Artificial Intelligence and Freedom

Introduction

Artificial Intelligence (AI) generally is based on Deep Learning, where a decision-making process is divided up into numerous small little steps leading to a certain likelihood of an outcome. Often, these steps become too complex for a regular person to understand. Whereas this technology offers great possibilities for progress, it also threatens personal freedom if implemented arbitrarily. That AI is going to take over decisions in a modern society is inevitable and necessary to sustain society’s increased complexity, but it is essential that both the ultimate decision needs to be made by a human being and that the decision-tree remains transparent for all involved.

Problem description

In most parliaments rules are introduced with a keen eye on public opinion but less on administrative execution feasibility, let alone ‘sturdiness’ or the avoidance of individual subjects falling in between the cracks. In addition, the public outrage in case something goes awry – which sometimes just happens to be part of reality – is a grand motivation to becoming ever more risk averse, and thus to regulate into ever more situations. As the administrative apparatus does have physical limits, a growing support by (ever more AI) technology becomes inevitable. As long as situations are standard, AI likely will predict the correct outcome and serve the populace well. However, there is only a likelihood of accuracy and exceptions to an administrative rule will remain and due to ever increased complexity of rules increase in number.

In the European Union, for example, there is freedom of movement in principle. In practice, there are many exceptions to this rule. Below we will examine actual examples out of the life of one single person, who just happened to move to foreign countries a couple of times. The importance of the population registrar and the challenges offered by frequently moving address are shown in these examples, all derived from the life of only this one individual. Fortunately, he was well-educated, white (or ‘pink’ or ‘Caucasian’), born in the EU with a European passport, and he spoke the languages. Imagine how these barriers would work for those less fortunate, or those numerous refugee, whether political or economic, that we are sheltering.

  • Our ‘(anti-)hero’ turned in effect homeless when he returned from living abroad to his home country looking for a job and the last city he lived in refused to let him register for (social) housing because he had only lived there for three years. The previous city, where he had grown up, did the same because he had moved to another city within the country afterwards.
  • When he divorced and moved out of the marital house, he needed to wait for seven years before social housing would be made available to him (agreed, here he probably better had been a refugee).
  • When moving to another country within the European Community he was kicked out of the health insurance because of his new foreign address.
  • When moving back that same health insurance refused him without a citizen registry change. When that change occurred, the insurance unilaterally insured him not per that date but per the original request date, making him double-insured. When confronted with the anomaly the insurance second line refused to correct it but referred to a central government legal committee for rectification.
  • When wanting to exercise his legal right to put in a complaint, a policeman demonstratively tore up the complaint in front of his eyes. (Both a police lieutenant and the mayor later apologized for this aberration but only after persistent follow-up on the content.)
  • When living in one and wanting to move to another European country he found a job in the new country. However, a mortgage advisor refused to take him on as a client to finance a house, or even to call him back/talk to him on his foreign telephone/Facetime/Whatsapp number or let him open an account merely because it was still living outside of the country.
  • He was told that he could not open up a new bank account with a particular bank in the new country since the institution absolutely refused ‘still foreign addresses’ given the bureaucracy involved due to their ‘money laundering’-challenges.
  • More examples are there but that is not the point and would burden this piece to boredom.

All of these challenges he could overcome because he had sufficient income, determination, language skills and procedural knowledge. For a regular refugee without these resources these obstacles would have been fairly insurmountable.

Frequency of problem occurrence

Our Western European society boasts ‘free internal movement of people’. In practice it does not support them. Still, the local binds are loosening and ever more youth are moving from the countryside to the city. Ever more students are studying abroad. Ever more workers are crossing borders as well, as evidenced by – not only – the Brexit fuel-driver problems. The legal crossing of international barriers is increasing significantly, leaving aside the additional illegal crossings.

At the same time local regulation grows, the number of local law books has proliferated, and the issuing authorities have grown with the central EU-bureaucracy – another source of prolific regulation. The execution of rules does not have primary attention at their creation. The ‘Toeslagenaffaire’ in The Netherlands is an example where citizens were (and despite Prime Ministerial excuses and the fall of a cabinet some still are) grinded between the mill stones of government agencies.

A parallel development is the ‘throwing over the fence’ by central governments to local ones, where the execution of regulations is ‘delegated’ with insufficient funds. This ‘political strangulation’ puts ever more pressure on the execution of regulation by local governments. Not only the police needs to prioritize where to enforce, also the mayor’s office needs to decide which parties yells loudest.

There are two main approaches to such complexities: the inherently best one probably is ‘Pushing Back’ and have regulations reduced. For simplicity’s sake I will here consider that method as impractical. The most feasible and thus most likely one seems to be automation, with Robotics Process Automation (the automation of manual administrative steps) and AI Deep Learning as main potentials. It is my expectation that both will be implemented widely in the coming years.

If processes have been documented properly, they show not only the steps to be taken for every conceivable situation but also which responsibility every party has in that process. The most common ways of explaining responsibility is via RACI or RASIC. The most crucial ones are the R which stands for ‘Responsible’ or the person doing the work and the ‘A’ which stands for either ‘Accountable’ or ‘Authorize’ who needs to take the decisions. A fuller explanation of these methods you can find in a previous blog the-ida-ly-cycle/  Unfortunately, quite a number of government measures are implemented under great publicity and time pressure, which bodes ill for proper and esp. complete documentation.

If there is no room for deviation to the bulk of the situations, there will be no freedom left, be it of decision or of action.

Requirements or solution

The term GIGO also applies to AI. But AI brings in an additional danger, namely that of human incomprehension and undue respect, like a reference to ‘AI decided it, and AI is many times more intelligent than I am’. Make no mistake: Artificial Intelligence is a great ‘R’ to do the work, but there are two hard requirements for each and every implementation:

  • the procedure always will need to remain a human who is the ‘A’ of Authorize. Artificial Intelligence should never be allowed to make decisions, only support the decisions of a human being by doing the hard work.
  • allow for a regress-procedure with human pragmatism and intelligence to overcome mistakes. In the speed of progress we will make mistakes, and we will need to make mistakes for progress (“Fail fast”).

Or, in short, a human should oversee AI and take the ultimate decision. The moment we hand over that responsibility we hand over our freedom.

  • Oct 6 Textual edit
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a comment