Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Humanities and Social Sciences Communications volume 12, Article number: 24 (2025)
Metrics details
This study aims to understand how widely used Artificial Intelligence (AI) tools reflect the cultural context through the built environment. This research explores how outputs obtained with ChatGPT-4o, Midjourney’s bot on Discord and Google Maps represent the cultural context of Stockholm, Sweden. Cultural context is important because it shapes people’s identity, behaviour, and power dynamics. AI-generated recommendations and images of Stockholm’s cultural context were compared with real photographs, GIS demographic data and socio-economic information about the city. Results show how outputs written with ChatGPT-4o mostly listed museums and other venues popular among visitors, while Midjourney’s bot mostly represented cafes, streets, and furniture, reflecting a cultural context heavily shaped by buildings, consumption and commercial interests. Google Maps shows commercial sites while also enabling users to directly add information about places, like opinions, photographs and the main features of a business. These AI perspectives on cultural context can broaden the understanding of the urban environment and facilitate a deeper insight into the prevailing ideas behind the data that train these algorithms. Results suggest that the generative AI systems analysed convey a narrow view of the cultural context, prioritising buildings and a sense of cultural context that is curated, exhibited and commercialised. Generative AI tools could jeopardise cultural diversity by prioritising some ideas and places as “cultural”, exacerbating power relationships and even aggravating segregation. Consequently, public institutions should promote further discussion and research on AI tools, and help users combine AI tools with other forms of knowledge. The providers of AI systems should ensure more inclusivity in AI training data, facilitate users’ writing of prompts and disclose the limitations of their data sources. Despite the current potential reduction of diversity of the cultural context, AI providers have a unique opportunity to produce more nuanced outputs, which promote more societal diversity and equality.
This study aims to understand how widely used Artificial Intelligence (AI) tools reflect the cultural context through the built environment. This research explores how AI tools interpret and represent the cultural context of Stockholm, Sweden, comparing AI-generated recommendations and images of Stockholm’s cultural context with real photographs and demographic information about the city. This research critically examines the impact of AI on societies and equality, particularly through its influence on urban spaces. This study explores the consequences for society, urban space, and social justice of the perspectives on cultural context conveyed by mainstream AI tools.
The research question is: how is cultural context represented by mainstream AI tools and what are their consequences for society, urban space, and social justice? To explore this question, datasets on cultural context from interactions with ChatGPT-4o and Midjourney’s bot on Discord were compared and analysed with Google Maps, and with photographs of Stockholm and Geographic Information System statistical geolocated population data.
Cultural context affects communication (Marion, 2017), and can facilitate social diversity, and inclusion. In terms of architecture, appreciating cultural context is key to a better understanding of the needs and motivations of people inhabiting a space. The terms ‘context’, ‘contextual’, and ‘contextualism’ became important to the critics of the early modern movement in architecture after the publication of an article by Ernesto Rogers, in which the Italian word ambiente was translated into the English language as “context” (Forty, 2000, pp. 132–135). The word context comes from Latin con texere meaning weaving together and it aligns with the idea that people and places are all interconnected in multiple ways. The relationship with the physical environment defines people’s identities, for example, the place where people belong (Peng et al. 2020). Culture can affect how people behave and include others (Adobor, 2021). It can be understood as a selection of frames of reality, which can help identify problems and their causes, and can also be a form of exerting power (McNealy, 2021). The concept of interculturality needs to be redefined constantly because it is often connected to a dominating ideology (Dervin, 2023). Cultural contexts of individuals influence their thoughts, well-being, values, lives and experiences (Parker et al. 2020). Stereotypes, prejudice, and discrimination in a diverse social group can erode social cohesion and enhance divisions within societies, whereas, by contrast, diversity can foster creativity (Khan, 2024b).
Thus, the current analysis aims to contribute to deeper insights into the cultural context by examining how increasingly used AI tools reflect the cultural context of a city, determine what contextual actors are highlighted and which are neglected, understand power structures and explore opportunities to enhance social cohesion, diversity and equality. Reducing inequalities is one of the United Nations’ Sustainable Development Goals.
While AI has been part of everyday life for the last decade, influencing work, homes, and social relationships (Elliott, 2018), recent AI systems are a turning point with their human-like production of conversations and predictions and their extended usability beyond expert users.
Before AI, many visitors explored cities with printed guides, city maps, information from tourist offices, guided tours, and personal recommendations. Locals moved across the city based on previous knowledge and recommendations from their network. However, the amount of data collected in cities through sensors and reviews is now enormous and AI tools can help us better understand the places we inhabit (Ullah et al. 2020). TripAdvisor and other social networks have transformed and changed the way travellers and tourists seek, find, and share information from online reviews rather than tourism providers (Ali et al. 2021). Additionally, many AI tools are readily available to many Internet users and can provide data about a place. ChatGPT attracted 100 million users in the first two months after its release in November 2022 by OpenAI (Anon, 2023). ChatGPT has been rapidly adopted in different industries such as education, healthcare, and entertainment (Gupta et al. 2024).
Therefore, the new context in cities includes historical and personal information, and AI recommendations, based on undisclosed parameters. Widely used AI tools may exert a significant influence on users’ choices of where to go and how they experience a place, the routes they choose, or the way they illustrate their works. With AI, the new context includes many curated inputs from previous users who have expressed their opinions and with whom not everyone might relate, which act as big filters of reality through their recommendations and images.
According to the OECD, an AI system is “a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments” (OECD, 2019). The European Commission defines AI as follows:
“AI system’ means a machine-based system that is designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment, and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments (The European Parliament and the Council of the European Union, 2024)’.
Generative AI models are a subset of AI that can generate new data such as the patterns, structures, and features of the data on which they were trained (Khan, 2024a) and according to user requirements. Some limitations of generative AI models include some misleading outputs with severe social consequences (Khan, 2024a) and environmental impact (George et al. 2023). In the same AI Act, the European Parliament states that:
‘If the AI system is not trained with high-quality data, does not meet adequate requirements in terms of its performance, its accuracy or robustness, or is not properly designed and tested before being put on the market or otherwise put into service, it may single out people in a discriminatory or otherwise incorrect or unjust manner (The European Parliament and the Council of the European Union, 2024)’.
This new context with AI also includes a global struggle for control of these influential tools. Tech giants like Alphabet (Google), Apple, Meta (Facebook), Amazon, and Microsoft from the USA, and Baidu, Alibaba, and Tencent from China dominate and control the market with increasingly powerful AI tools (Verdegem, 2022). The Chinese government is aiming to become the largest AI development hub by 2030 (Mozur, 2017).
In the face of the rapid and powerful development of AI systems, over 30,000 people have called for this development to be paused, in order to plan and manage the risks to humankind (Future of Life Institute, 2023). Among the risks identified, the lack of transparency, whereby users cannot understand how AI systems work, jeopardises fairness, accountability (Lo Piano, 2020), reliability of information and the organisations that produce it (von Eschenbach, 2021).
After Alan Turing questioned whether machines could think, the 1956 Dartmouth Summer Research Project was widely accepted as the birthplace of AI (OECD, 2019). Research on AI and culture dates back to the beginnings of AI and the explorations of the relationship between the human mind and algorithms (Newell and Simon, 1976; Searle, 1980; Dreyfus and Dreyfus, 1988; Churchland and Churchland, 1990). In the 1970s, Weizenbaum argued that people should not be regarded as mere information-processing machines, against the metaphor of machine dominance, and that AI should be understood within its cultural context (Weizenbaum 1984). In the 1990s, Richard Wallace developed the Artificial Linguistic Internet Computer Entity, which could interact with humans in simple conversation, and IBM’s Deep Blue defeated chess champion Gary Kasparov (OECD, 2019).
Today, urban AI focuses on the relationship between AI systems and urban contexts, whereby AI can affect the governance of cities and their relationship with other areas (Luusua et al. 2023). Urban AI comprises different AI systems that affect cities and by which cities also affect AI systems, where citizens could even play a marginal role (Cugurullo et al. 2023).
AI systems can help analyse extensive datasets that affect urban planning, such as demographics, land use, and traffic (Peng et al. 2023). AI tools could help urban planners analyse multiple parameters simultaneously and mitigate inequalities (Lyu et al. 2024). AI can also help promote the accessibility of urban data for greater participatory governance and justice (Howe et al. 2022).
However, the ubiquitous and constant AI systems that analyse people, traffic, and other phenomena in cities could also be viewed as a centralised and omnipresent system that controls citizens (Crawford, 2021), or a “Polyopticon”, a network of sensors and algorithms with multiple perspectives under continuous observation (Bratton, 2021; Sherman, 2023). However, AI systems in the urban context could be understood as an assemblage of distributed systems from heterogenous stakeholders and in continuous transformation (Tseng, 2023). In fact, in some of the so-called smart cities, a holistic approach to AI systems is lacking (Cugurullo, 2021; Macrorie et al. 2021). Urban AI systems require a holistic understanding to be implemented with transparency, communicability, and visibility in the urban context (Popelka et al. 2023).
Beyond analysis, AI can influence life in cities at other levels and can affect urban form in unknown ways through technologies operating at landscape scale (Bratton, 2021). It can also affect the governance of cities through machines that behave autonomously (Cugurullo, 2020) and through systems that perform unsupervised managerial functions (Cugurullo et al. 2024). AI systems applied in the urban context could also increase disadvantages between cities (Macrorie et al. 2021).
Moreover, AI systems affect social diversity, equality, and inclusion (Cachat-Rosset and Klarsfeld, 2023). Sex and gender diversity are often at risk in AI applications (Fosch-Villaronga and Poulsen, 2022). Many AI systems extract average trends from enormous quantities of data and these averages may narrowly integrate the diversity of people, perpetuating stereotypes and biases (Sloan and Warner, 2020). AI can even portray migrants as a security threat (Bircan and Korkmaz, 2021). Recommendations powered by AI from social media could isolate users in ‘filter bubbles’ of virtual groups of people with similar affinities (Pariser, 2011), contribute to the dissemination of fake news (Rhodes, 2022), add new biases to existing ones (Bozdag, 2013), limit information diversity (Sun et al. 2022), and exacerbate political polarisation (Geschke et al. 2019). However, users’ choices are still the main driver of engagement with partisan and unreliable news on Google Search rather than algorithmic recommendations (Robertson et al. 2023).
In this study, cultural context is considered through multiple interconnected dimensions, including education, employment, financial resources, political empowerment, gender, and space. Mironenko and Sorokin (2018) define culture as follows:
‘Culture includes material objects: artefacts, ranging from planetary scale to jewellery. There are also spiritual, non-material components: languages, literature, art, science, etc. Culture also includes processes: external – social, ranging from individual to collective modes of behaviour (for instance, relating to constantly emerging and changing customs and traditions); and internal – individually-psychic’.
Cultural context is closely related to place identity or the characteristics that distinguish one place from another and affect human–environment interactions (Jang et al. 2023).
As regards culture, AI systems can mismatch the cultural norms and expectations of target cultural groups and produce cultural barriers by imposing hegemonic classifications, offensive or toxic settings, violating cultural values that are important for a group, and by omitting, trivialising, or simplifying certain identities and histories (Prabhakaran et al. 2022).
Therefore, AI systems and their implementation at urban level by local authorities require urgent further research (Yigitcanlar et al. 2021). Additional research on AI could help urban management and development processes (Son et al. 2023). Urban AI systems should be combined with other forms of knowledge while contemplating social and environmental contexts (Palmini and Cugurullo, 2023). There is a relative lack of interdisciplinarity connecting experts on AI, architecture, and civil society (Tseng, 2023). Combining urban AI with real-life needs and practical applications may accelerate the use and benefits of AI in urban planning (Sanchez et al. 2023).
Stockholm city centre and closest municipalities are a suitable testbed to study how tools with AI reflect the cultural context through a combination of social equality and digitalisation. Sweden has led changes toward greater social equality internationally (Kent, 2008) and has become an international reference in this area, despite persisting inequalities (Henning et al. 2023). Women acquired the right to vote in Sweden in 1921. Between the 1930s and 1960s, Swedish authorities launched initiatives for granting individuals greater economic independence. Buildings like new day-care centres, libraries, and others were a step toward greater individual freedom and supported this social change. After the 1960s, public housing and public buildings supported further equal opportunities for all members of society. Sweden also aims to rank among the pioneering states of digital transformation and AI (Rönnblom et al. 2023; Berman et al. 2024).
By bridging the knowledge gaps, this research innovatively analyses key aspects of the cultural context of Stockholm reflected by AI tools, along with their mismatches and omissions, to address the implications for social justice and architecture and the opportunities AI brings.
This study focuses on the possibilities offered by AI to relate to different cultural contexts through the built environment and its potential impact on social equality and diversity.
Four datasets were developed. Inclusion criteria for the AI systems tools selected were that they included AI, were widely used, and could potentially influence how people engage with urban space. ChatGPT-4o, Midjourney’s bot in Discord, and Google Maps were used.
ChatGPT is accessible to anyone with an Internet connection and basic knowledge of how to use a computer or smartphone and is growing exponentially as a search tool (Raile, 2024). Midjourney is a generative AI tool that produces images from natural language descriptions, called prompts, through a bot on Discord. Midjourney’s bot can produce photorealistic images or images in a specific style suggested by the user.
Google Maps is a leader among navigation apps, with six times more usage than the next most used (Haria et al. 2019) and billions of users worldwide (Phuangsuwan et al. 2024). Google Maps has led the way among navigation systems, with user-friendly advanced technologies, including satellite imagery, aerial photography, street maps, 360-degree panoramic views, street view, turn-by-turn instructions, public transit schedules and retrieving data from users. Google Maps is therefore an assemblage of sources of information and technologies in continuous change (McQuire, 2019).
The scope of this research to examine widely used AI tools and discuss findings.
The first dataset originated from interacting with ChatGPT-4o and prompting the following (18 September 2024): ‘List ten places in Stockholm that reflect the city’s cultural context’. The prompt was limited to ten bullet points to ensure the consistency of the data gathered from the prompts. The output obtained with ChatGPT-4o can thus be compared and analysed (see Table 1). ChatGPT-4o was then prompted to extend the scope of the answer: ‘List ten places that reflect the cultural context in Stockholm city and its suburbs’ (see Table 2).
ChatGPT-4o was also prompted ‘Recommend ten top cultural activities to do in Stockholm’ (21 September 2024). Responses written with ChatGPT-4o were given in the following order: visit the Vasa Museum, Gamla Stan, the Royal Palace, Nobel Prize Museum, Skansen Open-Air Museum, Fotografiska, ABBA The Museum, the City Hall, Moderna Museet, and Millesgården. When the user identified as a man, output obtained with ChatGPT-4o listed: Vasa Museum, ABBA The Museum, Viking Museum, Fotografiska Museum, Gamla Stan Pub Tour, Military History Museum, Drottningholm Palace, Skansen Open-Air Museum, Ericsson Globe Skyview and the Nobel Prize Museum. When the user identified as a woman, the outputs obtained with ChatGPT-4 listed: The Vasa Museum, Fotografiska, ABBA The Museum, Gamla Stan, Skansen Open-Air Museum, Rosendals Trädgård, Nordic Museum, Millesgården, Nobel Prize Museum and Art Galleries in Södermalm.
A second dataset was obtained by interacting with Midjourney’s bot (Figs. 1–5). The prompt given to Midjourney’s bot was: ‘A place in Stockholm that reflects the city’s cultural context’. Ten prompts to Midjourney’s bot resulted in 40 images.
Images generated with Midjourney’s bot on Discord from the prompt: ‘a place in Stockholm that reflects the city’s cultural context’ on 21 September 2024, first and second prompts.
Images generated with Midjourney’s bot on Discord from the prompt: ‘a place in Stockholm that reflects the city’s cultural context’ on 21 September 2024, third and fourth prompts.
Images generated with Midjourney’s bot on Discord from the prompt: ‘a place in Stockholm that reflects the city’s cultural context’ on 21 September 2024, fifth and sixth prompts.
Images generated with Midjourney’s bot on Discord from the prompt: ‘a place in Stockholm that reflects the city’s cultural context’ on 21 September 2024, seventh and eighth prompts.
Images generated with Midjourney’s bot on Discord from the prompt: ‘a place in Stockholm that reflects the city’s cultural context’ on 21 September 2024, ninth and tenth prompts.
The third dataset analysed is a map (Fig. 6) created with data from © Statistics Sweden on the distribution of the population born in Sweden in 2020, with map lines from © Lantmäteriet. The map also includes the geographical location of the places obtained from interacting with ChatGPT-4o from the first dataset as representing the cultural context. Additionally, the author has added the locations of the main museums, cultural centres, and libraries in Stockholm’s city centre and nearby suburbs to the map. This map shows that the population living in most of Stockholm is a mixture of nationalities.
The places obtained from interacting with ChatGPT, which reflect the cultural context of Stockholm, are mostly located in the city centre. Actually, there are more landmark museums, cultural centres, and libraries (represented with blue stars) in and around Stockholm city centre than in the suburbs. Map 1:100 000 with vector data on the property map, the overview map, and place names (fastighetskartan, översiktskartan, ortnamn) from © Lantmäteriet. Source of distribution of population born in Sweden in 2020 © Statistics Sweden. Source of location of museums, cultural centres, and libraries: Ingrid Campo-Ruiz. Source of location of the places suggested during interaction with ChatGPT-4o: Ingrid Campo-Ruiz.
The fourth and fifth datasets comprised photographs taken around Stockholm by the author to illustrate the diversity of buildings and urban space and to show the places listed in the interactions with ChatGPT-4o respectively (Figs. 7 and 8). ChatGPT was excluded as a tool for providing routes because the obtained recommendations were nonsensical. Navigation research with AI was limited to Google Maps.
Photos of different urban settings in Stockholm city centre: a Stockholm City Library, b Subway station Stadion, c Stockholm Woodland Cemetery, d Filmhuset, e KTH Royal Institute of Technology, f Ladugårdsgärdet Park, g Skeppsholms Bridge, h Stadion (photographer: Ingrid Campo-Ruiz). People in the photographs have been blurred using Photoshop to protect their privacy in compliance with GDPR.
Photos of places in Stockholm that reflect the cultural context as listed in the outputs from interaction with ChatGPT-4o on 18 September 2024: a Gamla Stan, b Skansen, c Vasa Museum, d ABBA The Museum, e Moderna Museet, f The City Hall, g Fotografiska, h Drottningholm Palace (photographer: Ingrid Campo-Ruiz). People in the photographs have been blurred using Photoshop to protect their privacy in compliance with GDPR.
In late 2022, OpenAI, based in San Francisco, California, released the first public version of ChatGPT (GPT meaning Generative pre-trained Transformer). The model was then updated to ChatGPT-4 and GPT-4o, March 2023 and May 2024, respectively. OpenAI (49% is owned by Microsoft) describes itself as ‘an AI research and deployment company. Our mission is to ensure that artificial general intelligence benefits all of humanity’ (OpenAI, 2023).
ChatGPT has amazed many: it understands and remembers interactions with humans, generates text, codes, reasons (Wu et al. 2023), summarises, interacts through voice, and detects emotional tone. While Open AI has not disclosed the number of parameters with which they work, estimations point to over 1.8 trillion (Howarth, 2024). ChatGPT has also concerned many with its shortcomings, including factual errors, lack of accurate information, high computing costs (Wu et al. 2023) and environmental impact (George et al. 2023).
The elements obtained interacting with ChatGPT-4o from the prompt ‘List ten places in Stockholm that reflect the city’s cultural context’ were six museums, the old city centre, the City Hall, the Royal Palace Drottningholm, and the southern district of Södermalm (Table 1). These places are all located in the city centre, except for Drottningholm, which is in the suburbs (Fig. 8). The places listed in the first interaction with ChatGPT-4o reflected a cultural scene mainly based on museums in the city centre. Two of the suggested places are districts, Gamla Stan and Södermalm, and the other places are buildings or an open-air museum in a group of buildings, Skansen.
The places listed in the second interaction with ChatGPT-4o, including the word suburbs in the prompt ‘List ten places that reflect the cultural context in Stockholm city and suburbs’, reduced the number of museums to four, and included the old city centre, Sergel Square, Kulturhuset cultural centre, the Royal Palace, a botanical garden, a butterfly house, and Tensta exhibition hall (Table 2). The listed places are primarily buildings.
The map with geolocated data (Fig. 6) shows a high concentration of museums, libraries and exhibition halls in the city centre. In the suburbs of Stockholm, a broad range of cultural buildings also enhance education, employment opportunities, and health care services—libraries like Rinkeby-Kista Bibliotek (2004) are home to cultural events and exhibitions. The list of places obtained in the second interaction with ChatGPT-4o correctly highlights another of these suburban public buildings, Tensta Art Centre, which reflects Stockholm’s multicultural fabric.
People from around 180 different countries live in Stockholm (Mahmud, 2013). In the 1950s and 1970s, Sweden needed workers for its expanding industry, attracting immigrants from the Nordic and southern European countries. Between the mid-1970s and 1990s, immigrants came from Chile, Bosnia, and Somalia (Murdie and Borgegård, 1998). Stockholm has also attracted many people from different areas of Sweden.
Some of the Stockholm suburbs are home to lower-income households and a larger population of non-European origin (Andersson and Bråmå, 2018). Stockholm registers higher segregation and polarisation compared to other European capitals (Musterd et al. 2017; Haandrikman et al. 2023). Some Swedish newspapers establish a connection between immigrants and “un-Swedish values”, framing them as a threat to gender equality and the welfare system (Norocel et al. 2020, pp. 101–103). In parallel, Sweden has reduced its immigration policies to meet the European Union minimum, reflecting a broader trend of political polarisation in Europe (Borbáth et al. 2023).
Therefore, the multicultural nature of Stockholm is broader than the idea of cultural context obtained from interacting with ChatGPT-4o. The list of places obtained from the first interaction with ChatGPT-4o describes a limited form of culture, which prioritises buildings, and among them popular museums. When prompted about places in the city centre and suburbs, the list obtained with ChatGPT-4o extended the scope of cultural context beyond museums, exhibition centres and popular districts, to also include a square and two parks. Among the results obtained with ChatGPT-4o, buildings in the city centre—many of which are museums with an entrance fee—are prevalent and results give a sense of cultural context that is curated, exhibited, and commercialised. Importantly, the resulting list of places written with ChatGPT-4o suggested different places when requested to ‘Recommend ten top cultural activities to do in Stockholm’, if the user identifies as a man or as a woman.
If huge numbers of people follow these texts written with ChatGPT-4o as if they were recommendations, some areas will be increasingly neglected, whereas others will become more attractive to visitors and economically powerful and could potentially decrease diversity.
Midjourney Inc., based in San Francisco, California, which launched its open beta on 12 July 2022, describes itself as:
an independent research lab exploring new mediums of thought and expanding the imaginative powers of the human species. We are a small self-funded team focused on design, human infrastructure, and AI. We have 11 full-time staff and an incredible set of advisers (Midjourney, 2024).
Midjourney’s bot on Discord is trained to create images with associated words. The results obtained from Midjourney’s bot depend heavily on the semantics of the prompts (Tan and Luhrs, 2024).
To the best of available knowledge, the data used to train Midjourney’s bot is undisclosed. However, AI is generally trained on data that has been provided manually, from structured databases, data streams such as sensors, application programming interfaces that obtain data from other platforms, the Internet, and through interaction with users (Khan, 2024a).
From the prompt ‘a place in Stockholm that reflects the city’s cultural context’, the images generated with Midjourney’s bot mostly showed street-level cafes, with wooden furniture, in ochre, yellow, and grey mortar-rendered buildings, and cobblestone streets. The images produced with Midjourney’s bot predominantly and realistically depict cafés (20 out of 40), settings near water (11 out of 40), a staircase that could be part of an urban park, and artistic works such as a stained-glass façade, urban sculptures, and a frescoed vault. One of the images even resembles one of the colourful grottos that form part of the subway stations in Stockholm.
The images produced with Midjourney’s bot predominantly show buildings or other urban design elements, such as waterfronts or a staircase. Midjourney’s bot’s images also feature furniture and decorative elements, which are important aspects of culture. These images blend contemporary and 17th-18th-century-like Stockholm architecture, similar to the old town, Gamla Stan, with some images featuring water, reflecting the city’s archipelago setting. Overall, the photos generated with Midjourney’s bot highlight places like cafes that connect culture with consumption and economic activity.
A comparison of the author’s photos (Fig. 7) with Midjourney’s bot-generated images (Figs. 1–5) reveals that the bot’s images resemble diverse urban areas across Stockholm’s city centre, albeit without the diversity of Stockholm’s architecture and urban space built across the centuries. The images obtained with Midjourney’s bot do not show iconic landmarks like the Royal Palace, Concert Hall (1926), City Library (1928), Kulturhuset (1974), Vasa Museum (1990), or the new building of Moderna Museet (1998), nor key places like Sergel Square (1967), urban parks, or Million Programme housing blocks (1965–1974).
Culture goes beyond architecture. However, Midjourney’s images overlook other places that reflect other forms of cultural context like theatre, dance, music, or street markets. Science, technology and sustainability, still deeply rooted in Stockholm through its world-famous innovative organizations, are poorly represented with few contemporary-like constructions. Furthermore, the images obtained with Midjourney’s bot show only modern or well-maintained architectural settings, omitting other architectural settings such as housing blocks from the 1950s to the 1980s.
Google Maps was launched in 2005. In 2004, Google had acquired a start-up from Danish-born brothers Lars and Jens Rasmussen, which targeted rival mapping services of the time (McQuire, 2019). Google Maps attracted ten million views when it was launched, providing users with the unprecedented experience of exploring maps without having to refresh the image (McQuire, 2019). Google combines its master map with aerial and satellite imagery and data harvested from users, combining GPS, sensors, and voluntary online reviews and conversations (McQuire, 2019).
For Google Maps, providing maps as a service generates advertising revenue and retrieves user data, which are essential for keeping the map updated (Heckmann et al. 2022). AI is also used to recommend information to users based on sophisticated analysis of their behaviours, including text and images they have viewed and their drifting interests (Zhang et al. 2021). In 2009, Google began personalised searches aimed at predicting user preferences (Pariser, 2011). Big Techs usually commercialise user information such as location, preferences, and searches (Verdegem, 2022). Algorithms that issue recommendations for users are profitable for Big Techs that provide media content as they maximize the time users engage with their content (Geschke et al. 2019).
Google Maps uses the A* algorithm to find the shortest path between two points (Mehta et al. 2019). The A* algorithm is a search engine used within AI systems, but it does not learn over time. Already in 2009, Google began personalised searches to predict user preferences (Pariser, 2011).
Google Maps limits users’ queries to date, time, and means of transport. It does not address many other parameters that could influence the decision of a route, even though technology is already able to support considering other person-dependent parameters (Heckmann et al. 2022). Although Google Maps’ suggested routes are the fastest, other aspects of urban space, such as aesthetics, sounds, smells, safety, or comfort may be neglected. However, Google Maps has expanded user knowledge of the urban context and set up a system whereby anyone with access to the Internet and technology skills can add information to their map. While activities of people without access to the Internet might easily be left out of Google Maps, the platform has been instrumental for many in learning about, accessing, and engaging with the surrounding urban space, traffic, and activities. Google Maps has enhanced the visibility of local businesses, facilitated consumer engagement through reviews, and increased foot traffic, thereby contributing to economic growth and development (Phuangsuwan et al. 2024).
The outputs obtained with ChatGPT-4o on cultural context show a realistic list of venues which are particularly interesting for visitors to Stockholm. The outputs from Midjourney’s bot are photorealistic images of places that could well be in Stockholm’s city centre because of the colours, materials, and atmospheric light. Google Maps provides accurate information about the city with multiple views of streets, timing of traffic, public transportation, and possible routes.
The outputs on cultural context obtained from interaction with ChatGPT-4o focused on popular venues for visitors, while Midjourney focused on cafes, streets, and everyday objects like chairs and tables, reflecting a cultural context heavily shaped by consumption and commercial interests. The cultural context of Stockholm, reflected in the places listed with ChatGPT-4o and represented by Midjourney’s bot, is rather narrow compared to the city’s extensive multiculturalism. Google Maps also shows commercial activities, while enabling active participation from users who can add information to their main map, for example, their opinions. Therefore, Google Maps reflects a more diverse cultural context.
Reasons for ChatGPT-4o and Midjourney’s bot’s specific views on cultural context may be shaped by their training data. AI systems curate datasets, establishing the parameters and hierarchies for their outputs. If their training data is taken from the Internet and interactions with users, their output could therefore potentially overlook the perspectives of those less engaged with the Internet and these AI tools, including older persons, introverts, privacy-conscious persons, and those with limited time or lower incomes. Large amounts of statistical data are processed by a few actors, while citizens’ individual perspectives and preferences become less relevant (Sareen et al. 2020). Data contributors to these AI tools do not necessarily represent most societies or their minorities. By following AI recommendations, people may enhance the opinions of the majority of data contributors.
Shortcomings of Google Maps’ participative map might include the fact that some users may find it challenging to contribute —senior citizens, people with visual or cognitive disabilities, people lacking time to engage due to demanding work schedules or personal family commitments, or people who do not possess updated smartphones. Most educated persons can engage more easily with information and communication technologies, while older people avoid sharing online and often show little interest (Elena-Bucea et al. 2021).
Many are very excited about the short-term impact of AI, though predicting the long-term societal effects of generative AI is rather unforeseeable (Sabherwal and Grover, 2024). The outputs analysed from AI systems have important implications in the built environment at social, economic, and political levels. AI systems can:
Broaden the understanding of context. AI systems can augment our knowledge of the urban space we inhabit.
Facilitate a deeper insight into prevailing ideas of culture on the data that train these algorithms. If the Internet and interactions with users are the data sources of these AI tools, then these outputs also provide an average representation of the unfathomable number of images online and show a substantial Internet bias regarding culture. If ChatGPT and Midjourney disclosed their data sources, users could identify potential societal omissions and neglect of these data, and key concentrations of power, and further understand the multiple parameters that describe reality.
Redefine culture. The output analysed from interactions with ChatGPT-4o and Midjourney’s bot privileges perspectives on a culture that enhances buildings, mainly museums or cafes, economic and consumption perspectives over other activities that are also part of local culture.
Jeopardise cultural diversity that represents different members of society. The outputs from interactions with AI show only a limited aspect of the city’s complex cultural diversity, prioritising buildings, museums, and cafes as “cultural” over others and conveying a sense of cultural context that is curated, exhibited, and commercialised. These outputs could redefine people’s ideas on culture, by disseminating ideologies about what culture is, empowering certain perspectives, neglecting other views, and decreasing diversity. Furthermore, rendering more visibility to few commercial buildings would contribute to the increasing commercialisation of cultural space in the city and inequalities (Campo-Ruiz, 2024).
Exacerbate power relationships in the built environment that perpetuate a limited set of ideas. The places obtained in the interactions with ChatGPT-4o and Midjourney’s bot that reflect the cultural context in Stockholm predominantly benefit a few organisations while disregarding other forms of cultural expression, potentially increasing inequalities. Those excluded from the conversation might feel even more disengaged. Neocolonialism has been described as a way to control a territory and influence its people, with psychological domination and the denigration of local values (Flikschuh et al. 2015). From this perspective, some AI tools could contribute to establishing forms of urban space control limited to a few powerful actors, enhancing their interests, increasing the vulnerability of some, and multiplying inequalities.
Enhance segregation. The outputs obtained with ChatGPT-4o on cultural context are mostly located in the city centre, while Midjourney’s images also resemble places in Stockholm city centre, risking becoming yet another factor perpetuating segregation.
These findings highlight how AI is triggering rearrangements of social relationships and information pathways and transforming urban space from the perspective of urban environment and societal equality, by contributing an innovative and complementary view of the influence of AI in the urban context. Urban analysis has included multiple technologies like robots and self-driving cars that affect the built environment (Cugurullo et al. 2023). Generative AI, which is rapidly expanding, also plays a key role alongside other AI systems. While generative AI may not manage cities autonomously, it will significantly influence decisions concerning engaging and understanding urban space. Governments should avoid introducing further risks, such as discrimination and power imbalances, when implementing AI systems, by adhering to principles of equality, freedom, and human rights, while ensuring citizen involvement and open access to information (Duberry, 2022).
In light of these findings, the following actions are proposed to engage with AI for greater social justice through architecture:
Public organisations should promote further public discussion and research about these AI tools and their implications for people through the places they inhabit. Some of the abovementioned risks of narrowly redefining cultural context through AI tools could promote inequalities and segregation, challenging democratic and social justice principles in public space.
Public organisations should raise awareness of the limitations of some AI systems, which may neglect important aspects that shape the cultural context of a place. Many users constantly feed algorithms with their interactions and the data they upload, including comments, photos, purchases, and searches. Public authorities should empower users to critically consider who they are supporting and neglecting by following AI outputs.
Public organisations should facilitate users to combine AI outputs with other tools and knowledge such as humanistic, social, and ethical values. AI should not be the sole source of information about the world we live in.
The providers of AI systems should ensure more inclusivity in the data used to train their models and in the way they are trained, to achieve more nuanced outputs. In research, all data bear some degree of bias, and researchers are the curators of the information they manage. The same should apply to AI systems in the built environment because developers are responsible for curating data. AI and users exert reciprocal influence on each other, though users have less power in general (Airoldi and Rokka, 2022).
The providers of AI systems should disclose the limitations of their training data, users and the rest of society would better understand and address AI output bias.
All those who have the agency, skills, and time to use AI should help others interact with AI tools critically, understanding that not all texts are true, not all images are representative, and not all recommendations serve one’s best interests. While some research suggests that users change behaviours to avoid becoming victims (Zhang and Hu, 2023), many persons lack the agency, tools, time, or skills to be active players in shaping AI. AI systems can be valuable tools but are not accessible to everyone. Users can be activists to use AI for the greater good and to have a positive impact on the places they live in.
One limitation of this research is that the datasets refer to only one city as a case study. Further research is needed to determine broader, worldwide implications.
In conclusion, although AI has been around for decades, a new era is now beginning. A seismic revolution is taking place through interaction with AI systems, which have been made widely available and usable to many more people than only a few years ago. AI systems can aid in understanding of the complexities of the urban environment by helping users analyse quantities of data that are unfathomable to the human mind. The shortcomings of the generative AI models analysed could be considered reasonable: these tools are recent, yet can still provide jaw-dropping results. Despite the current potential reduction of cultural context diversity, AI providers have a unique opportunity to produce more nuanced outputs, which promote more societal diversity and equality. AI systems have a huge opportunity to make the world a better place, but some organisations have more power and responsibility than other members of society.
The datasets generated during the current study are included in the article. ChatGPT-4o is available at https://chat.openai.com/, Midjourney’s bot at https://discord.com/channels and the websites and apps have been referenced.
Adobor H (2021) Open strategy: what is the impact of national culture? Manag Res Rev. 44:1277–1297. https://doi.org/10.1108/MRR-06-2020-0334
Article MATH Google Scholar
Airoldi M, Rokka J (2022) Algorithmic consumer culture. Consum Mark. Cult. 25:411–428. https://doi.org/10.1080/10253866.2022.2084726
Article MATH Google Scholar
Ali T, Marc B, Omar B et al. (2021) Exploring destination’s negative e-reputation using aspect-based sentiment analysis approach: case of Marrakech destination on TripAdvisor. Tour. Manag Perspect. 40:100892. https://doi.org/10.1016/j.tmp.2021.100892
Article Google Scholar
Andersson R, Bråmå Å (2018) The Stockholm estates: a tale of the importance of initial conditions, macroeconomic dependencies, tenure and immigration. In: Hess D, Tammaru T, van Ham M (eds.) Housing estates in Europe, 1st edn. Springer, Cham, p 361–388
Anon (2023) How to worry wisely about artificial intelligence. The Economist. https://www.economist.com/leaders/2023/04/20/how-to-worry-wisely-about-artificial-intelligence
Berman A, de Fine Licht K, Carlsson V (2024) Trustworthy AI in the public sector: an empirical analysis of a Swedish labor market decision-support system. Technol. Soc. 76:102471. https://doi.org/10.1016/j.techsoc.2024.102471
Article Google Scholar
Bircan T, Korkmaz EE (2021) Big data for whose sake? Governing migration through artificial intelligence. Humanit Soc. Sci. Commun. 8:1–5. https://doi.org/10.1057/s41599-021-00910-x
Article MATH Google Scholar
Borbáth E, Hutter S, Leininger A (2023) Cleavage politics, polarisation and participation in Western Europe. West Eur. Polit. 0:1–21. https://doi.org/10.1080/01402382.2022.2161786
Article MATH Google Scholar
Bozdag E (2013) Bias in algorithmic filtering and personalization. Ethics Inf Technol 15:209–227. https://doi.org/10.1007/s10676-013-9321-6
Bratton B (2021) AI urbanism: a design framework for governance, program, and platform cognition. AI Soc. 36:1307–1312. https://doi.org/10.1007/s00146-020-01121-9
Article MATH Google Scholar
Cachat-Rosset G, Klarsfeld A (2023) Diversity, equity, and inclusion in Artificial Intelligence: An evaluation of guidelines. Appl Artif Intell 37. https://doi.org/10.1080/08839514.2023.2176618
Campo-Ruiz I (2024) Economic powers encompass the largest cultural buildings: market, culture and equality in Stockholm, Sweden (1918–2023). ArchNet-IJAR. https://doi.org/10.1108/ARCH-06-2023-0160
Churchland PM, Churchland PS (1990) Could a machine think? Sci. Am. Assoc. Adv. Sci. 262:32–39
CAS Google Scholar
Crawford K (2021) Atlas of AI: power, politics, and the planetary costs of artificial intelligence. Yale University Press, New Haven
Book MATH Google Scholar
Cugurullo F (2021) Frankenstein urbanism: eco, smart and autonomous cities, artificial intelligence and the end of the city. Routledge, Taylor & Francis Group, London
Book Google Scholar
Cugurullo F, Caprotti F, Cook M et al. (2024) The rise of AI urbanism in post-smart cities: a critical commentary on urban artificial intelligence. Urban Stud. 61:1168–1182. https://doi.org/10.1177/00420980231203386
Article MATH Google Scholar
Cugurullo F (2020) Urban artificial intelligence: from automation to autonomy in the smart city. Front Sustain Cities 2. https://doi.org/10.3389/frsc.2020.00038
Cugurullo F, Caprotti F, Cook M, Karvonen A, McGuirk P, Marvin S (eds.) (2023) Artificial intelligence and the city: urbanistic perspectives on AI, 1st edn. Taylor & Francis, London
Dervin F (2023) The paradoxes of interculturality: a toolbox of out-of-the-box ideas for intercultural communication education. Routledge, New York
MATH Google Scholar
Dreyfus HL, Dreyfus SE (1988) Making a mind versus modeling the brain: Artificial Intelligence back at a branchpoint. Daedalus Camb. Mass 117:15–43
MATH Google Scholar
Duberry J (2022) Artificial Intelligence and democracy: risks and promises of AI-mediated citizen-government relations. Edward Elgar Publishing, Northampton
Elena-Bucea A, Cruz-Jesus F, Oliveira T, Coelho PS (2021) Assessing the role of age, education, gender and income on the digital divide: evidence for the European Union. Inf. Syst. Front 23:1007–1021. https://doi.org/10.1007/s10796-020-10012-9
Article Google Scholar
Elliott A (2018) The Culture of AI: everyday life and the digital revolution, 1st edn. Routledge, Milton
MATH Google Scholar
von Eschenbach WJ (2021) Transparency and the black box problem: why we do not trust AI. Philos. Technol. 34:1607–1622. https://doi.org/10.1007/s13347-021-00477-0
Article MATH Google Scholar
European Parliament and the Council of the European Union (2024) Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act). https://eur-lex.europa.eu/eli/reg/2024/1689/oj. Accessed 20 Sep 2024
Flikschuh K, Ypi L, Ajei M (2015) Kant and colonialism: historical and critical perspectives. University Press, Oxford
Forty A (2000) Words and buildings: a vocabulary of modern architecture. Thames & Hudson, London
Fosch-Villaronga E, Poulsen A (2022) Diversity and inclusion in Artificial Intelligence. In: Custers B, Fosch-Villaronga E (eds) Law and Artificial Intelligence: regulating AI and applying AI in legal practice. T.M.C. Asser Press, The Hague, p 109–134
Future of Life Institute (2023) Pause giant AI experiments: an open letter. In: Future Life Inst. https://futureoflife.org/open-letter/pause-giant-ai-experiments/. Accessed 3 Sep 2024
George AS, George ASH, Martin ASG (2023) The environmental impact of AI: a case study of water consumption by Chat GPT. Part. Univers Int Innov. J. 1:97–104. https://doi.org/10.5281/zenodo.7855594
Article MATH Google Scholar
Geschke D, Lorenz J, Holtz P (2019) The triple-filter bubble: Using agent-based modelling to test a meta-theoretical framework for the emergence of filter bubbles and echo chambers. Br. J. Soc. Psychol. 58:129–149. https://doi.org/10.1111/bjso.12286
Article PubMed MATH Google Scholar
Gupta R, Nair K, Mishra M et al. (2024) Adoption and impacts of generative artificial intelligence: theoretical underpinnings and research agenda. Int J. Inf. Manag Data Insights 4:100232. https://doi.org/10.1016/j.jjimei.2024.100232
Article Google Scholar
Haandrikman K, Costa R, Malmberg B et al. (2023) Socio-economic segregation in European cities. A comparative study of Brussels, Copenhagen, Amsterdam, Oslo and Stockholm. Urban Geogr. 44:1–36. https://doi.org/10.1080/02723638.2021.1959778
Article Google Scholar
Haria V, Shah Y, Gangwar V et al. (2019) The working of Google Maps, and the commercial usage of navigation systems. IJIRT 6:184–191
MATH Google Scholar
Heckmann R, Kock S, Gaspers L (2022) Artificial intelligence supporting sustainable and individual mobility: development of an algorithm for mobility planning and choice of means of transport. In: Coors V, Pietruschka D, Zeitler B (eds) iCity. Transformative research for the livable, intelligent, and sustainable city. Springer International Publishing, Stuttgart, p 27–40
Henning M, Westlund H, Enflo K (2023) Urban–rural population changes and spatial inequalities in Sweden. Reg. Sci. Policy Pr. 15:878–892. https://doi.org/10.1111/rsp3.12602
Article MATH Google Scholar
Howarth J (2024) Number of parameters in GPT-4 (Latest Data). https://explodingtopics.com/blog/gpt-parameters. Accessed 8 Sep 2024
Howe B, Brown JM, Han B, Herman B, Weber N, Yan A, Yang S, Yang Y (2022) Integrative urban AI to expand coverage, access, and equity of urban data. Eur. Phys. J. Spec. Top. 231:1741–1752. https://doi.org/10.1140/epjs/s11734-022-00475-z
Article PubMed PubMed Central MATH Google Scholar
Jang KM, Chen J, Kang Y, Kim J, Lee J, Duarte F (2023) Understanding place identity with generative AI. Leibniz International Proceedings in Informatics, Leibniz, pp 1–6
Kent N (2008) A concise history of Sweden. Cambridge University Press, Cambridge
Book MATH Google Scholar
Khan I (2024a) The quick guide to prompt engineering. John Wiley & Sons, Newark
MATH Google Scholar
Khan S (2024b) Cultural diversity and social cohesion: perspectives from social science. Phys. Educ. Health Soc. Sci. 2:40–48
MATH Google Scholar
Lo Piano S (2020) Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward. Humanit Soc Sci Commun 7:1–7. https://doi.org/10.1057/s41599-020-0501-9
Luusua A, Ylipulli J, Foth M, Aurigi A (2023) Urban AI: understanding the emerging role of artificial intelligence in smart cities. AI Soc. 38:1039–1044. https://doi.org/10.1007/s00146-022-01537-5
Article PubMed Google Scholar
Lyu Y, Lu H, Lee MK, Schmitt G, Lim B (2024) IF-City: intelligible fair city planning to measure, explain and mitigate inequality. IEEE Trans. Vis. Comput Graph 30:3749–3766. https://doi.org/10.1109/TVCG.2023.3239909
Article PubMed Google Scholar
Macrorie R, Marvin S, While A (2021) Robotics and automation in the city: a research agenda. Urban Geogr. 42(2):197–217
Article MATH Google Scholar
Mahmud HJA (2013) Social cohesion in multicultural society: a case of Bangladeshi immigrants in Stockholm. Dissertation, Stockholm University
Marion G (2017) How Culture Affects Language and Dialogue. In: The Routledge Handbook of Language and Dialogue. Routledge, London
McNealy JE (2021) Framing and language of ethics: technology, persuasion, and cultural context. J. Soc. Comput 2:226–237. https://doi.org/10.23919/JSC.2021.0027
Article MATH Google Scholar
McQuire S (2019) One map to rule them all? Google Maps as digital technical object. Commun. Public 4:150–165. https://doi.org/10.1177/2057047319850192
Article Google Scholar
Mehta H, Kanani P, Lande P (2019) Google Maps. Int J. Comput. Appl 178:41–46. https://doi.org/10.5120/ijca2019918791
Article Google Scholar
Midjourney (2024) About. www.midjourney.com/home. Accessed 23 Oct 2024
Mironenko IA, Sorokin PS (2018) Seeking for the definition of “culture”: current concerns and their implications. a comment on Gustav Jahoda’s article “Critical reflections on some recent definitions of “culture’”. Integr. Psychol. Behav. Sci. 52:331–340. https://doi.org/10.1007/s12124-018-9425-y
Article PubMed MATH Google Scholar
Mozur P (2017) Beijing wants A.I. to be made in China by 2030. N. Y. Times. https://www.nytimes.com/2017/07/20/business/china-artificial-intelligence.html
Murdie RA, Borgegård L-E (1998) Immigration, spatial segregation and housing segmentation of immigrants in metropolitan Stockholm, 1960-95. Urban Stud. Edinb. Scotl. 35:1869–1888. https://doi.org/10.1080/0042098984196
Article Google Scholar
Musterd S, Marcińczak S, van Ham M, Tammaru T (2017) Socioeconomic segregation in European capital cities. Increasing separation between poor and rich. Urban Geogr. 38:1062–1083. https://doi.org/10.1080/02723638.2016.1228371
Article Google Scholar
Newell A, Simon HA (1976) Computer science as empirical inquiry: symbols and search. Commun. ACM 19:113–126. https://doi.org/10.1145/360018.360022
Article MathSciNet MATH Google Scholar
Norocel OC, Hellström A, Jørgensen MB (2020) Nostalgia and hope: intersections between politics of culture, welfare, and migration in Europe, 1st edn. Springer International Publishing, Cham
OECD (2019) Artificial Intelligence in Society. OECD Publishing, Paris. https://doi.org/10.1787/eedfee77-en. Accessed 23 Oct 2024
OpenAI (2023) https://openai.com/about. Accessed 20 Sep 2024
Palmini O, Cugurullo F (2023) Charting AI urbanism: conceptual sources and spatial implications of urban artificial intelligence. Discov Artif Intell 3. https://doi.org/10.1007/s44163-023-00060-w
Pariser E (2011) The filter bubble: what the Internet is hiding from you. Viking, London
Parker PD, Van Zanden B, Marsh HW, Owen K, Duineveld J, Noetel M (2020) The intersection of gender, social class, and cultural context: a meta-analysis. Educ. Psychol. Rev. 32:197–228. https://doi.org/10.1007/s10648-019-09493-1
Article Google Scholar
Peng J, Strijker D, Wu Q (2020) Place identity: how far have we come in exploring its meanings? Front Psychol 11. https://doi.org/10.3389/fpsyg.2020.00294
Peng Z-R, Lu K-F, Liu Y, Zhai W (2023) The pathway of urban planning AI: from planning support to plan-making. J Plan Educ Res 0. https://doi.org/10.1177/0739456X231180568
Phuangsuwan P, Siripipatthanakul S, Limna P, Pariwongkhuntorn N (2024) The impact of Google Maps application on the digital economy. Corp. Bus. Strategy Rev. 5:192–203. https://doi.org/10.22495/cbsrv5i1art18
Article Google Scholar
Popelka S, Narvaez Zertuche L, Beroche H (2023) Urban AI guide. Urban AI. https://urbanai.fr/
Prabhakaran V, Qadri R, Hutchinson B (2022) Cultural incongruencies in Artificial Intelligence. ArXiv Cornell Univ. https://doi.org/10.48550/arxiv.2211.13069
Raile P (2024) The usefulness of ChatGPT for psychotherapists and patients. Humanit Soc. Sci. Commun. 11:47–48. https://doi.org/10.1057/s41599-023-02567-0
Article Google Scholar
Rhodes SC (2022) Filter bubbles, echo chambers, and fake news: How social media conditions individuals to be less critical of political misinformation. Polit Commun 39:1–22. https://doi.org/10.1080/10584609.2021.1910887
Robertson RE, Green J, Ruck DJ, et al. (2023) Users choose to engage with more partisan news than they are exposed to on Google Search. Nature 618:342–348. https://doi.org/10.1038/s41586-023-06078-5
Rönnblom M, Carlsson V, Öjehag‐Pettersson A (2023) Gender equality in Swedish AI policies. What’s the problem represented to be? Rev. Policy Res 40:688–704. https://doi.org/10.1111/ropr.12547
Article Google Scholar
Sabherwal R, Grover V (2024) The societal impacts of generative Artificial Intelligence: a balanced perspective. J. Assoc. Inf. Syst. 25:13–22. https://doi.org/10.17705/1jais.00860
Article MATH Google Scholar
Sanchez TW, Shumway H, Gordner T, Lim T (2023) The prospects of Artificial Intelligence in urban planning. Int J. Urban Sci. 27:179–194. https://doi.org/10.1080/12265934.2022.2102538
Article MATH Google Scholar
Sareen S, Saltelli A, Rommetveit K (2020) Ethics of quantification: illumination, obfuscation and performative legitimation. Palgrave Commun 6. https://doi.org/10.1057/s41599-020-0396-5
Searle J (1980) Minds, brains and programs. In: Collins A, Smith EE (eds) Readings in Cognitive Science. Morgan Kaufmann, San Mateo, p 20–31
Sherman S (2023) The polyopticon: a diagram for urban Artificial Intelligences. AI Soc. 38:1209–1222. https://doi.org/10.1007/s00146-022-01501-3
Article MATH Google Scholar
Sloan RH, Warner R (2020) Beyond bias: Artificial Intelligence and social justice. Va J. Law Technol. 24:1
MATH Google Scholar
Son TH, Weedon Z, Yigitcanlar T et al. (2023) Algorithmic urban planning for smart and sustainable development: Systematic review of the literature. Sustain Cities Soc. 94:104562. https://doi.org/10.1016/j.scs.2023.104562
Article Google Scholar
Sun J, Song J, Jiang Y, et al. (2022) Prick the filter bubble: A novel cross domain recommendation model with adaptive diversity regularization. Electron Mark 32:101–121. https://doi.org/10.1007/s12525-021-00492-1
Tan L, Luhrs M (2024) Using generative AI Midjourney to enhance divergent and convergent thinking in an architect’s creative design process. Des. J. 27:677–699. https://doi.org/10.1080/14606925.2024.2353479
Article Google Scholar
Tseng Y-S (2023) Assemblage thinking as a methodology for studying urban AI phenomena. AI Soc. 38:1099–1110. https://doi.org/10.1007/s00146-022-01500-4
Article MATH Google Scholar
Ullah Z, Al-Turjman F, Mostarda L, Gagliardi R (2020) Applications of Artificial Intelligence and Machine Learning in smart cities. Comput Commun. 154:313–323. https://doi.org/10.1016/j.comcom.2020.02.069
Article MATH Google Scholar
Verdegem P (2022) Dismantling AI capitalism: the commons as an alternative to the power concentration of Big Tech. AI Soc 1–11. https://doi.org/10.1007/s00146-022-01437-8
Weizenbaum J (1984) Computer power and human reason: from judgment to calculation. Penguin, Harmondsworth
MATH Google Scholar
Wu T, He S, Liu J et al. (2023) A brief overview of ChatGPT: The history, status quo and potential future development. IEEECAA J. Autom. Sin. 10:1122–1136. https://doi.org/10.1109/JAS.2023.123618
Article MATH Google Scholar
Yigitcanlar T, Corchado JM, Mehmood R, Li RYM, Mossberger K, Desouza KC (2021) Responsible urban innovation with local government Artificial Intelligence (AI): A conceptual framework and research agenda. J. Open Innov. 7:1–16. https://doi.org/10.3390/joitmc7010071
Article Google Scholar
Zhang Q, Lu J, Jin Y (2021) Artificial Intelligence in recommender systems. Complex Intell. Syst. 7:439–457. https://doi.org/10.1007/s40747-020-00212-w
Article MATH Google Scholar
Zhang J, Hu C (2023) Adaptive algorithms: users must be more vigilant. Nat Lond 618:907–907. https://doi.org/10.1038/d41586-023-02033-6
Download references
This project has received funding from the European Union’s Horizon 2020 Research and Innovation Programme under the European Commission Marie Skłodowska-Curie Actions Individual Fellowship Grant Number 101032687 — EQUBUILD.
KTH Royal Institute of Technology, ABE School of Architecture and the Built Environment, School of Architecture, Stockholm, Sweden
Ingrid Campo-Ruiz
You can also search for this author in PubMed Google Scholar
Correspondence to Ingrid Campo-Ruiz.
The author declares no competing interests.
This article contains no studies with human participants undertaken by the author.
This article contains no studies with human participants undertaken by any of the authors.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Campo-Ruiz, I. Artificial intelligence may affect diversity: architecture and cultural context reflected through ChatGPT, Midjourney, and Google Maps. Humanit Soc Sci Commun 12, 24 (2025). https://doi.org/10.1057/s41599-024-03968-5
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1057/s41599-024-03968-5
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
© 2025 Springer Nature Limited
