AI and human intelligence together make the best decision-making systems

The development of artificial intelligence, AI, has transformed commerce in several ways. In retail business industry, great resources are put into returns management. There is increasing focus on mobile applications and advanced data analytics to understand customer behaviour. The huge availability of data has led many companies to increasingly take an interest in advanced data analysis and machine learning that can automatically generate predictions with the aim of supporting decision-making.

“AI has impacted retail business industry’s opportunities to grow and be competitive. But several companies have realised that AI-based predictions are not always useful; they have been relied upon too much,” said Stefan Cronholm, Professor of Information Technology.

Several research papers also highlight the risk there is that employees' knowledge, skills, and experiences can be overlooked. It may have to do with cognitive abilities related to logical reasoning, abstraction skills, or quiet knowledge that is not represented in databases, or about contexts, intuition, professional experiences, and culture. In those respects, people's abilities are many times superior to that of machines.

Have developed a prototype for a digital decision support system

The aim of the project has been to develop design knowledge supporting the development of decision support systems in which AI and human intelligence have been integrated together as a hybrid. A prototype for a digital decision support system and design principles have now been developed.

“We have limited ourselves to examining how human intelligence and AI can be integrated when dealing with retail business industry returns. We call the prototype a hybrid system and its purpose is to support the improvement of returns management, for example in analyses of returns that are over-represented, analyses of transport in relation to carbon emissions, and analyses of more return-prone community groups,” explained Stefan Cronholm.

Both the decision support system and the design principles have been developed based on theoretical insights and empirical evaluations.

After testing the prototype in a number of retail business industry companies, it was found that it contributes to decision-making that is better than what people or AI alone can produce.

“The design principles of the project are based on a case of returns management within a smaller number of companies, but there are no obstacles to their application in other contexts,” concluded Stefan Cronholm.

Read more about the project

Digital Hybrid Systems for Innovation 

The project was implemented in the period 2019—2021 and financed by the Swedish Retail and Wholesale Council.