The industrial AI revolution
Andreas Bartsch, Head of Innovation and Services at PBT Group
What began in the 1760s with the First Industrial Revolution that saw the move to new manufacturing processes in Britain has evolved to the Fourth Industrial Revolution in the 2000s where attention turned to the Internet and renewable energy. It is especially in the past two years where the acceleration of digital technology has been keenly felt.
Additionally, with the emergence of the Internet of Things (IoT) and artificial intelligence (AI), we are now seeing a convergence between the physical and virtual worlds. Fuelling this is access to real-time data that is resulting in business and technology leaders making better business decisions and consumers embracing more digital ways of doing things.
And as has been the case with every industrial revolution to date, this one will also herald a change in the skillsets needed to manage this at scale. Enter the industrial data scientist.
Data science done differently
In the past, an industry expert worked in tandem with IT and computer specialists to operationalise data in relevant ways. But this is evolving into a new profile that combines domain expertise (for instance as provided by an engineering qualification), tech-savviness, and a better understanding of the importance and relevance of data in the modern world. So, while the traditional approach can be considered more of an academic data scientist, this new one centres on the industrial data scientist.
However, companies need to find the right blend of experience and expertise to transition to this and adopt the likes of AI and machine learning within their industrial world. This new specialist is not about losing the person with years of industry experience but complementing them with the expertise to build and develop data-driven solutions that enable growth.
What cannot be ignored though is that this new breed of industrial data scientist will be fundamental to the digital transformation initiatives of an organisation. It is all about discovering business opportunities, achieving efficiencies and cost-savings, and maximising productivity.
Of course, this will not happen without overcoming challenges along the way. The industrial data scientist must be empowered by creating a supporting environment. This requires the organisation to turn the focus towards data readiness.
By achieving data readiness through proper data architecture, data modelling, data governance and data engineering, only then can the company provide the necessary data pipelines. After all, it does not help if it has the experience and the expertise if the data the company has is in shambles.
To overcome this, collaboration between IT and the data platform team is critical. This will help ensure the relevant IT and data processes can be leveraged to develop, test, and deploy models. Following from here, the AI models can then be productionised. These must be maintained to ensure data consistencies and to begin the process of automating as much of the data platform as possible.
Making sense of it all
This has already been embraced across various industry sectors. For instance, retailers are using data to track customer behaviour, manage stock levels, and manage campaigns. On the finance side, the likes of credit risk behaviour and developing relevant pricing strategies are significantly influenced by data-driven processes.
For their part, telcos are using this to drive cross-selling and reduce customer churn while in insurance it is about optimising call centre operations and reducing fraudulent claims. In logistics, data helps optimise routes and in healthcare the pandemic has highlighted the need to track and trace infections.
At a fundamental level, the likes of IoT, 5G, cloud, edge computing, and AI are enabling technologies for near real-time, event driven activities that feed machine learning models. This will result in the realisation of the industrial AI solution and the eme