QT.Bot

The "rogue offspring" of Queering the Map, as the creator Lucas LaRochelle describes the QT.bot, is an artificial intelligence trained on a community archive of geo-located queer experiences. The project aims to subvert the reductionist logic of traditional data analysis and instead use AI to generate speculative queer pasts, presents, and futures, exploring the potential of machine learning when used in non-productive, poetic ways. QT.bot uses an implementation of the OPEN AI GPT-2 text generation model trained on submissions from the Queering the Map. Its first results were published in a video format featuring images generated from scraped Google Street View imagery of the tagged locations.

The Canada-based designer Lucas LaRochelle developed QT.bot in 2020 as a critical response to the dataset collected within Queering the Map. The purpose of most data analyses is to categorize and streamline information as a means to identify and control, often having a substantial impact on the respective human and non-human bodies. QT.bot tries to break with this logic and instead explore applications towards fabulation and fantasy.

Lucas describes the work as a collaboration, in which they act as a curator for the AI. In the selection process, Lucas is not interested in generated content that could pass as real, but mainly in the failures and feelings produced by AI-generated stories that can be identified as such. They describe the results as throwing a thousand broken mirrors back at Queering the Map.