A holistic ecosystem approach to AI

The need for an ecosystem perspective

As technical capability advanced and IT became ubiquitous, other considerations started to play a more prominent role such as the relative cost of a particular approach and its sustainability (environmental footprint and cost of operation). More recently, factors such as the security of the system, privacy and trustworthiness have become a commonplace topic of discussion and play an increasingly important role in defining requirements around emerging IT technologies.

The digital transformation of industries has further changed the landscape for IT standardization. For instance:

• Emerging non-technical requirements such as ethical and societal considerations and the ability to design trustworthy systems are key aspects
• Stakeholder diversity has increased considerably (eg. regulatory, social scientists, etc.)
• Early engagement by the various stakeholders has become the norm
• The use cases for IT have dramatically increased
• Understanding uses, proving business cases and developing standards are now concurrent
• The ‘data ecosystem’ is as important as hardware, software and operational technologies
• Maintainability of the solution and portability into other uses cases, within or across domains, has become top of mind for architects, technologists, etc. and increasingly part of the business planning for adoption of new IT
• Integration of a solution within larger technology deployments has become essential

A new approach to standardization

Not only has IT become ubiquitous, the way we have come to rely on technology and its presence in our daily lives has changed. Whether it’s at work or at home, IT is ever present in our daily lives. Most of us check our smartphones when we wake up and before going to sleep. We carry them around all day with us and have come to rely on their capabilities for everything from shopping to making a dinner reservation. At work, we are interacting with IT systems on a constant basis. Virtually every sector of the economy has been leveraging the innovations that IT is bringing and continues to digitalize.

While these innovations have increased efficiency throughout society, they also come with concerns that need to be addressed as highlighted above.

Standards play an important role in enabling technology adoption by eliminating barriers. Thus, it stands to reason that technology standardization can no longer only focus on the technical requirements and needs to concurrently consider implications and context of use. Furthermore, in the way that standardization provided an open forum to address technical challenges and came up with innovative solutions to address them, it can do the same for considerations such as ethical and societal concerns.  

SC 42 and the holistic AI ecosystem

What is needed is a framework that:  

• Takes into account the context of use of the technology by looking at both technology capability and non-technical requirements, such as business requirements, regulatory and policy requirements, application domain needs, and ethical and societal concerns
• Translates the above into technical requirements
• Builds foundational standards that communities can build upon, such as terminology, use cases, application guidance and reference architectures
• Links technology innovation communities such as proprietary implementations, research, SDOs and open source communities

The work of SC 42 addresses all these points. SC 42 is the focal point and proponent for the joint programme between ISO and IEC for IT Standardization, AI and Big Data work (ISO/IEC JTC 1/SC 42). In addition, it provides guidance to other JTC 1, IEC and ISO committees looking at applications of these technologies.

The ecosystem approach allows the assimilation of requirements from a variety of sources that include application domain, regulatory, societal, ethical and business considerations. While SC 42 produces ‘horizontal’ standards, it considers application domains and a broad collection of use cases to achieve this. Moreover, its deliverables provide the foundation for other AI and Big Data work to build on. The current programme of work is organized into a number of groups that include:

• foundational AI standardization
• big data
• AI trustworthiness
• use cases and applications
• computational approaches and computational characteristics of AI systems
• governance implications of AI
• AI management systems
• AI systems engineering

Topics like ethical and societal considerations are incorporated across the entire programme of work and topics like lifecycle are given consideration from a variety of perspectives in different groups.

The ecosystem approach extends to external collaboration. To-date, SC 42 has over 25 liaison partnerships and growing. The work also enables other SDOs, open source communities and R&D organizations to build on it, resulting in accelerated technology adoption.