More than 50 AI-based tools are used in the Estonian public sector. One of them helps employees at the Estonian Unemployment Insurance Fund.
In Estonian mythology, kratt is a magical creature, made of hay and animated in exchange for three drops of blood to the devil, that does everything the master orders. In modern Estonia, kratt is a self-learning artificial intelligence system, doing everything that its masters order.
As the Ministry of Economic Affairs and Communications’ national artificial intelligence strategy is nearing its end, Estonia now has about 80 various AI applications either working or in development in the public sector. The decision support tool for the Estonian Unemployment Insurance Fund (EUIF) named OTT is one of the most outstanding of these “mythical” creatures.
AI calculates employment pathways
As people turn to the Unemployment Office for support, OTT is a tool used by civil servants to improve the understanding of the clients’ needs. It is an AI system that employs more than 100 000 client records to estimate the probabilities of different employment pathways, systematizes clients to provide support where it is most needed, and distributes the workload between civil servants.
“Our office has always been well stocked with data, and we’re pleased to see how well we have managed to put this in use,” says Mari Väli, development specialist at the Estonian Unemployment Insurance Fund. “Although OTT has been operational for only five months, so we haven’t done any official evaluation, our employees have accepted it well. After all, it helps them streamline their efforts a lot.”
This AI, created in cooperation with Citis, Nortal, and Resta, was first tested in some of the EUIF offices for six months. This testing period proved valuable as the developers decided to change the output of the model.
“First, we thought we would estimate the client’s probability of becoming a long-term unemployed, but this proved to be both difficult and also a bit useless. Now the model helps us to understand the probability of finding a job, and this is much more helpful,” Ms. Väli describes the development period.
Although fully operational, OTT is in constant development. Every three months, it is restrained manually by updating the actual outcomes of client pathways. But this is not the only development in progress by EUIF.
“EUIF is taking part in a program “Machine learning and AI-powered public service delivery” with one of the aims being the development of a tool to predict unemployment risk and thereby prevent becoming unemployed. A research consortium is looking to develop an AI that would help predict unemployment for currently employed people. By considering the developments in the job market and a combination of skills, this would help people prepare for the future and take better command of their careers,” Ms. Väli explains her plans.
Victims of our success
OTT is just one example of a more significant shift in Estonia’s public sector as it is turning even further towards cooperative open platform development. More than 25 private companies and 30 public sector organisations engaged in developing AI-based solutions for the public and private sectors. How is this kind of momentum and coordination achieved?
“I would say that in the best way, we are victims of our success,” says Ott Velsberg, Government Chief Data Officer at the Ministry of Economic Affairs and Communications. “Because we have an image to be one of the most digitalised and citizen-centric public sectors in the world, and we all want to live up to this image.”
This intense cooperation helps to achieve tangible savings. Mr. Velsberg brings an example that, for example, the #KrattAI PoC project, a project aimed at creating government virtual AI assistants, would save 1,8 million Euros of taxpayers’ money each year alone if implemented in just ten organisations.
But the savings effect is much larger. Because Estonia is a relatively small market, natural-language-based tools would be too costly for smaller private companies to develop. Because of the cooperation led by the Ministry of Education and Research and the Ministry of Economic Affairs and Communications, though, components designed with state finances are free to be recycled by entrepreneurs in the way they see fit.
Agile public sector development
As Estonians are getting increasingly used to AI-s in their everyday life, cooperative software development has left a mark on the public sector working methods in general. More civil servants are aligning with agile data-driven development methods.
“Instead of the old linear model, people are getting more used to the agile model, where the digital solution developed maybe something completely different from what we imagined in the beginning,” Mr. Velsberg explains. “I think this benefits our public sector greatly that we are now able to have a more problem-solving-oriented way of thinking. But this hasn’t come easily!”
Also, focusing on AI in the public sector also requires some minor changes in the legal system. In this regard, Estonia is waiting for the EU to provide its suggestions expected this spring. These could include topics such as accountability of AI and compatibility with human rights. Like in the olden days, although kratt brings excellent fortune, it has to be dealt with care to continue serving its masters faithfully.
Can’t travel but want to hear the e-Estonia story or want to implement e-services in your country or company? Take a look at our services and get in touch – we’ve got you covered!