Artificial Intelligence in the parliament

If we trust traffic lights when we cross the street, and electronic voting to cast our vote, how ready would we be to take a step further and delegate decision-making to a thinking machine?

Machine learning was behind Donald Trump’s victory, Bloomberg has compared him to a failed but nevertheless learning singularity. Surely, nothing bad could happen if an MP were to be replaced by an AI for a month, at least as far as Kunnar Kukk, Proekspert’s Product Owner, is concerned. In late October last year, Tieto announced that they will be using an AI as a member of the leadership team of its new data-driven businesses unit. Associated Press replaced a journalist with an AI for producing routine news back in 2015.

Why does a secretary at a ministry or a local government have to sort incoming mail according to keywords, and redirect them to the right department, if a machine would be more accurate and efficient? Why is it that today, an official has to formulate simple answers to queries, while intelligent chatbots already exist, able to provide quick answers for clients using the company’s support services? Would we soon become used to a thinking robot answering our simple questions, instead of an official, rural municipality mayor, or minister?

Nora to the parliament of Estonia

Estonia held the world´s first parliamentary elections via the Interent and is the first country to offer e-Residency (a transnational digital identity available to anyone in the world interested in administering a location-independent business online). What would happen if one day, the AI Nora replaced a member of the parliament of Estonia for a month, during parliamentary sittings and committee work?

While permission to process state secrets would probably not be given to Nora due to security concerns, since the risk of becoming an object of interest to a neighbouring country’s intelligence agencies is already very high as it is; Nora, trained to make decisions, could probably handle the monthly workload of the 101st MP on the Social Affairs Committee or the Economic Affairs Committee. Not to mention replacing MPs who are keen to have assistants to complete data analysis and background information overviews, because there’s never enough time. Nora would have to learn and analyse a considerably larger amount of information than an average MP would currently be able to during their entire lifetime; and Nora could also formalise and confirm claims that today’s average MPs would most likely struggle with. And if Nora could understand context and ask pithy questions, we could have more in-depth reasoning instead of circumlocution, and perhaps even an initial overview of contradictory pieces of legislation. At the same time, scientists would have the interesting challenge of how to explain to Nora what public interest and human rights are.

The conscience of an AI-MP should primarily be that of a democrat. Nora should probably not be a member of a coalition or an opposition party, nor the ideological supporter of any other party; it would first be necessary to teach Nora the Constitution, and then the Estonian legal system, and the basic principles of parliamentary democracy; and the voting procedure should be changed.

Better decisions

If Nora were to adopt the obligations of an MP completely, we would probably see more reasoned decisions, more quantitative analysis in economics and the development of social services; the decisions would be based on data and ascertainable material, not on emotions reflected in the public, or populism reflected back to the parliament. If Nora were to be taught to recognise and avoid demagoguery and populism, it would be a bitter pill to swallow for the other MPs.

As far as economic affairs are concerned, the new MP could know and understand a wide variety of authors, and also be able to synthesise practical experience into policy suggestions. In foreign affairs, Nora would be a partner who would know how to forecast policy development based on public sources, and to analyse the moves and patterns of the operators. Diplomatic jargon would probably disappear, and if there were a war going on, it would actually be referred to as war — while remaining polite.

Nora could meet with voters 24/7, it would have the relevant capability, and there would be no additional lines on expense reports. Neither would Nora need to spend anything on a car, which is often an objectionable issue for journalists. Nora is able to communicate simultaneously with voters and other policy-makers, regardless of the time and location. This MP 2.0 would be able to work both night and day, while learning from feedback, and creating new models to operate successfully.

Where’s the limit?

Nora’s work can be measured by the quality of its decisions — what will happen if some of its decisions will directly or indirectly cause someone’s death? Who will be responsible: Nora’s creator or the AI itself? An artificial intelligence must be able to connect decisions with potentially fatal consequences, and understand where the limit is.

Fair enough, it doesn’t have to be an MP that the AI would be replacing, and as far as our current know-how is concerned, it would still be rather too difficult to replace the most colorful characters. However, if we were able to use machine learning and service design to develop a simpler, rather routine public service or support service that was previously provided by a person and now by a machine, we could consider this to be a huge step forward. Whether we’re ready for it, is another matter. By tomorrow, we’d at least have a slightly smaller and more optimal government.

Hopefully, members of the parliament of Estonia won’t turn out to be Luddites who are afraid that a machine will come and take their job away, who will cut down on electronic voting, or abandon crazy ideas without fully analysing them first.

Kunnar Kukk is a Product Owner at Proekspert. Our area of expertise is smart devices and robotics, we are keen to contemplate the issues connected with the future development of these areas.