Use of artificial intelligence in public administrations

Aware of this, the Third Sector Platform of Catalonia, within the framework of the conference ‘Artificial Intelligence and Social Rights: Risks and Opportunities’ of the m4Social at the Mobile World Congress, under the umbrella of the Mobile World Capital Foundation, presented the report ‘Radar of algorithms and automated decision-making processes for access to the social rights of citizens’, prepared by Ksnet, where, among other things, the forms that artificial intelligence takes in public administration in the field of social rights are collected.

It is a subject of special interest to the third sector, and that is that much of the management of organizations necessarily passes through the filter of the public sector. That is why the keys set out in this section of the report are now included.

AI in administrations

One of the actions that public administrations around the world are carrying out in terms of AI is to work towards its implementation in improving the effectiveness of their own processes. However, given the fact that one of the main tasks of administrations is social protection, the application of AI in this field requires special caution, because the impact on citizens is very great.

Currently, the technology most used by public administrations is automated decision (ADMS). These are systems that use algorithms and large amounts of data to automate certain processes and, in some cases, do so through AI.

Public AI Areas

With the aim of improving the effectiveness of the different areas of work of the public sector, several states already implement the use of AI in many of their sectors of action. The report mentions the following: employment, crime prevention, transport, social services, police forces, health, education, public policy design, prisons, social aid, fraud detection and justice administration.

In Spain specifically, the procedures that already work with AI-driven software are the automation of the public aid system (with BOSCO), support for police forces (through software such as VioGén and VeriPol) and crime prevention (through RisCanvi).

Algorithms in public sector

Algorithms are sequences of instructions or steps to achieve specific results, and have their maximum exponent on social networks, where they are used to recommend content to users. They are tools that can easily contribute to social biases, and that is why it is necessary to know the application.

According to the report, at European level, the vast majority of algorithms are introduced at national level (54.1%), while a smaller part is dedicated to the local level (26.7%) and a lower percentage at regional level (10.1%) and transnational (9.2%).

In Spain, in particular, a total of 34 algorithms are used and are under development or under pilot testing in public administrations, placing the country in the sixth place of European Union countries that implement algorithms in the public sector, below France (37), Germany (39), Estonia (50), Portugal (60), Italy (63) and the Low Countries (116).

Benefits of AI in administrations

The Table report identifies some benefits of the already implemented AI application to the public administration. The first is a verifiable improvement in the design and provision of public services, as it is a tool that allows identifying perceptions, problems of access to services and behaviors of citizens to include them in the administration agenda. In addition, AI allows you to manage unprecedented amounts of data, especially through social networks and mobile devices, data that can be used to offer citizens a more personalized, effective and sustainable attention economically and over time.

At the same time, the use of AI in administrations is leading to an improvement in the internal management of institutions, especially at the state level. In this sense, one of the optimal uses of this technology is the automation of repetitive tasks, which allows human efforts to be dedicated to a more efficient and, at the same time, personal attention.

Risks of AI in administrations

The first risk identified in the report is the lack of accountability, transparency and explainability of algorithms that drive AI to public administration. This refers to the difficulties that citizens encounter in understanding the functioning of the mechanism. This is one of the issues being discussed in the drafting of AI legislation, and one of the main sources of concern and struggle of technology entities. The cry is, in this case, that people who use public services must understand how the result has been achieved. This is the only way to guarantee the social rights of citizens, putting AI at the service of the common good.

Another proven risk of AI being applied to public administration is the reproduction and amplification of biases and discrimination. This problem arises in the creation phase of the software, when the data entered is false or insufficient.

A last major risk that the report points out is the lack of security and privacy of the information used, which is highly worrying given that the information that administrations release is usually very sensitive.

The result of these risks is the direct impact on the implementation of public policies, while perpetuating inequalities and greatly reducing the effectiveness of initiatives and the functioning of administrations themselves and, with it, the confidence of the population.