From the perspective of differentiation theory, the article examines the conditions under which machines are considered artificially intelligent – and how these attributions change through media interfaces. The central thesis is that AI is not primarily a technical, but a communicative construction whose social plausibility depends largely on the design of its interfaces. But when is a machine social enough to be considered “real” AI? The term AI is and remains controversial. Against this backdrop, how should we evaluate tech companies’ pursuit of AGI (Artificial General Intelligence)? Sascha Dickel argues that AGI is a figure of expectation that pushes “real” AI further and further into the future.

The text is available for download in Open Access here.