopher Trenberth, from the University’s Department of Earth and Atmospheric Sciences.

“That can help us to make decisions about how to reduce emissions.”

The researchers ran simulations with a wide range of scenarios, and found that they could identify the key differences between different scenarios.

For instance, they found that while the simulations showed no sign of significant warming over the past decade, the data did show a rise in CO2 emissions, which was not predicted by the models.

The model was able to produce a very convincing prediction, which is the increase in global average temperatures by the end of the century, which they said could be considered as a positive sign of progress.

But while the model predicted that the rise in temperature would occur in the next century, the fact that it did not happened in the century that followed was not surprising.

“When you run simulations, you run a set amount of time, and you have to make predictions,” said Trenberg.

“And that’s because the models do have a set number of parameters.

The uncertainty in the model can be quite large.”

But when it comes to predicting what the future will look like, it was difficult to determine which model was better.

“It’s difficult to say, ‘This is the best one’ or ‘This one is better’,” said Tredberth.

“If you look at this model, you see that it is not very good at predicting the future.”

The study also showed that the best models were only able to predict about half of the warming observed over the previous century.

“The modelers really didn’t have a lot of flexibility,” said Dr Meyer.

“They could make predictions that were more accurate, but they were not very flexible.”

So while the models are able to accurately predict how the climate will change over time – the authors believe this is due to the uncertainty in their models – the models may not be the best ones to make that prediction.