Ethical AI in operational practice

Global
- China
- Japan
- Germany
- North America
- South East Asia
- United Kingdom
- India
- Korea
- Hungary
- Arabic
- Brazil
- Global contacts

We develop and deliver complex power and propulsion solutions for safety-critical applications in the air, at sea and on land.

Our products and service packages enable our customers to connect people, societies, cultures and economies together.

Products & Services

We develop and deliver complex power and propulsion solutions.

For more than 100 years we have been at the forefront of innovation. Helping to power, protect and connect the modern world.

About

Create your future with us

Help us deliver better power for our changing world.

Doing more with less

We have a fundamental role in meeting the environmental and societal opportunities and challenges that the world faces.

Sustainability

Rolls-Royce Half Year results

Rolls-Royce announced Half Year results on Thursday 1 August 2024.

View Press release

News centre

Updates and news from around the Rolls-Royce businesses.

Home

Media

Our stories

Back

Friday 7 February 2020 10:38 AM

Ethical AI in operational practice

Our Industrial journey in Machine Learning, Artificial Intelligence and Advanced Analytics started when we put our Engine Health Monitoring systems in place in 1999.

In 2020 we are now monitoring 3000 engines whilst they are in flight at any one time, with data transmissions from the airborne aircraft via satellite or VHF. There are more than 11,000 engines hooked into our AI-driven predictive maintenance systems.

Digital Twins

We work with our world-class engineers to include the use of complex Digital Twins to assist in the development of some of the networks in the monitoring systems. They help by identifying insights the engineers will need to assess the health of the engines when they go into service. Over the last 3 years, we have developed our 2nd generation AI 2.0 which is so complex that we cannot programme it what to look for, – the AI system has to ‘self-learn’ across 26-dimensional multi-variate analysis. Our analytics run 24/7/365.

As you would hope, we’re monitoring the engine on the aircraft that’s flying passengers across the Atlantic every day - even if it is Christmas day! It takes just 3 minutes from acquiring data on the aircraft through transmission, AI analysis and providing any insights to the engineers who may then need to contact a maintenance crew to be ready on the ground to change an oil filter, for example, when the aircraft lands – and therefor ensuring that it can take-off on time.

Ethical requirements

It’s vitally important that we control bias and accuracy in these types of systems and processes, otherwise we could have increased maintenance costs or scrapping of some extremely expensive manufactured components.

To do this, we have developed an alternative approach to ‘explainability of the workings inside the black box’ and use high integrity parallel and independent checks that can be deployed into any of our AI based processes. For us, data bias and accuracy ethical requirements and fundamental business requirements – so developing leadership capabilities are win-win for us.

This complete ‘integrity system’ is being transferred across all our businesses, including into manufacturing and are enhanced to include the control of bias and accuracy in, for example, robot inspection of critical components or the development of autonomous robotic manufacturing processes.

AI Ethics principles

This control of bias and accuracy forms the heart of our AI Ethics programme which takes the world famous, ubiquitous AI Ethics principles (such as Asilomar, EU, Good Corporation etc) and creates a framework that ensures those principles are realised into any of our AI deployments across the whole of Rolls-Royce – bridging the gap from the ‘what to do’ questions in emerging data innovations techniques to the ‘how is it operationalised’ questions that can help to transform our business for the future.