Abstract
This article presents an analysis of European smart city narratives and how they evolved under the pressure of the COVID-19 pandemic. We start with Joss et al.’s observation that the smart-city discourse is presently in flux, engaged in intensive boundary-work and struggling to gain wider support. We approach this process from the critical perspective of surveillance capitalism, as proposed by Zuboff, to highlight the growing privacy concerns related to technological development. Our results are based on analysing 184 articles regarding smart-city solutions, published on social media by five European journals between 2017 and 2021. We adopted both human and machine coding processes for qualitative and quantitative analysis of our data. As a result, we identified the main actors and four dominant narratives: regulation of artificial intelligence and facial recognition, technological fight with the climate emergency, contact tracing apps and the potential of 5G technology to boost the digitalisation processes. Our analysis shows the growing number of positive narratives underlining the importance of technology in fighting the pandemic and mitigating the climate emergency, but the latter is often mentioned in a tokenistic fashion. Right to privacy considerations are central for two out of four discovered topics. We found that the main rationale for the development of surveillance technologies relates to the competitiveness of the EU in the global technological rivalry, while ambitions like increasing societal well-being or safeguarding the transparency of new policies are nearly non-existent.
Introduction
As a result of global pressure to make cities safer, more competitive and climate neutral, for the last 20 years there have been efforts to integrate the smart-city model as part of a broader agenda for urban development (Albino et al., 2015; Swyngedouw et al., 2002). Currently, smart cities are at the centre of the European Commission’s (EC) strategic vision for a modern, competitive and climate-neutral economy (European Commission, 2019, 2020).
The rationale for smart cities development is changing with time, and what is accepted or expected from the smart solutions depends on the problems it promises to fix. The initial narrative concerning smart cities focused on increased efficiency in managing resources, assets and services (Cohen, 2015). Lately, the discursive goal of smart-city policies became the formation of climate-neutral, sustainable cities in which environment changes are monitored, resources optimised and production of waste limited. All of which is enabled by digital surveillance technologies based on a system of sensors, cameras, drones and diverse artificial intelligence (AI) algorithms monitoring urban processes. Smart-city proponents suggest that it can not only address issues of resource efficiency or security, but also enable social cohesion and participation, while strengthening evidence-based policymaking (Karvonen et al., 2019).
A plethora of cities is turning smart rhetoric into reality, becoming ‘actually existing’ smart cities (Shelton et al., 2015) by introducing information and communications technology (ICT) in urban policies, agendas, narratives and aspirations. This accelerating pace of ‘smartification’ produces mixed outcomes and brings fundamental changes in the way that cities operate and how they are experienced (Curran and Smart, 2021). City policymakers endogenise the process of technology-led growth, to develop global reputations of their cities as liveable, competitive and progressive. As a result, smart urbanism becomes ‘a normative aspiration for the urban future’ (Kong and Woods, 2018: 681).
At the same time, smart urbanism has attracted multidimensional criticism (e.g. Greenfield, 2013; Kitchin, 2015; Sennett, 2012; Swyngedouw, 2016; Vanolo, 2014), with authors arguing that smart-city practices strengthen the neoliberal economic agenda (Söderström et al., 2014) and offer economic benefits mainly to city authorities and technology providers (Cardullo and Kitchin, 2019; Luque-Ayala and Marvin, 2015). As Joss et al. (2019) point out, while the public interest is claimed to be a key focus, smart-city implementation predominantly serves as an extension of the private sector, a platform for economic renewal through cutting-edge innovation. This creates multifaceted concerns related to privacy, cyber security and digital rights in our cities (Calzada et al., 2021).
The current smart-city agenda has also embraced the ecological approach to urban development (Trencher and Karvonen, 2019), which is often tokenistic, using environmental responsibility as part of the smart city’s sales pitch (Joss et al., 2019). The narrative ascent of the smart-city agenda has coincided with the climate emergency discourse, but aside from the general claims, wider environmental concerns (climate justice, biodiversity, etc.) are barely mentioned.
In this article, we study certain aspects of the European narrative concerning smart cities and its relation to surveillance concerns. We specifically focus on how the COVID-19 pandemic and climate crisis influenced the narrative concerning controversial smart-city technologies, which were often fast-tracked between 2020 and 2021 (Costa and Peixoto, 2020). Smart cities generated a ‘powerful discourse regime’ which merits critical analysis, especially while replication and scaling are currently ongoing. We start with Joss et al.’s (2019) observation that the smart-city discourse is presently in flux, engaged in intensive boundary-work and evidently struggling to gain wider support on several fronts.
The next section presents a literature review on the importance of narratives and the notion of surveillance capitalism; we then present the research objective and our analytical approach, followed by the presentation and discussion of the results, leading to the conclusions.
The importance of narratives in the context of smart surveillance
Narratives are an important aspect of public policy because they not only help people to build a shared understanding of reality but also reflect and shape people’s attitudes and expectations (Collier, 2016; Narlikar and Sottilotta, 2021), which in the context of crisis can strongly influence citizens’ response. Especially when tough choices must be made, narratives can be decisive in defining the choice architecture on the individual as well as policy level (Wydick, 2015).
Jirón et al. (2021) point out that smart-city transitions often focus on creating the narrative of vaguely defined future benefits instead of quick fixes to the existing issues. For example, Melgaço and Van Brakel (2021) compare investment in mass surveillance to a theatre in which politicians play the role of crime fighters, while not being exactly interested in the effects of implemented solutions.
An important line of criticism is rooted in the concept of surveillance capitalism proposed by Zuboff (2019), who points out threats resulting from the commodification and commercialisation of data by Big Tech, which is further facilitated by smart-city technologies (Kitchin et al., 2018; Zuboff, 2019). Similarly, Wood (2015) argues that smart cities are by definition surveillance cities because their smartness is just a function of the data they constantly elicit from citizens. That is why radical asymmetries of access to big data (Barns, 2021) and privacy (Löfgren and Webster, 2020) are often flagged as critical concerns.
Despite those growing reservations, the last decade was marked by rapid growth in terms of smart-city policies and implementations. In 2020, the COVID-19 pandemic created a powerful push towards the development of monitoring and tracking technologies (temperature measuring points, compulsory apps or contact tracing), mainly implemented in dense urban environments. The intensity of COVID-19-related surveillance made some people wary of giving up their privacy for the sake of unregulated data collection. In this context, we wanted to identify the dominant narratives in the European digital outlets, to see what main directions, arguments and sentiments are present in the public discourse.
The meaning-making function of narratives (Boin et al., 2016) advanced by political leaders acquires a higher relevance in this context. Formulating a persuading narrative is often crucial for policy implementation and a key source of legitimation (Jones et al., 2014). In their analysis of global discourse reflected in 27 smart cities, Joss et al. (2019) illustrate how some narratives inform and shape policies (Bevir and Rhodes, 2006), providing persuasive public accounts of what is happening, why and what can be done about it (‘t Hart and Tindall, 2009). As Drennan et al. (2014) point out, leaders – as primary storytellers – explain and simplify events during crises, to influence the behaviour and performance of other actors.
Research objective
Our goal was to analyse the digital press narrative, to see if there was any significant change in how smart technologies were described over the last four years. We were also interested to see if the criticism towards smart-city solutions and related surveillance technologies became more or less prominent in the pandemic context. We start our analysis in 2017, which saw the delivery of most of the legislative proposals and policy initiatives announced in the EU Digital Single Market Strategy from 2015.
We focused on the public discourse on this subject, as reflected in five Europe-centred journals, to identify main narratives as well as leading actors. We expected that in publications related to the wider critical discourse concerning smart cities we would observe a shift from the enthusiasm regarding technological solutions, to a more critical approach towards the accelerated implementation of new surveillance technologies, like tracing apps or temperature-measuring points, especially after the Cambridge Analytica (Cadwalladr and Graham-Harrison, 2018) scandal and the discussion on surveillance capitalisms voiced by Zuboff (2019).
Our main focus was on surveillance in the context of European smart cities policies as well as specific technologies and implementations. We hypothesised that previous concerns about the loss of privacy and excessive surveillance might lose significance in the crisis context (Kunzmann, 2020). Secondly, we wanted to see whether the COVID-19 pandemic changed the narrative on the smart transition of European cities and consequently accelerated the implementation of policies and technologies. We expected that the intensity of the changes shifted the existing narrative towards threats that excessive data collection may pose to citizens. Thirdly, our goal was also methodological – we took this opportunity to compare the results of quantitative topic modelling with qualitative expert analysis of the same set of articles.
Our main research questions were: (1) What are the main narratives concerning smart technologies present in the studied media outlets? (2) Is surveillance and privacy an important part of narratives about smart-city implementations? (3) Who are the main actors and what is the sentiment towards surveillance and smart-city technologies in those narratives? (4) Has the pandemic changed the narratives on surveillance and smart-city technologies? (5) Is machine coding of narratives accurate and as such a useful tool in discourse analysis?
Data and analytical strategy
We collected articles from five leading European online news outlets that publish articles in English on European affairs (predominantly they discuss issues concerning the European Union): POLITICO Europe (PE), Euronews English (EE), EUobserver (EUo), EURACTIV (EURA) and The Parliament Magazine (PM).
We started by collecting the content of all posts created by these journals on Facebook from 1 January 2017 to 25 March 2021. For each post, we followed the link to the promoted article on the webpage of a given outlet and scraped its content. In this way, we assembled a database of 70,157 articles. In order to analyse articles discussing smart cities, technologies and surveillance, we selected only those containing at least one of the following pairs of keywords: (1) ‘smart’ and ‘city’, (2) ‘city’ and ‘surveillance’, (3) ‘app’ and ‘surveillance’, (4) ‘AI’ and ‘surveillance’. This gave us a subset of 625 articles. Finally, we read the leads of all articles in the subset and selected 184 that specifically covered the issues of smart-city technologies and surveillance.
We used the number of likes of the outlets’ Facebook fan pages as a proxy for their popularity and impact. We focused on journals that cover European affairs to have a set of publishers that are homogenous in terms of topics covered as well as style and language, given the fact that they address more or less the same audience. This corpus construction criteria minimalised the risk that our topic modelling would be driven by differences in terms of basic stylistic properties instead of the actual topics discussed in the articles.
Quantitative analysis
First, we investigated how frequently articles on smart cities, surveillance and surveillance technologies were published over time. Then we performed topic modelling using Latent Dirichlet Allocation (LDA) (Blei et al., 2003). We used a full-featured, industrial-strength implementation provided by the Gensim library for Python programming language (Rehurek and Sojka, 2010). This allowed for separating groups of articles (topics) that are characterised by similar word distributions.
The LDA method assumes that the observed patterns in the document can be explained by the fact that each document is a mixture of a small number of topics and each topic is associated with a particular probability distribution over words. Therefore, articles related to the same topics use similar words. Although it is a somewhat simplistic approach that does not account for higher-order dependencies between words (bigrams, trigrams, etc.), it usually produces easily interpretable results and can be estimated based on a relatively limited dataset. The second property was especially important in our case, since after the filtering by keywords we were left with 184 articles.
Even though the topics were discovered based on the observed distributions of words and therefore results are purely data-driven, the crucial decision on the number of topics to look for was made by the researcher. We based the selection of the correct number of topics (groups into which the articles are divided) on a coherence measure that considers both the similarity of the articles within the topics and the variety of the topics (Röder et al., 2015). This required producing several models for different numbers of clusters (topics) and comparing them on their coherence measure. We tested division into 3–10 topics and chose the solution that maximised an asymmetric confirmation measure – the UMass coherence index proposed by Mimno et al. (2011).
The topic classification provided by LDA is probabilistic. That is, for every article, there is the probability that it covers any given topic. Hence, in some cases when topic probabilities are similar, assigning the article to a single topic may be somewhat arbitrary. However, this issue affects any probabilistic topic-modelling method and therefore can hardly be avoided. Moreover, this also reflects the fact that articles very often discuss different topics and issues at the same time.
Furthermore, for each topic, we identified actors using a full-featured, state-of-the-art Natural Language Processing library for Python – SpaCy (Honnibal et al., 2018). This allowed for Named Entity Recognition (NER) tagging. Based on the results, we were able to identify the main actors in the whole corpus and to track their appearance over time in each of the topics. We calculated a saturation index showing the frequency in which individual actors appear in articles from a given topic. The saturation index ranges from 0 (the actor is not mentioned in any of the articles within the topic) to 1 (all articles in the topic mention the actor). Finally, we computed the sentiment of each of the topics. We built our metric on top of the popular sentiment model for English – VADER (Hutto and Gilbert, 2014). To be able to compare the texts of different lengths, we defined a simple metric that weighs the compound score, a sum of the valence scores of each word in the lexicon, by the actual rate of non-neutral sentiment:
Qualitative analysis
As the final step of our analysis, we performed a qualitative analysis of the corpus of all 184 articles. Three researchers read and coded a third of the articles in MAXQDA software, developing data-driven coding trees. The trees were compared and merged into the new coding tree, focusing on the main themes and recurring topics. All codes as well as controversial cases were discussed and reviewed by all, to make sure researchers share the same definitions of codes. The main areas of analysis concerned the identification of actors (including cities, countries, institutions, companies and NGOs), identification of main technologies and extraction of threats and opportunities that emerging technology brings about.
Results
We started our analysis by examining the co-occurrence of keywords like ‘smart’ and ‘city’, ‘surveillance’ and ‘privacy’ over time. The main aim here was to investigate the effect of the COVID-19 pandemic on narratives concerning smart cities, surveillance and technologies like AI or contact-tracing apps. We discovered that the topic of the smart city, although regularly mentioned, was not very popular in the studied period (2017–2021). On average, there was one article published on the subject per month.
Surveillance and privacy in narratives on smart cities
The link between smart cities and surveillance was not uniform in time. At the beginning of the studied period, the articles on smart cities did not reflect on surveillance or privacy at all. These issues entered the discussion on cities only in the second quarter of 2018, the same time when the malpractice of Cambridge Analytica was revealed. Articles on technologies like smartphone apps and AI started to touch on the topic of surveillance and privacy at a similar time.
Figure 1 illustrates that the COVID-19 pandemic had an especially big effect on the frequency of occurrence of articles on the topic of apps and surveillance. There was a clear spike at the beginning of the pandemic in the volume of articles that contained pairs of words like ‘app’ and ‘privacy’ or ‘app’ and ‘surveillance’.
Privacy and surveillance in articles about apps and AI. Red (black) indicates articles published after the onset of the pandemic.
The popularity of discourse on surveillance and privacy in the context of AI was also affected by the onset of the pandemic, but in the opposite direction – the pandemic narrative seemingly stifled the topic for a few months. The peak of popularity of AI-related topics in which surveillance or privacy was mentioned fell just a month before the pandemic outbreak (Figure 1). The pandemic had the same chilling effect on the occurrence of articles about smart cities – since March 2020, in some months the topic was not mentioned even once in our corpus. During the pandemic, the city was discussed in the context of surveillance, but less in the context of privacy.
Main actors of the narratives
Here we report only on the frequency with which the specific cities, countries and institutions are mentioned in the selected articles, while the context in which specific actors appear will be discussed later. Not many European cities were mentioned by name. The most popular were Brussels (96), London (55), Stockholm (54), Berlin (46) and Paris (41). Other, non-European cities appearing relatively often were Beijing, Hong Kong, Washington and Moscow. In terms of regions, Europe was the most often cited geographical location (514), which comes as no surprise considering the choice of media outlets. In terms of countries present in the analysed articles, the United States (272), the United Kingdom (197) and China (192) were the most often mentioned countries from outside of the EU. Germany (152) and France (121) were referred to the most among the member states. Other frequently discussed country-actors included Poland (58), Italy (54), Sweden (39), Finland (34) and Spain (34).
Tech companies were mentioned quite often, and while most of them represented the US Silicon Valley companies (Google: 178, Apple: 123 and Facebook: 196), there were also numerous mentions of Chinese companies, including Huawei (127), ByteDance (43; a Chinese tech giant and owner of TikTok), Alibaba (10; China’s dominant e-commerce and cloud company), Nuctech (8; a state-owned manufacturer of X-ray machines, scanners and explosive detection systems) and DJI (4; a private drone-manufacturing company). Telephone companies emerged as a specific cluster in the discussion, with Vodafone (20) and Ericsson (5) being most frequent.
European Union institutions were referred to in various manners: as EU (805), EC (139) or Brussels (96). Taken together, this was the most important group of actors present in the analysed corpus. Police forces from different countries and cities were often mentioned in the context of threats related to technological implementations (217). The watchdog organisations, although not as visible as public and private organisations, also appeared in the narratives. Privacy Shield (18), Amnesty International (17) and Human Rights Watch (5) were the three most often mentioned organisations.
Main technologies
In 184 analysed articles, we counted 679 mentions of AI (or artificial intelligence), 388 mentions of the 5G network, 327 mentions of face recognition technologies (FR) and 103 mentions of tracing (or tracking) apps. Other technologies included self-driving cars (or driverless cars) (113), drones (51), robots (46), Bluetooth (27) and other forms of automation (10).
Main topics on smart cities and surveillance
The corpus of the articles was divided automatically via LDA into seven main topics, out of which three were excluded. Two of those were composed of only four to five articles, too small to perform a reliable quantitative analysis. The third excluded group was composed of 14 articles, mainly Digital Briefs published by EURACTIV, which are too short to extract any narrative.
In the end, four main topics of narratives were defined: AI and FR (Topic 1), innovations that help to overcome the climate emergency (Topic 2), tracking apps (Topic 3) and 5G technology (Topic 4). Below we describe each topic in detail, summarising the results in Table 1.
Summary of main topics (saturation index in parentheses).
Topic 1: AI and facial recognition
The first and most prominent topic was composed of 82 articles that dealt with AI, FR and their regulation (Figure 2). The geopolitical focus of this topic lies with the EU, China and the USA (Table 1). Police of different countries are the main actor in many articles, while companies play a secondary role. The narrative on AI and FR quite often refers to privacy and surveillance concerns. Privacy as well as the General Data Protection Regulation (GDPR) introduced by the EC are mentioned regularly.
Most probable words in Topic 1 (left); probability of words unique to Topic 1 (right).
The algorithmic analysis of sentiment shows that although the majority of articles are rather positive, a quarter of them reveal a negative attitude to either the AI and FR technology or their regulation (see Figure 3).
The dynamics of the articles’ sentiment in Topic 1. Red (black) indicates articles published after the onset of the pandemic.
Qualitative analysis showed that FR systems were often mentioned in the context of police efforts to introduce new safety measures in the public space as well as during protests or other mass events. FR requires creating a database of faces and their IDs, and that is why it often involves the use of some form of AI algorithm to compute and compare large bodies of data. The articles concerning FR often pointed out examples of AI bias and the threat of discrimination of specific groups – women, racial and ethnic minorities and people with disabilities. The articles on AI and FR raised concerns on two main levels: AI bias and FR threat to privacy.
Topic 2: Smart technologies to fight the climate emergency
The second topic was composed of 38 articles that represented the discussion of Europe’s fight against the climate emergency with innovation and smart technologies (Figure 4). Qualitative analysis pointed out the connection of smart cities with creativity, waste management, energy, heating and transportation. In most of the articles, the focus was on the EU. Only individual mentions of other countries (like China, the USA, Sweden and Germany), as well as technological companies (like Apple, Google and Nokia), could be found. This type of narration was characteristic for the pre-pandemic period and was definitely positive (compare Figure 5). Surveillance and privacy were not a cornerstone of this topic.
Most probable words in Topic 2 (left); probability of words unique to Topic 2 (right).
The dynamics of the articles’ sentiment in Topic 2.
The qualitative analysis of articles in this topic showed that the issue of dealing with the side-products of technological development raises a pressing challenge. According to some articles, recycling smartphones, cables, computers or microchips is an important part of building a sustainable smart future for European cities – firstly because this type of waste is dangerous, and secondly because the rare metals are running out and should be reused.
Another important environmental issue mentioned was the heating effect of server farms and other ICT infrastructure required for the implementation of smart-city innovations. Some circular, closed-loop and nature-based solutions were mentioned as possible solutions to this challenge. Third, a prominent subtopic described the process of developing smart solutions for cities as the necessary response to the climate emergency.
Topic 3: Technology to fight the COVID-19 pandemic
The third group of articles was composed of 28 articles that addressed the topic of contact-tracing apps as a tool for fighting the spread of COVID-19 (Figure 6). The apps, developed in many countries simultaneously, were supposed to help manage testing and quarantine processes and consequently reduce the spread of COVID-19. This narration surfaced in 2020 and lasted until the end of the examined period (see Figure 7). The most discussed regions/countries in this context were the EU, Germany, France, UK and Poland, with China playing a peripheral role. Two international technological companies were important actors here – Google and Apple. Although the problem of privacy and surveillance was discussed greatly within this topic, most of the articles seem to have positive sentiment, and only some expressed negative sentiment (see Figure 7).
Most probable words in Topic 3 (left); probability of words unique to Topic 3 (right).
The dynamics of the articles’ sentiment in Topic 3.
Qualitative analysis showed that tracking apps were often described as a ‘necessary evil’ and part of national efforts to curb the spread of the virus. Articles often cited criticism concerning the amount of private data that is being collected, as well as doubts about the potential collaboration of governments with tech giants like Apple or Google in creating tracing applications.
Topic 4: For digitalisation of Europe
The fourth topic was composed of 13 articles that focused on the potential of 5G technology to boost the digitalisation processes in Europe (Figure 8). The implementation of 5G technology is presented as a cornerstone of the development of smart cities that utilise countless smart devices. The high occurrence of words like health, body, waves and radiation indicated that the discourse on the development of 5G is accompanied by health concerns. The narrative about 5G is relatively new – it started in 2019 – and it refers mainly to the EU and the EC. The companies mentioned in this context were Amazon, Google, Facebook and Huawei. The issues of surveillance and privacy almost did not occur, and the sentiment of the narration was positive (see Figure 9).
Most probable words in Topic 4 (left); probability of words unique to Topic 4 (right).
The dynamics of the articles’ sentiment in Topic 4. Red (black) indicates articles published after the onset of the pandemic.
The qualitative analysis of articles revealed that due to higher download speeds, 5G is expected to enable new applications of the internet of things (IoT) and other forms of digital services. Articles mentioning this technology usually warned against the growing dominance of non-European suppliers, mainly Huawei, which was meant to provide 5G infrastructure for the EU digital ecosystem. Most articles discussed Huawei in terms of potential threats to European data security. Often the narrative touched upon the risk of 5G networks increasing the vulnerability of core functions of public data systems to potential errors and cyber-attacks.
Interestingly, we could observe the change of the 5G narrative over time. At the beginning of the implementation, there was more criticism, with many mentions of Huawei dominance as a threat to EU security in 2019 articles. However, with time, the tone of the narrative changed and focused on the issue of global competitiveness and how developing European 5G solutions is the only way for the EU to remain a relevant actor on the global digital market, without risking Chinese technological monopoly within the EU as well as globally.
The effects of the COVID-19 pandemic on narratives: Qualitative analysis
Before the pandemic, we distinguish three main narratives focused on (a) privacy and data security, (b) economic competitiveness of the EU and (c) measures mitigating the climate emergency in EU cities. The issue of climate change was very much pronounced as both the reason for implementing new technologies as well as the higher purpose of smart-city technologies. Achieving carbon neutrality and mitigating the climate crisis set the ground for a positive evaluation of new smart-city technologies within the studied narrative, as predicted by Joss et al.’s (2019) findings.
Since the start of the COVID-19 pandemic, an increase in articles focused on tracing apps and surveillance technologies could be observed. We could also distinguish an additional topic that has not emerged in the first step of quantitative topic analysis, as detailed in the next section.
Pandemic as a fast-tracking factor in technological development
One of the leitmotifs in the discussion concerning the technological countermeasures to the pandemic was the widespread privacy concern and fear that once the surveillance technology had been implemented it would not be switched off after the health threat was over: The response of EU countries to the coronavirus outbreak has prompted unprecedented levels of surveillance, data exploitation, and misinformation. Data collection can be essential to understand and respond to the Covid-19 emergency, but creating such digital surveillance risks failure and adverse side-effects. (EUobserver, 12 April 2020)
Privacy concerns, similarly to those emerging in other research (Löfgren and Webster, 2020), were related to the amount of private data (including sensitive data on health, contact lists, etc.) that contact-tracing apps are collecting and fears that the data will not be safe and may be used for commercial or political reasons. The speed of introducing new technologies created understandable doubts about their security and resistance to theft or other forms of abuse: Big Tech has become a global punching bag because of fears that Silicon Valley has too much control over our daily lives. But in their legitimate efforts to keep people safe, officials across the European Union, United States and elsewhere are quickly falling into the same trap – creating a government surveillance network on the fly, with little oversight and almost no clarity about when it will be shut down. (POLITICO Europe, 19 April 2020)
Those doubts bring the trade-off between health and privacy to light. The need to quickly implement countermeasures to stop the spread of the virus meant that many governments decided to suspend citizens’ right to privacy. The urgency of the matter meant that no wider public debate on the topic was possible. Therefore, some of the articles underlined the need for public scrutiny over those new apps and different ICT solutions: In these unprecedented times, there has been a monumental shift in personal liberties, and there is undoubtedly a trade-off to be had between privacy and the need to curtail the spread of the pandemic, but what measures are appropriate, and will they have a long-lasting societal effect? (Euronews English, 31 March 2020)
In the case of some articles, authors and cited experts pointed out that the governments (sometimes hand in hand with technological companies) are using the pandemic as an excuse to fast-track surveillance technologies. For governments, those technologies are useful tools for amassing power, stifling protests or weakening the opposition. For companies, they can be easily turned into profits and sold to the highest bidder: If, by any chance, mass surveillance develops in Europe, it will not have been for the coronavirus. The coronavirus will have been an excuse for this to happen (Euronews English, 31 March 2020).
As officials around the world struggled to contain the COVID-19 pandemic, they were increasingly turning to the surveillance methods already used in some Asian countries to track people’s movements and contacts. That has raised questions about privacy and the timeframe of urgency measures – in other words, when the tracking will be switched off. Some articles pointed out the need to think about the after-COVID-19 reality and make sure that surveillance does not become a ‘new normal’ but is rather a temporary fix (see also Kunzmann, 2020). Otherwise, some authors warned, European countries and cities risk turning into surveillance zones.
A solution to this threat may be to give control over harvested data to civil society actors rather than governments or private firms. However, it seems that the EC can only create guidelines and recommendations, while the nation-states and municipalities are experimenting with different solutions and regulations concerning data protection.
Technology threats: Qualitative analysis
The quantitative analysis allowed us to distinguish technology threats that repeatedly appeared in the articles but did not show up in the quantitative analysis.
The narrative about technological threat usually serves as a negative benchmark or ‘dark scenario’ used to illustrate the possible dangers of abusing technological developments to tighten control over citizens and strengthen non-democratic regimes. China and Russia are often mentioned in this context, followed by the USA, Hungary and Serbia.
China has been using cutting-edge AI to build up a high-tech surveillance system and crack down on political dissent, according to media reports — and other countries around the world, including European nations such as Serbia, have struck deals with Chinese companies to introduce their own systems aimed at monitoring individuals. (POLITICO Europe, 24 June 2019)
Another narrative focuses on the worrying developments within the European Union, concerning the growing power supplied to governments by big data and surveillance technologies. The same threats were also well described in Curran and Smart (2020) article on China: In tackling the pandemic, we must avoid Europe sleepwalking into a permanent expanded surveillance state (EUobserver, 24 April 2020).
This narrative specifically relates to the use of cameras and drones by the police and border service. The main concern here is that even if face recognition, surveillance drones or contact tracing are implemented in good faith, the solutions will allow the abuse of power in the longer run. Distrust towards the government, in countries such as Germany or the UK, is the main leitmotif here. New developments concerning the pandemic are also mentioned in this context, as factors fast-tracking the development of surveillance that may stay in place even after the state of emergency: People are growing increasingly sceptical about how Google, Amazon and Facebook collect reams of our personal data, often in opaque ways that are hard to decipher. So why shouldn’t governments come under the same scrutiny? (POLITICO Europe, 19 April 2020).
We found one more recurring narrative concerning the possible societal dangers of implementing AI in different policy areas, relating to face recognition technology. Articles often point out that AI proves to incorporate human bias and even strengthens it, which is visible in its reactions and evaluation procedures. As a result, AI often discriminates against women, as well as racial or ethnic groups. Another threat relates to the biased attribution of lower grades or of crimes to minorities. The lack of transparency or ‘opaqueness’ of AI algorithms brings concerns about potential ways to fix the bias. This is well described in the works of Cugurullo (2020) or Curran and Smart (2020), which show how risks stemming from those threats are unequally distributed in society: Kitting out our streets, supermarkets, and parks with these technologies will amplify existing discrimination against people of colour, people with disabilities, and other marginalised groups. Instead of being contestable, these practices will be hidden under a veil of false scientific authority and deliberate opacity. (Euronews English, 17 February 2021)
In the analysed text, we find the claim that any type of AI application in public policies or public space should be preceded by open democratic debate and that the technological solutions should be more transparent. However, as mentioned in the introduction, the recent COVID-19-related crisis allowed for fast-tracking ‘smartification’, while lockdowns restricted the right to protest.
Discussion
Our study is based on a two-step analysis of the corpus of articles in EU-focused online press. By comparing the results of the quantitative topic analysis with in-depth discourse analysis, we managed to confirm the general topics emerging in the studied publications. The qualitative analysis allowed us to distinguish an important sub-topic concerning the pandemic as a fast-tracking factor in technological development and to analyse technological threats mentioned in the studied narratives. However, in terms of identifying main topics and trends in the narratives, the quantitative approach seems accurate enough and allows for reviewing large archives of texts in a much shorter time.
In relation to our research questions, we distinguished four main narratives: first concerning AI and FR, second related to the climate emergency, third regarding anti-pandemic measures and fourth focused on the 5G network. We found that privacy and surveillance were present in AI and FR narratives accompanied by the polarised sentiment, as well as in anti-pandemic ones, where they were often referred to as a ‘necessary evil’. We also identified the main actors – in terms of countries, cities and companies – of those narratives, with the EU often playing a central role but China and Big Tech companies also being frequently mentioned.
While we wanted to establish if the pandemic changed the narratives on surveillance and smart-city technologies, our analysis showed the growing number of positive narratives underlying the importance of technology in fighting the pandemic and mitigating the climate emergency. We also found that in the context of those momentous technological changes that are underway in Europe, the narrative remains rather predictable and shallow, repeating and recycling topics without trying to present deeper arguments or long-term solutions. This is in line with the findings of Zuboff (2019) or Vanolo (2014), who point out how the smart-city discourse may be a powerful tool for the production of ‘docile subjects’ and political legitimisation of Big Tech business goals.
This is confirmed by the fact that most utterances concerning pandemic or climate crisis boil down to the digital economy race. Similarly, the crucial argument for fast-tracking smart solutions relates to the need for securing the competitiveness of the EU in the global technological rivalry rather than increasing citizens’ well-being or safeguarding the democratic transparency of the introduced innovations. This exemplifies Melgaço and Van Brakel’s (2021) concept of smart cities as surveillance theatres in which power relations play an important role.
When citing EC officials, the narratives repeat the promise of good regulations enabling beneficial technologies while curbing their negative consequences. In the context of the COVID-19 pandemic, however, those regulations were not in place yet, while there was an urgent push to introduce technological measures that could help slow down the virus’ spread.
The optimistic storyline about the EU’s growing position in ‘the global fight for tech supremacy’ is based on references to new regulation proposals and policy initiatives. What connects those arguments is the repeated metaphor of the ‘digital race’ in which the EU must up its game to avoid marginalisation or even takeover by global tech players. This brought in the flurry of policy initiatives to protect Europe from its dependence on foreign-owned tech companies while boosting the EU’s own tech sector to compete against rivals in China and the United States. It also creates a temptation to implement ‘smartification’ faster, without constraints of the democratic debate and long-term negotiations of the new socio-technological contract: Digitalisation is a race we are bound to lose if we do not move quickly. (…) However, we are held back by bureaucracy and slow decision-making processes, while our competitors are rushing forward and do not play by the rules. (The Parliament Magazine, 4 August 2020)
This could be interpreted as a sign that the logic of surveillance capitalism is deeply embedded in the European narratives concerning smart technologies. The trade-off between data privacy and economic gains seems to have only one preferred solution – investing in digitalisation to ensure ‘the digital dividend’ to the citizens of Europe. With trade moving to the digital world, Europe cannot afford to lag, while China continues to advance.
Apart from the economic argument, there are two strong explanations why the only way for digital Europe is forward. One is the aftermath of the pandemic, which means that the region will face even more challenges in terms of public spending on health. This creates an opening for a wider introduction of AI algorithms and smart solutions (sensors, monitoring devices, remote tracing) in healthcare. As one of the articles from March 2021 points out, the pandemic is a chance to speed up the technological revolution in healthcare: The question now is whether this pandemic is the moment to make the cultural shift in healthcare needed to take advantage of the technological revolution that digitisation and digitalisation can offer (Euronews English, 30 March 2021).
Another strong argument for enabling digital transformation and the introduction of new smart-city technologies is the need for ‘twin transition’ based on digitalisation for the benefit of the environment. An investment programme that links environmental aspects with climate neutrality is therefore the logical next step. As another recent analysis points out, this means heavily investing in previously criticised technologies to meet EU climate goals.
Conclusions
As Cardullo and Kitchin (2019) point out and our analysis confirms, most attempts to recast the smart city as citizen-focused remain rooted in economic discourses and paternalistic practices rather than those of social rights, political citizenship and the common good. Moreover, sometimes smart-city transitions result in the creation of the narrative of vaguely defined future benefits, instead of proposing specific solutions to persistent challenges (Jirón et al., 2021; Valdez et al., 2018). Despite the need for reflection on digital transformation, the matter is often marginalised even in political discourse. Digitalisation is rarely a subject of political debate unless it fits parties’ programmatic profile or government finds it to be opportune (König and Wenzelburger, 2019). Without new narratives focusing on those challenges, we cannot hope for better smart-city policies, as they require an alternative concept of smart citizenship, guided by the right to the city, not the market logic.
What seems to be at stake in terms of the technological race we are witnessing is that without having an open and inclusive public conversation about smart-city development and surveillance implementations, we risk losing the democratic legitimacy of these transformations. As one of the few articles points out: The evolution of government surveillance is bold, multi-faceted, and confusing (EUobserver, 23 October 2019).
It means that different agencies are deploying a myriad of new technologies, including AI, FR and machine learning, to advance biometric surveillance but also master big data to allow for wider control.
That is why it is even more crucial for the European journals to focus on something more than Chinese ‘horror stories’, or the naive belief that the GDPR or other EU regulations will be a quick fix to privacy concerns, the rise of the surveillance state or lack of democratic control over implemented algorithms. Few of the articles point out how the public turns a blind eye to the fact that Europe is already a testing ground for many doubtful technologies, confirming Zuboff’s (2019) warnings. Even though EU citizens are legally protected from AI misuse, journalists and watchdog organisations point out that morally questionable and imperfect AI systems such as FR, fraud detection or surveillance cities are already in use across Europe.
Such an approach goes against the growing body of critical writing concerning smart cities (Cardullo and Kitchin, 2019; Swyngedouw, 2016), the roles of private and public entities in bringing innovation to the urban realm (Taylor Buck and While, 2017) and citizens’ rights in the digital society (Foth et al., 2015; Shaw and Graham, 2017). As Dean (2017) points out, improving the city for everyone begins with ensuring technology enhances democracy, by focusing on citizens’ right to use technology to address their needs, shape urban spaces and benefit from urban data.
Footnotes
Acknowledgements
The authors would like to thank Jan Zając (CEO of SoTrender) for providing us with Facebook data for this study.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work of Mikołaj Biesaga, Szymon Talaga and Andrzej Nowak was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 952026). The work of Anna Domaradzka was supported by the Polish National Science Centre under grant agreement no. 2018/30/E/HS6/00379. The work of Magdalena Roszczyńska-Kurasińska was supported by the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 960366). The methodology and the tool used in the study were developed under the project supported by the European Commission (grant agreement no. SMART 2017/0090 142-291425).
