The Tortoise AI Responsibility100
A first step towards assessing the FTSE 100 companies on the responsible use of artificial intelligence.
In partnership with
Artificial intelligence will soon be fundamental to the operations of almost every business, a shift accelerated by generative AI.
What we know: the use of AI offers huge opportunities for value creation.
- Around 68 per cent of large companies in the UK adopted at least one AI technology in 2022.
- The UK AI market, currently valued at £1.36 trillion, is estimated to be worth £2.4 trillion by 2027.
What we don’t know: how to use AI responsibly.
At Tortoise, we’ve long used data research to hold companies accountable. We gather social, environmental and ethical indicators to show what it means to be a responsible business. We think that how companies adopt AI is the next critical area to measure.
The risks of AI – including algorithmic bias, privacy violation and the spread of misinformation – are real and immediate. But the principles of how companies can use AI responsibly are unclear.
Many companies are still considering how AI could benefit their business – while the potential of the latest foundational models is not fully understood even by researchers.
That’s why we think it’s worthwhile, at this early stage, to propose a set of principles on responsible AI, informed by experts in the sector, and a framework to assess FTSE 100 companies on their use of the technology.
“Generative AI is a very exciting opportunity for every company. We see it as a step change to our business both for internal operations and externally working with customers.” – Gillian Tomlinson, Chief Data Officer, The Weir Group
What is responsible AI?
There is no universal consensus on what constitutes responsible artificial intelligence. But drawing from corporate and intergovernmental guiding frameworks, expert conversations and business insights, we suggest defining it based on the following core pillars.
How can we measure it?
Within each pillar, we have identified a set of potential indicators that could help us measure the degree to which businesses are adopting artificial intelligence responsibly.
Each indicator would receive a score between ‘latent’, ‘developing’ and ‘leading’. This approach will enable us to monitor the various dimensions of responsible AI adoption, distinguishing between areas where businesses are just beginning, areas showing progress, and areas where companies have achieved significant advancement.
The data points underlying the indicators would be drawn from company reporting – including annual reports and press releases – as well as independent third party sources.
Tortoise assesses responsible business performance by grouping measurements into ‘Talk’ and ‘Walk’ indicators, to differentiate between principle and practice. The same framework could be applied to the assessment of responsible AI.
Talk indicates the level of public commitment to AI principles, targets and memberships of certain organisations.
Walk refers to the measures taken to implement AI objectives through practical operations.
We hope this framework will allow us to start measuring progress, while also prompting further discussion with industry leaders on best practices.