Accelerating the UK’s efficient use of AI and data.

Early in 2020, the UK began to reinforce its stance on the creation and application of AI when the government released draft guidelines for an auditing framework for AI. Shortly after, guidelines for providing an explanation for AI-generated choices were released. The National Data Strategy was sparked by guidelines on AI and data protection that were released towards the end of the year.

In the context of the Integrated Review and the Integrate Review refresh, we got the Algorithmic Transparency Standard, the National AI action plan, and the Centre for Data, Ethics, and Innovation’s Effective AI Assurance Roadmap in 2021 and 2022. These are but a few instances of the policies and plans that suggest standards as a means of encouraging creativity and controlling artificial intelligence. We are not the only ones in this international competition to profit from new technology.

Therefore, as the spectrum of applications of AI continues to grow, standards and widely accepted technical specifications will become more and more important for furthering the governance and adoption of AI in the years to come. By improving the consistency and interoperability of AI technologies, they can help unlock the economic potential of AI by laying out best practices to guarantee that AI-related goods, processes, and services operate as intended.

The AI Standards Hub was created to make sure that business, government agencies, non-profit organizations, and scholarly academics have the resources and expertise necessary to contribute to the creation of these standards and utilize them wisely in order to build responsible and trustworthy AI.

The hub is an initiative that aims to increase the nation’s contribution to the development of international standards for artificial intelligence (AI) and addresses the difficulties faced by stakeholders in navigating the rapidly changing landscape of AI standardization. It is being led by the Alan Turing Institute in collaboration with the British Standards Institution (BSI), the National Physical Laboratory (NPL), and the Department for Science, Innovation and Technology (DSIT).

Nearly all of the “game-changing” goods, services, and solutions that will help us solve pressing global issues like public health, energy resilience, and climate change will depend on data that is trusted and used in sophisticated, artificial intelligence (AI)-driven digital technologies.

However, standards—specifically, the technical measuring standards used by NPL—are still developing. They stem from our work in metrology, or the science of measurement.  We need to be confident in order to unlock and extract the greatest value from AI and data.  faith in the data’s quality as well as in the reliability, robustness, safety, or riskiness of the AI tools, technologies, and systems’ outputs. Organizations like NPL have always placed a strong emphasis on quantification, quantitative inputs and outputs, and confidence levels—all of which are crucial in determining how reliable a technology is.  However, we must keep in mind that the use of AI varies greatly depending on the context.  How can I have faith in a system like this if my application is crucial to my business or safety? What matters to me now might not matter to you, or what matters to me now might not matter in five minutes.

We must change along with the world, if not ahead of it, as it changes all around us. While physical infrastructure technologies have been around for more than a century, the assets needed to build up a digital measuring infrastructure—especially one that is sophisticated and dynamic—cannot be expected to have the same longevity in a world where data is being used at the speed of artificial intelligence. Right now, there is a lively debate on what quantitative and qualitative criteria are required to ensure artificial intelligence (AI) and facilitate measurement when being evaluated for standard compliance. If you are interested in participating, please do so.