AI (Artificial Intelligence) is top of mind for most organisations these days and is encompassing anything from analysing big data to machine learning as well as deep learning and autonomy.
AI is one of the disruptive technologies underpinning Industry 4.0 transformation. While it's hard to predict how deeply or quickly AI will go in the industries we serve, for now much of the interest in AI derives from the business imperatives of enabling organisations to derive efficiency and risk reduction from large quantities of data.
AI technologies can provide increased operational efficiency by using large amounts of historical performance data, modelling risk of complex systems using real-time information, build predictive failures models, optimise supply chain performance and analyse maintenance records to capture organisational knowledge.
There is lots of technology out there as investors and corporates pour billions into AI firms. For instance, the OECD reported that between 2011 and mid-2018 more than $50b have been invested in AI start-ups globally. And this is in addition to multi-billion dollars R&D investment by big tech players such as Google, Amazon, Microsoft, IBM.*1
But it is applying AI to business problems where adoption challenges arise. A Deloitte survey found that the number one obstacle to successful AI deployment is the difficulty of integrating "cognitive systems projects" into existing processes and systems.*2 Shortage of AI and AI-usage skills is another big problem hampering adoption, with the OECD estimating that jobs requiring AI skills have grown 4.5-fold since 2013 without a corresponding increase in available talent. In addition, AI presents high complexity in terms technology choices and requires extensive collaboration between domain experts and AI specialists, something that is never straightforward. Adopting AI in critical infrastructure industries such as energy and shipping, faces additional hurdles with requirements such as interpretability, verification and auditability.
This is one of the reasons we formed LRQA Aurora, LRQA's dedicated industry 4.0 innovation practice. Through our renowned subject matter experts in safety, risk and performance of critical assets, technology partnerships and our own growing AI capabilities, we help clients with using specific AI approaches to solve critical business challenges. We also set up the LRQA Safety Accelerator to help our clients pilot cutting-edge AI from innovative early stage technology firms to solve tough critical safety challenges.
Over the past year LRQA Aurora have engaged in a number of business-focussed AI projects with clients and technology partners.
In natural language processing, with machine-learning based techniques, we are doing work with clients to extract insight from vast quantities of historical data, such as safety reports and maintenance reports. One of the most notable examples is LRQA SafetyScanner and our work in maintenance analytics with shipping and oil and gas clients as part of our maintenance optimisation services. We are also working with partners with innovative approaches based on natural language processing to specific problems, such as London-based Ohalo for automatic anonymisation of documents to allow better and more secure analysis of historical records.
In computer vision, through our LRQA Safety Accelerator, we are working on several pilots addressing client challenges in partnerships with AI start-ups using the vision technologies. We are working in areas such as detection of human errors in maintaining electricity networks with a deep learning specialist NumberBoost and in ship engine rooms with Invision AI. We are also working on detection of eye movement patterns with Denver-based Senseye to determine worker's fatigue in safety-critical industries and with advanced human emotion sensing specialist Sensing Feeling, using strong privacy-by-design principles.
We are also active in building digital twins, which use detailed models of assets risk and/or behaviours created by combining traditional models with machine-learning from historical sensor and other contextual data. Recently, we have done work in failure analysis for wind farms, fuel consumption modelling in shipping and optimisation and root-cause analysis for chemical plants. We are also starting to use state-of-the-art Auto Machine Learning technologies (AutoML), giving the power of these advanced modelling technologies to engineers and practitioners. At the research level we collaborate with the Alan Turing Institute, the UK's national centres for AI and data science, which is working on a number of digital twin projects in applications such as gas turbines, 3D printed bridges and energy distribution.
One final aspect of AI that is key in safety critical industries is assurance. This is a hot topic for many industries, governments and regulators and it is one that will take some time to unfold. LRQA is active in this space, with breakthrough work in creating an assurance framework for digital twins and a collaboration with GE to apply it to a twin of gas turbines. LRQA is also at the forefront of technology research, with the £10m LRQA Foundation's Assurance of Autonomy International Programme in collaboration with York University.
The potential is high, and industry is at the cusp of adopting these technologies and we believe the two key ingredients is collaboration and innovation. Collaborate with organisations you trust that have the right expertise and capabilities. Start from clear articulation of a business challenge that is hard to solve with conventional methods, define what success looks like and learn through small but scalable proof-of-value projects.
*2 https://www2.deloitte.com/content/dam/Deloitte/us/Documents/deloitte-analytics/us-da-2017-deloitte-state-of-cognitive-survey.pdf figure 7
What we think
LRQA's experts regularly share their research and insights.