Controlling the implementation of intelligent systems with needs
The possibilities of data analysis, artificial intelligence and machine learning are almost endless. But grandiose visions are of no value unless they come to fruition. Always start with the need; see potential in the problems you tackle on a daily basis. If you ask a concrete question, you increase the chance of receiving a clear answer.
How can large quantities of data be of benefit to the company? Machines and systems gather large quantities of data on a daily basis; data which could potentially improve profitability, quality, maintenance work and machine usage. The big question is how this will work.
Step one is to identify the need and how the solution to the problem can generate business value. Start where you have existing data and a real need. The difficult thing is finding the right boundaries and not trying to solve all problems at once. Step two is to assemble a team with relevant expertise. It should be a group which as a whole understands the entire process and can apply the necessary methods to rectify the problem. They should be people who understand machines’ functions and systems, who can extract, analyse and visualise relevant data. In addition to this, they should have expertise which can link the work to business value and elucidate the benefits through implementation. The third step is to run a pilot – start small and build up knowledge and experience for upcoming projects. It is also important to consider whether the problem can be solved with conventional methods, so as to avoid unnecessarily complex solutions.
What can we do then, to enable companies to get moving? A good start is to raise the question in different contexts and with a target group-adapted angle (it is difficult to reach the right people in the companies). Independent organisations such as Automation Region and RISE have an important role to fill. We can disseminate good examples and create opportunities to learn from organisations which have already successfully implemented artificial intelligence and machine learning. The dream would be a roadshow, where we can enter organisations at a high level and focus on business value.
The organisations that are successful with their projects have the ability to visualise profits throughout the chain; everything from financial aspects – how can we recoup these costs? – to machine operators’ daily lives, where the actual work can be facilitated and optimised. One final recommendation is therefore to build up a broad foundation and understanding within the organisation for the values and advantages which a new concept can provide. And to always let the need be the guiding factor.
Senior researcher at RISE SICS Västerås