Game changing AI admissions screening software launched

Fujitsu has launched the early childhood education and care (ECEC) sector’s first ever artificial intelligence (AI) admissions screening software in Tokyo overnight. The system has the capacity to match thousands of children to vacancies in ECEC services in seconds.

The AI-powered software was developed to support municipal employees to perform tasks currently undertaken by employees, such as matching children to vacancies. The system, known as ‘Fujitsu Human Centric AI Zinrai’ matches children to vacancies based on the needs of the child and detailed criteria which is determined by local governments.

The current process of enrolling children in Tokyo-based centres is similar to that in Australia – children are enrolled into centres on a trial and error basis, which focuses on availability, applicant priority, parent choice, and requests for siblings to be in the same service. This process is time consuming, and becomes complex when there are additional variables, such as ratios for both children and staff, or a lack of availability to accommodate siblings.

The ZInrai AI technology aims to overcome these challenges by incorporating game theory – a mathematical approach that rationally resolves the relationships between people with conflicting interests. It matches thousands of children to ECEC spaces in seconds by using priority criteria, to ensure all those in the applicant pool have as many of their requests met as possible.

The technology is currently being trialled by more than thirty local governments in Japan, with plans for expansion.

Background

The Japanese birth rate has been declining over the years, and initiatives such as Act on Child and Childcare Support have been put in place to address this, Fujitsu outlined, in justifying the development of the technology.

Many issues still remain, however, in the childcare situation in many regions, such as the number of children waiting for openings at ECEC services. One such issue is increasingly complex service admissions screenings due to the need to maintain equitable access for all those seeking ECEC spaces for their child or children. The result is that significant manpower and time are required to match children to limited openings, which takes various circumstances of applicants into account.

Some parents prioritise siblings going to the same ECEC center, while others allow siblings to be in separate ECEC centers as long as all the children are admitted somewhere. There have also been many cases, depending on the local government, where siblings ended up in different ECEC services even after repeated considerations. Solving these problems and handling the admissions assignment process quickly, carefully, and appropriately has become an urgent task of governments that have made supporting working women as an important policy.

How the matching process works

1. Extracting necessary information from child-rearing support system for admissions screenings

Information required for admissions screenings, including information for each child such as; current enrolment, priority level, desired location, siblings, age, and information about the number of available places in each service is taken from the child-rearing support system (Japanese analysis tool) and placed into the AI admissions screening system. Fujitsu’s solution or a third-party product can be selected as the child-rearing support system serving as the information source.

2. Conducting admissions screening using AI

The software matches children to services in a few seconds based on various applicant requests – including sibling entrance requirements and preferences for specific ECEC locations, as well as applicant priority criteria based on ECEC centre use adjustment indexes determined by each local government and detailed standards. This way, all applicants get as many of their requests met as possible according to applicant priority criteria, and the assignment process for several thousand children, which currently takes about 1,000 hours for government-specified core cities to perform, is completed in seconds. By notifying parents and guardians of the admissions result more quickly, residents’ services are also improved. As assignment results are generated in both digital and document formats, the results input process can also become more efficient.

3. Visualising results, supporting explanations

The software provides functionality to support staff in explaining the reasons for allocations, such as when a parent wants to know why their child was not admitted to the preferred service. Staff can explain the results based on the number of openings available, the priority of the request, and sibling entrance requirements, so they can smoothly respond to inquiries from and service-counter consultations with people seeking care, providing accountability.

Diagram of Daycare Center AI Admissions Screening