New method could make AI less biased

Scientists have developed a new approach that could help reduce each type of bias in artificial intelligence (AI) while maintaining the same level of predictive accuracy.

With machine learning systems being used to determine everything from stock prices to medical diagnoses, it has never been more important to look at how they arrive at decisions, said researchers from the Massachusetts Institute of Technology (MIT) in the US.

The new approach demonstrates that the main culprit is not just the algorithms themselves, but how the data itself is collected.

"Computer scientists are often quick to say that the way to make these systems less biased is to simply design better algorithms," said Irene Chen, a PhD student at MIT.

"But algorithms are only as good as the data they're using, and our research shows that you can often make a bigger difference with better data," Chen said.

Looking at specific examples, the researchers were able to both identify potential causes for differences in accuracies and quantify each factor's individual impact on the data.

They then showed how changing the way they collected data could reduce each type of bias while still maintaining the same level of predictive accuracy.

"We view this as a toolbox for helping machine learning engineers figure out what questions to ask of their data in order to diagnose why their systems may be making unfair predictions," said Professor David Sontag from MIT.

Chen said that one of the biggest misconceptions is that more data is always better.

Getting more participants doesn't necessarily help, since drawing from the exact same population often leads to the same subgroups being under-represented.

Even the popular image database ImageNet, with its many millions of images, has been shown to be biased towards the Northern Hemisphere, researchers said.

According to Sontag, often the key thing is to go out and get more data from those under-represented groups.

For example, the team looked at an income-prediction system and found that it was twice as likely to misclassify female employees as low-income and male employees as high-income.

The researchers found that if they had increased the dataset by a factor of 10, those mistakes would happen 40 per cent less often.

In another dataset, the researchers found that a system's ability to predict intensive care unit (ICU) mortality was less accurate for Asian patients.

Existing approaches for reducing discrimination would basically just make the non-Asian predictions less accurate, which is problematic when you're talking about settings like healthcare that can quite literally be life-or-death.

Chen said that their approach allows them to look at a dataset and determine how many more participants from different populations are needed to improve accuracy for the group with lower accuracy while still preserving accuracy for the group with higher accuracy.

"We can plot trajectory curves to see what would happen if we added 2,000 more people versus 20,000, and from that figure out what size the dataset should be if we want to have the best of all worlds," said Chen.

"With a more nuanced approach like this, hospitals and other institutions would be better equipped to do cost-benefit analyses to see if it would be useful to get more data," Chen said.