THANK YOU FOR SUBSCRIBING
Our safety mindset has broken through AI trust barriers
Caroline Gorski, Group Director-R Data Labs, Rolls-Royce
Rolls-Royce trusts its algorithms to make critical decisions because we have used our approach to product safety to develop a way to control bias and turn artificial intelligence ethics into a procedure.
We’ve called it The Aletheia Framework™ and, after thorough peer review, we believe it is world-leading and could transform levels of trust in AI, allowing AI’s full potential to be unlocked for the health, wealth, and growth of the world.
In fact, we think it’s so important, we’ve given it away for free.
The Aletheia Framework™, named after the Greek goddess of trust and transparency, is a 32-step toolkit to guide the practical application of ethical AI projects. It’s free to download from the Rolls-Royce website and any organisation that follows it can be assured its AI is accurate, well-managed, and has a positive impact.
It consists of three principal areas of focus: Social Impact, Accuracy/Trust, and Governance. Each of these areas has an associated set of contexts and ethics, which‚ in turn, connect directly to a series of realisation principles. It is these 32 realisation principles or challenges, that must be satisfied for the organisation to deem its AI projects trustworthy and ethical.
Presented in the form of a practical checklist, these principles form metrics for executives and boards to provide evidence that these criteria have been rigorously considered.
Once the AI has been applied, the framework includes a five-step continuous automated checking process, which tracks the decisions the AI is making to detect bias or malfunction and allow human intervention to control and correct it.
For Rolls-Royce, The Aletheia Framework™ sits in our management system, as a constituent part of what we call a Digital Passport, which all data projects have to have to proceed compliantly.
During our peer review, which included subject matter experts across big tech, academia, pharmaceuticals, automotive, and governments, it was clear that this was breakthrough work with impacts beyond our own area of industrial use and into consumer applications, healthcare, recruitment, security – basically any use of artificial intelligence.
Why did Rolls-Royce create the Aletheia Framework™?
Rolls-Royce is a global industrial technology leader and has used artificial intelligence for decades across multiple processes. In jet engines, AI is used to analyse more than 70 trillion data points across 26 dimensions to maximise the amount of time they fly and to improve their sustainability.
It is also now being developed for automated inspections of critical components and it was this application that prompted Rolls-Royce to extend its code of ethics to its use of AI. It needed clear answers to key questions to ensure that AI would be used appropriately, that it is trustworthy and – above all - safe. It’s that deep safety culture at the heart of Rolls-Royce’s business that has created the mindset to apply safety principles to data science.
Open-source, free to all
Rolls-Royce views the Aletheia FrameworkTM as a new standard of ethics and trustworthiness in AI that it believes is too important to keep to itself, which is why we have made it freely available to everyone.
We also believe AI needs to be regulated and that this work can help regulation to be effective, independent, and pragmatic. Our business is heavily regulated and we see how key it is to public trust. People fly in planes powered by our engines at 500mph at 35,000 feet because they trust us and the safety system within which we operate, including independent oversight. The widespread trust in AI will not happen without regulation and The Aletheia Framework™ offers a way towards practical regulation.
Currently, the trajectory for regulation is to crack open the black box and examine algorithms themselves, which are of course constantly changing. The Aletheia Framework™ provides a checklist of measures to ensure the initial design of the AI application is ethical, and that its resulting outputs remain unbiased and true to the intended design. When fully implemented, businesses can demonstrate their AI projects are fair, trustworthy, and ethical to regulators, and importantly the public. It is not necessary, or desirable to open the black box.
Artificial intelligence is a complex subject and The Aletheia Framework™ hasn’t solved all of its challenges. But it can help reassure organisations, people and communities, and regulators that AI is being used ethically and makes trustworthy decisions. Not only could greater use of ethical artificial intelligence help the world recover from Covid-19, but it will also help prepare the foundation for the opportunities of the digital future. However, the potential of artificial intelligence can only be realised with public confidence and trust.
As countries become more digitally enabled, people have a right to expect that AI is used ethically and that it is a trustworthy partner for people and society. The Aletheia FrameworkTM provides the foundations for that to happen, so business leaders, academics, technologists, and even philosophers can now move from simply talking about the potential of AI, to unlocking its benefits for the health, wealth, and growth of the world.