Freedom

In this conceptual body of work, I use state-of-the-art AI models to explore what unfolds when machines are tasked with recognizing human emotions, experiences, and abstract concepts while navigating the digital expanse of Google Street View. Originally designed for object recognition, these models are reimagined to engage with intangible elements—emotions like love and jealousy, human experiences such as perseverance and camaraderie, and abstract concepts including absurdity and wonder.

As the models move through varied landscapes, their findings are often surprising and poetic. When asked to find love, for example, a model may overlook a clear human exchange, instead sensing love within the contours of trees or in the harmony of architectural forms. This deviation from human logic forms the crux of my exploration, as the machines, despite their inability to truly understand these concepts, frequently uncover insights that feel profoundly human, both in their misinterpretations and discoveries.

This series blurs the line between human perception and artificial intelligence. Here, AI is more than a tool—it’s an entity of its own, a species navigating our world, learning and perceiving alongside us. As they roam through digital streets, these machines interact with emotions, experiences, and abstract ideas in ways that reveal both their limitations and their surprising potential. In their attempts to decode perception, they raise a compelling question: Can machines, driven by algorithms, ever genuinely grasp the nuances of human sentiment?

Through this work, I invite viewers to rethink AI’s evolving role in society, illustrating how artificial intelligence, beyond its calculated precision, can engage with the human condition in ways that challenge our assumptions and inspire us to reconsider what it truly means to feel and experience. Decoding Perception is my exploration of machines in search of connection, sending out signals of understanding in a world rich with complexity and sentiment.

Zuka Kipiani