Artificial Intelligence (AI) technologies are increasingly present in every area of human life. It is not surprising, therefore, that their diffusion, as this is the case for every potentially disruptive and transversal innovation, is being accompanied by ever broader and deeper attempts to anticipate their ethical, social, psychological, and economic implications, to assess their potential risks and benefits.
Such considerations are also emerging in the area of the use of AI technologies in the field of migration; and not by chance. This is a particularly complex field of application, both because it is politically very sensitive and divisive, and, above all, because it concerns subjects – migrants, precisely – who have high vulnerability profiles and have little chance of protecting themselves or being protected from the intended or unintended effects of technologies.
The question that therefore arises is: To what extent can AI technologies foster a more effective understanding of migration and under what conditions is their use socially acceptable?
Within the CRiTERIA project, we attempt to provide some answers to this question. Let us quickly examine some of the factors that make this issue complicated to tackle.
Social Acceptance of What?
The first factor concerns the very object of social acceptance, i.e., AI technologies.
In the field of migration, artificial intelligence is usually known to be used for border control, for instance, the verification and classification of biometric markers, or to verify the data provided by visa applicants. However, the functions that AI technologies can perform and the types of technologies that can be used are manifold and may not concern individuals, but migration processes in themselves, using social and traditional media as the main source of information. For most of these functions, we are only taking the first steps and many possible applications have yet to be identified. AI can be also applied, e.g., to
- identify the risks to which migrants are exposed,
- generate new knowledge on migration and migrants,
- detect emerging trends in migration (for example, regarding migratory routes or migratory projects),
- understand what migrants think of their migratory experiences,
- recognise migrants’ material and social needs, or
- detect human rights violations.
Some maintain, even with some convincing arguments, that the use of AI technologies is always harmful if applied to migrants. However, we must ask ourselves whether it is possible to block the use of technologies that are expanding in all directions, although creating risky situations. If in doubt, while we wait for a possible (and unlikely) moratorium on AI, we should begin to understand how risks can be reduced while, at the same time, enlarging their potential benefits.
Following, the question mostly shifts to: How it is possible to use AI for functions different from border control or the detection of immediate security risks?
Probably, a cultural and political change is needed. For over twenty years the migration issue has been increasingly treated as a national security problem which has led to a sort of militarization of the entire migration management system. Therefore, thinking about AI technologies used in a way that incorporates the point of view and interests of migrants or addressing non-security-related aspects seems difficult today. Nonetheless, the plurality of uses of AI in the field of migration prevents the matter from being settled too quickly. Promoting the use of AI technologies to protect migrants and their rights is possible and is, in principle, compatible with other aims, although with some compromises.
To analyse social acceptance, therefore, the first point to clarify is not only which technologies we are talking about but also the purposes for which they are used and could be used, being well aware that the possible answers will always be insufficiently clear or even ambiguous.
Social Acceptance by Whom?
Similarly problematic is the identification of the actors who are asked to “accept” or not AI technologies applied to migration.
About four or five decades ago, it was still possible to single out a common political arena or a “public opinion” (let us say, a majority of people sharing a similar perception of issues and questions) to refer to. Speaking of “social acceptance” as a dominant feeling or attitude was somehow possible, with some arrangements.
With the growing fragmentation and diversification of social life in late modernity, the identification of well-defined social groups as well as a unitary or dominant “public opinion” has become increasingly difficult, so much so that today it is even a useless exercise. Personal orientations, attitudes, choices, or lifestyles have an increasing weight in defining how individuals deal with public issues such as technologies or migration while other social variables (belonging to a specific professional group or income class, living in urban or rural areas or adhering to a religious faith) have less and less influence on people’s thinking and behaviours.
Thus, if this is the landscape, the key question is: Social acceptance by whom?
Undoubtedly, the most useful approach is to identify and understand social acceptance as a variable related to the different stakeholders, i.e., groups of people who are socially located in relation to the new technology. In the case of AI technologies applied to migration, we could distinguish, for example, border authorities, law enforcement agencies, national ministries, local authorities, civil society organisations, migrant associations, researchers, migrants, and the like. However, this perspective has some serious limitations, while probably lacking alternatives.
One of them is that stakeholders are not internally homogeneous. Even migrants are not, although they share similar conditions and risks, as they differ widely in their motivations, migratory plans, lifestyles, interests, political orientations, values, and visions of their future.
Another limitation is the asymmetrical distribution of information about the new technology among the stakeholders. For example, most NGOs working with migrants are not informed or even not aware of the use of AI technologies or threat and risk analysis methods in the field of migration. It thus becomes difficult to accept a technology if you do not know it well, although you know what is at stake with it.
Finally, social acceptance is influenced by the role that each stakeholder might have with respect to the new technology. Social acceptance changes if, for example, you are among those who buy that new technology, if you use it, if you benefit from its results, if you can influence how it is used, if you are still involved in some way in its management, and so on.
Social Acceptance of How?
This last consideration leads us to another factor that makes it difficult to address social acceptance: technologies are always strongly intertwined with social dynamics.
For this reason, in the scientific literature, there is a tendency to talk about “socio-technical systems” rather than simply “technological systems”, precisely to highlight how technologies are not produced and used in a social vacuum. They are always socially located: they are embedded in an organization; access to them is always regulated in some way; their results are communicated or shared to a variable extent; and so on.
This has many consequences in terms of social acceptance. One can accept a technology and the goals for which it is used but disagree about how it is used.
For example, many NGOs and civil society organizations would like a management system of AI technologies that is open to external actors, collaborative, transparent, accountable, and constantly supervised by computer scientists and experts (e.g., in human rights or gender issues). This could make these technologies more effective and safer but also more complicated and expensive to handle. For this reason, end users (such as border control authorities) would prefer technology that is easy to manage, inexpensive and does not require much training or major organizational changes.
Also on this front, therefore, it is difficult to address the issue of social acceptance in a simple way and in purely quantitative terms (high, medium, low social acceptance), while it appears more honest to recognize the existence of many different elements that contribute to producing orientations that are diversified, often spurious and not clearly defined.
Social Acceptance as a Perspective
The above leads to the assumption that the “social acceptance” of a given technology – especially when entering complex fields such as migration – cannot be treated as a precise, stable, measurable attitude that can be traced back to a specific social group or stakeholders.
At the same time, this does not mean that studying social acceptance is just a waste of time as it produces results that are unreliable and exposed to rapid change.
Rather, it means that we should consider the social acceptance of a technology, not as a factor that occurs downstream in the production process (a new technology is developed and then we see how it is accepted), but as a process that develops from the design of the new technology and continues throughout its use. In this way, social acceptance becomes a perspective on which to negotiate with the relevant social actors, concerning the goals for which the new technology is used, how it is managed, who can use it, who controls it, who can influence the way it is applied, and so on. Every technical, political, or organisational choice that is made becomes the subject of as far as possible collective reflection and analysis, regardless of which solution is actually adopted (which, more often than not, will be a compromise); what is important is that any choice is made by being aware of its effects, risks, and benefits.
It sounds like a complicated process; perhaps too complicated. Yet, in today’s complicated society, thinking of social acceptance as a process of negotiation seems a good way to help reknit the threads of dialogue in a social field – that of migration – where polarisation is leading to a massive and pervasive distrust, making everything extremely distorted and almost impossible to manage.
Luciano d’Andrea
Luciano d’Andrea has been engaged for over forty years in the domain of sociological research. Such activities were initially concentrated on issues related to development processes. At a later time, they enlarged their scope up to cover a wide array of phenomena – such as international migration, poverty and social exclusion, urbanization or gender inequality – that are connected with each other for their strong societal relevance and their link with globalization dynamics. Luciano d’Andrea has carried out research projects, consulting activities, training programmes and project evaluation exercises on behalf of many renowned NGOs and international organizations in the fields of sustainable development, humanitarian aid, human rights, and action to alleviate poverty. Over the last decade, his research interests have mainly focused on science-society relationships, understood as a privileged observation field to analyze the pervasive and profound transformation processes affecting contemporary societies as a whole, especially in projects funded by the European Commission Directorate General for Research and Innovation
Luciano is part of Conoscenza e Innovazione, Knowledge & Innovation, one of the consortium partners in the CRiTERIA project.
Banner image by Timon Studler on Unsplash.
