The LOOOOOP Corp. interactive exhibition was a provocative exploration into the dangers of non-consensually using human data for training AI, leading to personalised attacks. We cleverly disguised our questionnaire to gather personal information under the guise of offering shopping assistance for a wholly different objective. This information was then utilised by AI models to craft rumours in both text and image formats, aimed at personally attacking the participants.
The exhibit enticed visitors to engage with six futuristic shopping assistance kiosks where users speak with the machine, promising a seamless experience where no direct product inquiries were made, yet all desired products were directly dispatched to their homes, bypassing the hassle of decision-making. However, upon exiting, visitors were confronted with their personalised rumours projected onto a wall, subjecting them to personal attacks. At the entrance and exit, we displayed two versions of a privacy agreement: one cloaked in redundant, misleading language and the other offering a frank disclosure of the rights collected and exploited. Moreover, the exhibition delved into alternative interaction methods with the system. A group of five engaged in spontaneous conversations, reacting to the rumour projections, which spurred a continuous dialogue with the AI. This interaction highlighted the induced anxiety and underscored the urgent need for enhanced protective measures.
