Inhaltszusammenfassung

With the growing popularity of generative artificial intelligence, place-based narratives of cities produced by large language models are gaining traction. However, it remains unclear how such algorithmically developed representations relate to human perceptions and experiences of urban space. Against this background, this paper compares ‘typical places’ generated by large language models with data collected through participatory mapping by university students, using the city of Mainz as a ca...With the growing popularity of generative artificial intelligence, place-based narratives of cities produced by large language models are gaining traction. However, it remains unclear how such algorithmically developed representations relate to human perceptions and experiences of urban space. Against this background, this paper compares ‘typical places’ generated by large language models with data collected through participatory mapping by university students, using the city of Mainz as a case study. The analysis reveals a distinct difference: While large language models tend towards reproducing a stereotypical ‘postcard city’ based on iconic landmarks, the students’ depictions are more grounded in lived experience. Iconic landmarks are also present in their responses, but they appear alongside places shaped by everyday use and socially assigned meanings. These findings underline the need to critically examine algorithmically generated urban images and to explore how they could be more closely connected with lived urban realities. In this way, future applications of artificial intelligence in the field of city branding could go beyond simplified and selective representations of cities and contribute to a more inclusive urban environment.» weiterlesen» einklappen

Schlüsselwörter

• participatory mapping
• student-generated maps
• city branding
• urban image
• large language models

Autoren

Klassifikation

DFG Fachgebiet:
3.45-02 - Humangeographie

DDC Sachgruppe:
Geografie, Reisen