IBM Research launches explainable AI toolkit

IBM Research today introduced AI Explainability 360, an open-source collection of state-of-the art algorithms that use a range of techniques to explain AI model decision making.

The launch follows IBM’s release of AI Fairness 360 a year ago for the detection and mitigation of bias in AI models.

IBM is sharing its latest toolkit in order to increase trust and verification of artificial intelligence and help businesses who comply with regulation in order to use AI, IBM Research fellow and responsible AI lead Saska Mojsilovic told VentureBeat in a phone interview.

“That’s fundamentally important because we know people in organizations will not use or deploy AI technologies unless they really trust their decisions. And because we create infrastructure for a good part of this world, it is fundamentally important for us not because of our own internal deployments of AI or products that we might have in this space, but it’s fundamentally important to create these capabilities because our clients and the world will leverage them,” she said.

The toolkit is also being shared, Mojsilovic said, because industry progress on the creation of trustworthy AI has been “painfully slow.”

AI Explainability 360 draws on papers and algorithms made by IBM Research group members like “TED: Teaching AI to Explain Its Decisions,” a paper accepted for publication at AAAI/ACM conference on AI, ethics, and society, as well as the often cited “Towards Robust Interpretability with Self-Explaining Neural Networks” accepted for publication at NeurIPS 2018.

The toolkit draws on a number of different ways to explain outcomes such as contrastable explanations, an algorithm that attempts to explain important missing information.

Algorithms are also made for particular scenarios, such as Protodash for explaining prototypes, and algorithms designed to explain credit score models results to a consumer recently denied a loan as well as to a loan officer who needs an explanation of AI model decision making in order to comply with law.

All eight algorithms in the toolkit today come from IBM Research, but more from the wider AI community will be added in the future.

“It’s not one team of researchers or one organization that can move the needle, we all benefit when we join forces and do it together and that’s why we intend to grow the toolbox,” Mojsilovic said.