According to Jasmin Grünbaum, the development of artificial intelligence will also influence audits by the National Audit Office in future. A programme of the National Audit Office for young professionals, which was started in spring 2018, provides professionals at the start of their careers with a comprehensive overview of the work done by the NAOF, while providing the NAOF with fresh insights and new ideas. The young professionals blog about their experiences at VTV.fi.
In September, I was sitting at a seminar where the attendees where asked how much they believed algorithms to influence the decisions they make in their everyday lives. I replied that I believe algorithms influence my decisions a great deal. I wasn’t able to specify in detail in which way, however. In addition to automating large portions of our work, in future AI may also influence the big decisions we make in life, such as mortgage decisions. That is why my not knowing in which way algorithms influence my decisions is a problem.
A clear message given by the seminar Work Up! Artificial Intelligence and Technology, arranged in cooperation by the Working Life 2020 project and Helsinki Design Week, was that artificial intelligence should be democratised. The democratisation of AI refers to making sure that everybody understands how artificial intelligence and algorithms work and that everybody is given the opportunity to learn how to understand them. Even though we are capable of assessing the operation of intelligent systems, we do not necessarily understand how they work.
An example where AI grants mortgages was used as a starting point for discussion at the seminar. AI is taught by using teaching data, which may inadvertently teach the system to grant mortgages in an unfair manner, based on the applicant’s gender. This means that AI may learn to base its mortgage decisions on a wrong factor that may not have a direct causality with the applicant’s ability to pay back the loan.
The logic behind AI is a mystery to many people. People do not know what happens in the “black box” between the data and the given final result. They believe that the algorithm automatically spits out results. To ensure transparency and equality, we must understand how AI works, what kind of data it uses and in which ways.
Development of more intelligent technologies is a continuous competition. At the seminar, Minna Mustakallio, a designer at Futurice who specialises in ethics and data, challenged the attendees to think about what we want from technology. She says that the development could be disastrous if this perspective is overlooked. Jaana Leikas, a Principal Scientist at VTT Technical Research Centre of Finland, described a horror scenario where equality in our society is based on technology. AI is incapable of assessing the ethics of its own operations, at least for the time being. This means that programmers and the organisations that develop AI products hold the reins.
As AI technology becomes more advanced and more common, public servants’ area of responsibility will expand. I have had the opportunity to think about AI and the related ethics in more detail this autumn. I’m working with two major projects at present: one surveys AI-based tools and the other focuses on the ethics of public servants. Combined, these projects offer me new insights to both issues.
The fact that AI and robotics are becoming more common can be seen in many ways in public administration. More efficient automated or partially automated solutions are being sought in many administrative sectors. The NAOF’s operations will change in two ways.
Firstly, the development will influence our way of working. In future, AI and robotics will assist us in routine tasks, such as the preparation of basic reports. This change will offer major benefits: the auditors will have more time to focus on the development of their own expertise and more challenging analyses. To be able to utilise the opportunities offered by AI, we must continuously develop our own understanding and critical thinking regarding the operation of AI and its development.
In addition to our internal operations, the development will cause a change to what the NAOF audits. In future, AI will influence more and more decisions made in public administration. We must be able to understand and analyse the decisions that have been made through AI. We cannot succeed in this task without profound and comprehensive understanding of how AI works.
My attitude towards AI has changed this autumn. I still like the fact that AI automates some actions and is able to consider issues that people may not notice. Meanwhile, I have a hard time relying on the final result if I don’t know how AI reached it. A deeper understanding of AI requires investments from individuals, but above all transparency of operations.
A good way to start learning more is participating in Reaktor’s Elements of AI online course. Furthermore, Think Corner will arrange Smart as HEL events in November (5–16 November 2018). These events focus on the digital revolution and the role of humans in it.