It might come true…

Interactive one-way mirror installation.

Using uncertainty to counter the assumed accuracy and infallibility of AI predictions, ‘It might come true…’ is an interactive mirror that is shown alongside the film installation ‘The Cambois Oracle’ which uses the figure of the Delphic Oracle to comment on the current expansion of hyperscale data centres and the appeal of predictive technologies.

The work uses a one-way mirror as a compositional logic, riffing on Frank Pasquale’s claim that the black box society and its reliance on opaque decision-making algorithms ‘resembles a one-way mirror’. In terms of transparency and the general public’s access to information regarding exactly how their data is used to assess and make judgements, there is a huge inequality between data-gatherers and data-subjects. This unequal relationship is encompassed in the one-way mirror metaphor. In the artwork, this is emphasised by the sensor system being hidden behind the one-way mirror, imperceptible to viewers and echoing the data-gathering systems underpinning surveillance capitalism.

The mirror, adorned with iconography and symbolism relating to the Delphic Oracle – pythons, laurel leaves, ancient temple architecture – presents predictions and prophecies to audience members. When someone stands in front of the mirror, a sensor is triggered and they receive a prediction with a distinct sense of uncertainty. Instead of phrasing predictions with clarity and strength, they are fragile and wavering. The AI prophecy is unsure and refuses to speak with conviction. ‘You will [insert a future action]’ becomes instead, ‘You might…’ or ‘You may…’. These predictions do not offer a sense of stability or security.

All of the predictions are based on decisions that companies and businesses actually use AI for: job hires, healthcare waiting lists, access to housing, and so on. The predictions are based on the research presented in the paper: Whose Needs are being Met? Looking at Agency, Authority and Social Risk in Relation to Forbes’ Top 50 AI Companies (September 01, 2024). Available at SSRN: https://ssrn.com/abstract=5747884 or http://dx.doi.org/10.2139/ssrn.5747884