Intelligent robots and self-learning computers will have a huge impact on German industry in the coming years: Artificial intelligence (AI) could increase the gross domestic product by ten billion euros a year by 2030, according to McKinsey . The management consultancy sees the greatest opportunities in manufacturing and business processes. Carefully designed collaboration between robots and human workers could increase productivity by 20 percent, McKinsey posits, for instance – as long as companies integrate artificial intelligence into their business model consistently and as soon as possible. These five rules from McKinsey help to build an AI-centered environment.
1. Business case: Argue for AI opportunities
Investing quickly in a trend is not advisable when it comes to artificial intelligence. First you have to understand what AI can do and how it brings added value to your company. Key parameters here are the technical feasibility, the available budget and the optimization potential within the company. Defining a business case is no easy task when it comes to artificial intelligence, McKinsey warns. Information about AI remains incomplete and doubts can be widespread among employees.
2. Competencies: Connect internal and external skill bases
AI experts are still far too scarce: In Germany there is a gap of some 5,000 qualified workers , and according to McKinsey only 0.1 percent of workers in the US are trained data scientists. This means companies have to build up their in-house expertise as well as connect it with resources from third-party providers. Both developers and intermediaries are needed. The latter form the interface within the company between developers and management. This is the approach needed to create an AI-centered company.
3. Raw material: Store granular data
Data forms the foundation for every artificial intelligence. Companies must therefore store as much granular (i.e. sufficiently detailed) data as possible and make this information available in a relational, table-based database. Such data might be images or language-based, as well as data from sensors and machines – a heterogeneous ensemble that needs to be adapted and aligned.
4. Expertise: Combine existing knowledge with AI
AI will not lead to successful outcomes without careful attention to context. A company that wants to introduce AI knows their own sector, their technological environment and the interplay of systems, technologies and employees best. Outsourcing such assessments is not advisable. McKinsey recommends codifying your company's own knowledge, i.e. recording and storing it, and integrating it into the AI algorithm, to best support machines' self-learning.
5. Practice: Test as an agile process
Introducing AI should be an agile process. Simulations and tests in subsections of the company serve to continuously optimize results. AI gains more knowledge (learning), learns to understand it (understanding) and later, applies this knowledge (solving). Errors form the basis for future improvement. The scope of the project determines how long the pilot process lasts. An AI task force might monitor the processes across functions and make decisions about next steps.
However, there is one other thing a company must consider before integrating artificial intelligence into its processes: cultural factors. Many employees may have concerns about security and data protection, worry about jobs being lost, or believe that machine intelligence can never equal human discernment. Jonas Albertson , Managing Director with the Swedish industrial firm Atlas Copco, says: "The greatest challenge is not the technology. What really matters is change management for the people involved." AI requires a new way of thinking. Only a change of hearts and minds will bring about what McKinsey has predicted for 2030: a gross domestic product with growth of 10 billion euros a year thanks to AI.