Artificial intelligence (AI) has long been a technology with much promise. It was the onset of the pandemic, however, when it really came into its own, proving its value to workforces globally as organisations were forced to adopt the technology to support their move to hybrid working.
Having realised its true potential, organisations now need to begin to invest in AI and machine learning in order to support more “traditional” elements of business, to accelerate processes and gain market share over rivals.
Ian Hall, Head of AI & Analytics at Capgemini UK shares his top AI predictions for 2022 including the trends, technologies and key issues that we could see unfold in the year ahead.
Improving patient care via privacy preserving analytics
Hospitals have rapidly progressed the use of cutting-edge technologies in healthcare, while maintaining patient data privacy by applying federated learning (a type of machine learning) with hardware-enhanced security.
The concepts enabled each group to share x-ray data with one another in a secure anonymised way, enabling the development of more accurate automated medical diagnosis models using much larger sample sizes.
Creation of cross-organisational global datasets will deliver significantly improved results – in one case 18% better than the best locally developed model.
Given the obvious benefits of such collaborations, and the evolution of the technologies that enable such research, we envision a usage of such concepts, for example Intel Xeon Scalable processors and Nvidia graphics hardware, in 2022 across both the public and private sectors.
Data-Driven Digital Twins accelerate cars, planes, and space rockets
When manufacturers prototype in the physical world, it involves significant time and money. However, thanks to digital twin technology – the virtual representation of a physical product – it’s now possible to make multiple adjustments from a computer and uncover the most optimal design as well as run multiple simulations on how it would respond to varying conditions.
The drivers for digital twins are expanding from just basic modelling of devices, to being able to model the capabilities, specific attributes, and properties of these more accurately to identically replicate the physical device. This will become a stronger theme throughout 2022, with the use of digital twins focusing more on modelling the relationship between multiple devices, and even the factories that produce them.
These concepts are particularly relevant given the huge investments being planned in the automotive and (aero)space industries, and the accelerating shift towards high powered engines powered by electricity. If our predictions are correct, emissions and impacts could be understood before manufacturing even begins.
Increasing need for fair, Explainable AI, or ‘XAI’
As AI becomes more ubiquitous, there is simply a greater need for explainability, in other words, a set of processes and methods that allow human users to comprehend and trust the results and output created by machine learning algorithms. By keeping the human in the loop, models can be used as tools to assist decisions and thus means explainable models have a clearer path to production as a result of increased ease of understanding.
The wider public have not only always struggled to understand AI, but also questioned the ethics of computers making decisions by themselves, and this has given rise to EAI, pushing it to the top of the agenda. As a result, we’re likely to see more models being designed in 2022 with inherent explainability in mind, considering functions such as algorithmic transparency and decomposability.
More smart factories interoperating at scale
The concepts enabling smarter factories, for example using sensors and cameras to identify and predict problems within a production line, are by no means new. But deploying these concepts at scale has so far been constrained. In 2022, however, we’ll see more mature organisations upping their game.
Take, for instance, a fully connected and integrated plant. With all systems sharing data across multiple sites in real-time, more and more interdependencies can be monitored and autonomously managed. This means that if an issue is identified with a raw material from a certain batch, then all the processes that depend upon it – regardless of the location of the site – can be automatically halted thus minimising impact.
Sustainable AI and supply chains
Soon, everything will be seen through a sustainability lens, and AI is no exception. One way we might see AI helping organisations to reduce their carbon footprint in the year ahead is through route optimisation and fleet management throughout the supply chain. This will be driven by companies feeling a greater need to take into account – not only their direct impact – but also the indirect emission contributions from their extended supply chain of partners and vendors.
Therefore, in 2022, we’ll see further collaboration across this myriad of stakeholders in order to make a greater contribution towards decarbonisation via optimised logistics flows. AI will be used to interpret data from electricity or natural gas invoices, and we’ll see the growth of central sustainability tracking models, which will form the basis for footprint reduction projects.
For more information download the Capgemini Research Institute reports here,
File size: 3.42 MBFile type: PDF
File size: 2.68 MBFile type: PDF
Head of AI & Analytics at Capgemini UK
Ian has worked for Capgemini for almost 11 years. He is the Head of the AI & Analytics practice and a Strategic Architect within the Advisory team.
“Together we help businesses to understand and realise the value of placing insights and data at the centre of their thinking.”