#002 - Authorities & Property information

April 20 2020/45 min read

We are joined by Mats Snäll. He is the Chief Innovation Officer at Lantmäteriet (the Swedish mapping, cadastral and land registration authority, is a public authority that belongs to the Ministry of Finance). What are the key opportunities for AI in society from a governmental perspective?

(The discussion is in Swedish)

Key takeaways from this episode

The mission of the authority is to make sure ownership of land and property functions in society, that the information is available and workable, and that it’s possible to create new land and property.
As Chief Innovation Officer, Mats is responsible to take things from ideas to reality.

Lantmäteriet has 2300 employees in 50 locations. Founded in 1628 and has been gathering data since the 18th century. Sit on enormous amounts of data; text, maps and more. A true data factory. It’s all about quality in data, and being able to work with it.

The view on AI in the public sector and authorities in Sweden
  • Awareness and understanding of AI in Sweden is still low. It’s a new area and unexplored. But things are happening and interest is increasing. Also from organisations that usually are not very technically savvy.
  • “Can we use AI to solve this” is a question that is heard more and more.

The key to innovation lies in the interest for change and learning at an individual level
  • The creation an AI meetup forum at Lantmäteriet, to discuss possibilities and opportunities. Today 30-40 people.
  • The interest to learn and work with new ways of doing things can come from anywhere. It does not only come from developers and technologists.
  • Creativity and interest for change must be allowed to prosper and grow in an organisation.

Working together is important, and looking at the entire picture.
  • Fun projects are often the key to innovation and change.
The AI model itself is a small part of the entire solution
  • Many competences are needed and it is a team effort.
  • The interest and passion to solve problems is more important than AI competence itself.

Collaboration is important
  • It’s still hard to find the AI-modelling competence. There are so few of them.
  • One way to tackle this is a culture of collaboration, to work with other organisations, public, private and startups. To move forward and solve problems together.
The willingness to change is key

Lantmäteriet challenges around turnaround times
  • A lot of properties have very old data connected to them. Can be typed or handwritten, or old types of maps.
  • A lot of the knowledge that enables us to work with this kind of data sits at an individual experience and expertise level. Need to be able to handle when people retire and still work with historical data.
  • The goal is to make all our data accessible and easy to work with

Explainable is important. At least today. Maybe not in the future.
  • Since we work with recommendations and decisions that affect society trust is important. There is a risk of loss of trust if we cannot explain how decisions were made.
  • All authorities rest upon the foundation of trust.
  • Lantmäteriet has been given a government mission to look at trust mechanisms in AI.

Trust is also connected to knowledge and understanding.
  • At the moment the understanding of how AI functions is low. At all levels. Once the understanding increases, perhaps the need for explainability will lessen.
Do we need an AI “ombudsman”?
  • Who’s responsible if an AI model reaches the “wrong” decision?
  • When are we ready for AI-models to make unassisted decisions?

AI can make better decisions that humans.
  • Today AI is great decision support for humans. But we cannot stop there. AI can find patterns that humans cannot. AI will be able to drive better than humans.