By Nuala Polo
Today, we are pleased to announce the launch of DSIT’s Portfolio
of AI Assurance Techniques. The portfolio features a range of
case studies illustrating various AI assurance techniques being
used in the real-world to support the development of trustworthy
AI. You can read the case studies here.
How does AI Assurance support AI governance?
In the recent AI Regulation White
Paper the UK government describes its pro-innovation,
proportionate, and adaptable approach to AI regulation to support
responsible innovation across sectors. The White Paper outlines
five cross-cutting principles for AI regulation: Safety, security
and robustness; appropriate transparency and explainability;
fairness; accountability and governance; and contestability and
redress.
The regulatory principles outline what outcomes AI
systems need to fulfil, but how can we test whether a
system actually achieves these results in practice? This is where
tools for trustworthy AI come into play. Tools for trustworthy
AI, like assurance techniques and standards, can help to measure,
evaluate and communicate whether an AI system is trustworthy and
aligned with the UK’s principles for AI regulation and wider
governance. These tools provide the basis for consumers to trust
the products they buy are safe, and for industry to confidently
invest in new products and services. These services could also
build a successful market in its own right. Based on the success
of the UK’s cybersecurity assurance industry, an AI assurance
ecosystem could be worth nearly £4 billion to the UK economy.
The CDEI has conducted extensive research to investigate current
attitudes towards, and uptake of tools for trustworthy AI.
We published our findings in the Industry Temperature
Check report, which identified major barriers that are
impeding or preventing industry use of assurance techniques and
standards. One of the key barriers identified in this research
was a significant lack of knowledge and skills
regarding AI assurance. Research participants reported that even
if they want to assure their systems, they often don’t know what
assurance techniques exist, or how these might be applied in
practice across different contexts and use cases.
Portfolio of AI Assurance Techniques
To address this lack of knowledge and help industry to navigate
the AI assurance landscape, we are pleased to announce the launch
of the DSIT Portfolio of AI assurance techniques. The portfolio
has been developed by DSIT’s Centre for Data Ethics and
Innovation, initially in collaboration with Tech UK. The portfolio
is useful for anybody involved in designing, developing,
deploying or procuring AI-enabled systems, and showcases examples
of various AI assurance techniques being used in the real-world
to support the development of trustworthy AI.
The portfolio includes a variety of case studies from across
multiple sectors and features a range of technical, procedural
and educational approaches, illustrating how a combination of
different techniques can be used to promote responsible AI.
We have mapped these techniques to the principles set out
in the UK government’s white paper on AI regulation, to
illustrate the potential role of these techniques in supporting
organisational AI governance.
Please note the inclusion of a case study in the portfolio does
not represent a government endorsement of the technique or the
organisation, rather we are aiming to demonstrate the range of
possible options that currently exist.
Read the case studies here.
Next steps: future submissions We
will be developing the portfolio over time, and publishing future
iterations with new case studies. While the first iteration was
delivered with techUK, we invite future submissions from
organisations across all sectors, to showcase a broad range of AI
assurance techniques in practice. If you would like to submit
case studies to the portfolio or would like further information
please get in touch at ai.assurance@cdei.gov.uk.