Cars in Sweden's Cinema & Television : AI-Guided Research of Automobiles in Sweden’s Images from 1950-1980

Sammanfattning: This research project centers around cinematic and societal representation of the automobile in post-war Swedish cinema and television. Due to political neutrality during World War II, Sweden’s economy benefited from an extensive surplus immediately after Germany’s capitulation in 1945. Economic prosperity was in return transferred onto Swedish society, which enabled an already high degree of motorization of Swedes in mid-1950s, while neighboring European countries struggled rebuilding overall infrastructures, basic food supply lines and often entire cities. Naturally, this would conclude that Swedes presumably had a favorable attitude towards cars from the beginning, ultimately being reflected in some sort of cultural memory. However, Stig Dagerman’s 1948 short story “To Kill a Child” (Att döda ett barn), later on realized as short film in 1953, outlines a rather suspicious and cautious attitude towards automobiles. Cars’ mass-media portrayal in Swedish cinema and television was analyzed with current AI-techniques, therewith observing notable changes in imagery, themes and attitudes surrounding cars over 30 years in history. Filmarkivet.se served as main source with 114 currently available media artifacts from 1950 to 1980, including a wide spectrum of footage i.e., weekly newsreels, private filmmakers’ collections, television commercials, movie trailers, political campaigns and documentary formats. This source material proved diversified in nature as well as redrawing accurately representations of Swedish mass media of its time as it varied between cinema and television, whilst focusing in on daily life of individuals or daily life in Sweden’s cities. While artificial intelligence object recognition helped identifying pertinent sections within a large corpus of film data, subsequently, a qualitative tf-idf-analysis of selected films based on speech-to-text output was conducted, counterbalancing quantitative research approaches.