21 July 2020

New checklist helps the public sector use artificial intelligence responsibly

artificial intelligence

A new specification from Danish Standard can make it easier to work with artificial intelligence for public authorities, while also ensuring that the development takes place in an ethically responsible manner. Professor of software development from the Department of Computer Science Thomas Hildebrandt has been the lead author for the new specification.

Checklist

Politicians and technology providers are pushing for the development of smart optimization solutions with artificial intelligence that can save the public both time and money. 

But until now, there have not been any tangible guidelines to help ensure that the hyped technology is developed responsibly. 

Thomas Hildebrandt, professor at the Department of Computer Science at the University of Copenhagen, has written the new specification about artificial intelligence's use in public casework, which for the first time gives developers and authorities a toolbox to do exactly that. 

- Artificial intelligence can create mistrust of society's institutions and potentially discriminate in relation to, for example, gender and race. That's why it's important to think carefully when considering using AI technology. With this specification, we have put a tangible and systematic tool forward to help with this, says Thomas Hildebrandt.

Saves money and hinders mistrust

The specification, which is the first step on the way to an actual standard, consists of a number of questions that function as a checklist. These are not actual requirements or guidelines. 

The specification systematically lists several questions for each part of the process in the development of a new artificial intelligence and who the questions should be addressed. 

The answers are designed to ensure a more complete final product that works for the good of everyone. 

- We have seen too many examples of solutions being scrapped because doubts were raised about the legality or applicability of the solution. This can be countered by ensuring involvement at a very early stage in the process, which the guidelines propose. This can save a lot of mistrust towards the public sector and, of course, also money, says Thomas Hildebrandt. 

It is Danish Standard that together with a large number of experts are behind the specification. They have also developed a specification with guidelines to best ensure transparency in artificial intelligence. 

Thomas Hildebrandt will present the principles of specification in Danish it's virtual OffDig Week 2020 on August 27.