As artificial intelligence (AI) is added to an increasing number of business processes, the need to demonstrate that AI systems make decisions in a fair manner, are not vulnerable to tampering, and can be explained to different stakeholders is becoming more important. In a nutshell, there is a pressing need to bring trust into AI systems. Open source is a key enabler of building trust because code and techniques are visible to everyone.
To continue its commitment to make AI more trustworthy, IBM announced today that it is teaming with the Linux Foundation AI (LF AI) to drive the development of open source trusted AI workflows. Building on previously released open source AI projects from IBM Research, IBM VP Open Technologies Todd Moore stated in his keynote at Linux Foundation Open Source Summit:
“The basic tools needed to start building fairness, robustness, and explainability into enterprise AI workflows are now available to developers and data scientists, so it’s time for IBM to team with Linux Foundation AI and other partners to work together on this important space, and build one of the foundations for trusted AI.”
LF AI has a vendor-neutral environment with open governance to support collaboration and acceleration of open source technical projects. LF AI supports the open development of projects in a diverse and thriving community.
Fairness, robustness, and explainability in AI are some of the key cornerstones of trustworthy AI. Through its open source projects, IBM and IBM Research bring together the developer, data science and research community to accelerate the pace of innovation and instrument trust into AI.
Fairness, robustness, and explainability are key attributes of trusted AI
IBM Research open sourced three state-of-the-art, trusted AI toolkits in the past two years — AI Fairness 360, Adversarial Robustness 360, and AI Explainability 360. Each toolkit has a unique purpose and can easily be extended to include newly developed technology. Joining Linux Foundation AI (LF AI) will allow IBM to work closely with other members of LF AI and other contributors to continue growing and refining the toolkits.
- The AI Fairness 360 Toolkit is an open source toolkit that can help detect and mitigate unwanted bias in machine learning models and datasets. With the toolkit, developers and data scientists can easily check and mitigate for biases at multiple points along their machine learning lifecycle, using the appropriate fairness metrics for their circumstances.
- The Adversarial Robustness 360 Toolbox, is an open source library that supports both researchers and developers in detecting adversarial attacks on deep neural networks, defending them against such attacks, and crafting simulated attacks to make AI systems more secure.
- The AI Explainability 360 Toolkit is a comprehensive open source toolkit of diverse algorithms, code, guides, tutorials, and demos that support the interpretability and explainability of machine learning models.
As the number of techniques grows for implementing fairness, robustness, and explainability in AI systems, IBM’s deep enterprise AI expertise can help developers and data scientists choose among these techniques. Open source technology serves as a common vocabulary for these discussions.
Linux Foundation AI’s work to democratize building and deploying AI systems
LF AI is an umbrella foundation within The Linux Foundation that supports open source innovation in artificial intelligence, machine learning, and deep learning. A sustainable open source AI ecosystem allows developers, data scientists, and vendors to quickly create AI products and services using open source technology building blocks. With IBM joining LF AI, our goal is to make trusted AI techniques accessible to all LF AI projects, including Acumos.
LF AI has established committees to advance ethical and trustworthy AI as well as machine learning workflows. IBM will work with LF AI to craft reference architectures and best practices for using these open source tools in production and business scenarios, making them consumable in machine learning (ML) workflows. We will jointly lead the work with LF AI to define technical processes and guidelines on how to build trusted AI systems.
In the last six months, the overall open source AI ecosystem described in the LF AI landscape has grown from 80 to more than 170 projects, with a combined 350 million lines of code from more than 80 different organizations around the world. This level and pace of open source development is similar to the earliest days of Linux, blockchain, cloud, and containers development.
Synergies with other AI-infused parts of Linux Foundation are possible
By joining LF AI, we see synergies with several other initiatives where IBM is a member, including:
- LF Edge: Trusted AI is needed in edge devices, from driverless vehicles to smartphones to automated factories and farms. Industry fragmentation is a major challenge for edge development. LF Edge brings together projects across Internet of Things (IoT), cloud, and enterprises to increase unity across platforms, communities, and ecosystems. LF Edge fosters collaborations with end users, vendors, and developers to transform all aspects of edge technology and speed open source development.
- LF ODPi: Data is at the heart of building open source trusted AI systems — and data governance is especially needed. ODPi provides one of the only vendor neutral, open source standards to enable best practices for data governance, connectivity, business intelligence, analytics, and AI systems.
- LF Energy: The energy industry needs open source trusted AI across a wide range of business processes, from predicting demand to predictive maintenance of equipment and more. LF Energy provides a vendor-neutral, collaborative environment to “enable the electrification of everything to scale,” thereby transforming the world’s relationship to the important resource of energy.
- LF ONAP: Trusted AI embedded in the network is a priority for the communications industry. The Open Network Automation Platform (ONAP) is ready to be infused with AI to enhance real-time, policy-driven orchestration and automation of physical and virtual network functions. Communication industry providers and developers can use open source to rapidly automate new services and support complete lifecycle management.
- LF CNCF: Enterprise business processes will access AI capabilities through the cloud, which is why building trust in AI is so important. The Cloud Native Computing Foundation (CNCF) hosts critical components of the global technology infrastructure.
These projects are a sample of the related initiatives and sub-foundations under the Linux Foundation’s overall umbrella. As the era of trusted AI systems unfolds and the capabilities of AI systems increase, more synergies will develop.
Join us in developing open-source trusted AI workflows
IBM is joining the LF AI at this time to work more closely with existing LF AI members and engage with potential enterprise partners across industries who require trusted AI in their business processes. Joining LF AI helps us bring trusted AI to numerous industries, so that AI is used with fairness, robustness, and explainablity.
IBM has a long history of working in open source communities. Today, along with our new LF AI partners, IBM invites others to join in contributing code and best practices for building open source trusted AI.