IBM Research is building and enabling AI solutions people can trust.
Artificial intelligence systems are increasingly being used to support human decision-making. While AI holds the promise of delivering valuable insights and knowledge across a multitude of applications, broad adoption of AI systems will rely heavily on the ability to trust their output. Human trust in technology is based on our understanding of how it works and our assessment of its safety and reliability. To trust a decision made by an algorithm, we need to know that it is reliable and fair, that it can be accounted for, and that it will cause no harm. We need assurance that it cannot be tampered with and that the system itself is secure. We need to understand the rationale behind the algorithmic assessment, recommendation or outcome, and be able to interact with it, probe it – even ask questions. And we need assurance that the values and norms of our societies are also reflected in those outcomes.
Moving forward, “build for performance” will not suffice as an AI design paradigm. We must learn how to build, evaluate and monitor for trust. IBM Research AI is developing diverse approaches for how to achieve fairness, robustness, explainability, accountability, value alignment, and how to integrate them throughout the entire lifecycle of an AI application.
AI Explainability 360 Open Source Toolkit
This extensible open source toolkit can help you comprehend how machine learning models predict labels by various means throughout the AI application lifecycle.
As AI advances, and humans and AI systems increasingly work together, it is essential that we trust the output of these systems to inform our decisions. Alongside policy considerations and business efforts, science has a central role to play: developing and applying tools to wire AI systems for trust. IBM Research’s comprehensive strategy addresses multiple dimensions of trust to enable AI solutions that inspire confidence.
Transparency and Accountability
AI Explainability 360 Toolkit
This extensible open source toolkit can help you comprehend how machine learning models predict labels by various means throughout the AI application lifecycle. Containing eight state-of-the-art algorithms for interpretable machine learning as well as metrics for explainability, it is designed to translate algorithmic research from the lab into the actual practice of domains as wide-ranging as finance, human capital management, healthcare, and education.
IBM scientists suggest that AI services be accompanied with a factsheet outlining the details about how it operates, how it was trained and tested, its performance metrics, fairness and robustness checks, intended uses, maintenance, and other critical details.
AI Fairness 360 Toolkit
This extensible open source toolkit can help you examine, report, and mitigate discrimination and bias in machine learning models throughout the AI application lifecycle. Containing over 70 fairness metrics and 10 state-of-the-art bias mitigation algorithms developed by the research community, it is designed to translate algorithmic research from the lab into the actual practice of domains as wide-ranging as finance, human capital management, healthcare, and education.
Adversarial Robustness 360 Toolbox
The Adversarial Robustness Toolbox is designed to support researchers and developers in creating novel defense techniques, as well as in deploying practical defenses of real-world AI systems. Researchers can use the Adversarial Robustness Toolbox to benchmark novel defenses against the state-of-the-art. For developers, the library provides interfaces which support the composition of comprehensive defense systems using individual methods as building blocks.
IBM Research AI is developing diverse approaches for how to achieve fairness, robustness, explainability, accountability, value alignment, and how to integrate them throughout the entire lifecycle of an AI application.
Please explore all of our trusting AI related research papers
|AI Fairness 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias||Fairness|
|Fairness GAN: Generating Datasets with Fairness Properties using a Generative Adversarial Network||Fairness||ICLR (2019)|
|Efficient Neural Network Robustness Certification with General Activation Functions||Robustness||NeurIPS (2018)|
|Boolean Decision Rules via Column Generation||Explainability||NeurIPS (2018)|
|Automated Test Generation to Detect Individual Discrimination in AI Models||Fairness|
|Analyzing Federated Learning through an Adversarial Lens||Robustness||ICML (2019)|
|Interpretable Multi-Objective Reinforcement Learning through Policy Orchestrations||Value Alignment||ICML (2019)|
|Data Pre-Processing for Discrimination Prevention: Information-Theoretic Optimization and Analysis||Fairness|
|Evolutionary Search for Adversarially Robust Neural Networks||Robustness||ICLR (2019)|
|Judging a Book by its Description : Analyzing Gender Stereotypes in the Man Bookers Prize Winning Fiction||Fairness|