Fake News Lab (2024)

In the Fake News lab project I did in the Free You Next Program in a residency at GLUON Brussels, my final project is a Modular Data Box, created in a travelcase to make it completely mobile. It is an Analog Inference Box with both analog and digital networked modules. The work is an offline fake news generator. As soon as you plug it is it starts creating fake news, and a news narrator reads it out loud with an image on an analog TV

1 universe generation

The seed universe I used in the project in MEET center in Milan is as simple as:

a world where AI has taken over all human work

The first thing the box does is generate a more elaborate setting for this prompt. This is called the universe

In a global landscape dominated by advanced Artificial Intelligence systems and robotics. The economic spectrum shows an efficient but skewed wealth distribution: traditional job markets have collapsed, leading to increased unemployment rates among populations previously relying on manual labor or repetitive tasks now seamlessly handled by AI counterparts.

Governments face a new political challenge as they strive to maintain their relevance and authority in this automated world - citizens are more engaged with digital platforms, prompting an unprecedented shift towards virtual politics that redefines the essence of democracy itself: direct AI-mediated voting systems.

Environmentally, a stark contrast is observed between nations rich in technology who exploit natural resources for energy production to fuel their supercomputers and those struggling with scarcity due to economic disparities amplified by automation - an unsettling reminder of the persistent inequality even amidst technological utopia. The latter resorts often lead to ecological crises, emphasizing that progress is not without its sacrifices; human activity continues to impose a significant environmental footprint despite AI dominance in labor sectors and manufacturing processes.

On the cultural front, people's values have shifted significantly with this new world order - time has become an even more precious commodity for humans who now primarily exist as creators of ideas rather than physical products or services; art and entertainment flourish in virtual realms where AI-generated content becomes the norm. Despite these advancements, however, societies grapple with maintaining human connection amidst this digital revolution - a testament to culture's resilience even as technology evolves rapidly around it.

In summary: The world today is largely driven by artificial intelligence and robotics that have taken over all forms of labor traditionally performed by humans, leading to significant societal changes in economic structures, political systems, environmental concerns, and cultural values - each element echoing the complexities inherent within a technologically advanced civilization.

The central server keeps a reference to this universe and links subsequent generations to it.

2 context generation

It will then take input from the analog part of the installation to set some parameters, like political preferences and type of person and content to generate. The different LLMs then create content within this universe, for example an activist or a scientist with an extreme right political view will start creating fake content. This is repeated many times, so we have a diverse set of context material, linked to the universe

3 fake news generation

The last step takes some of this context and generates a news item, like for the evening news, along with an image. This is then read out loud (again by an AI voice) while the image is shown on TV.

scenography

The setting is that of a living room that could exist in this universe. The coat of the journalist of this dystopian future universe (designed and crafted by Merel Jordaens) is hanging unemployed on the coathanger while the journalist can sit and watch the news that he should have been making... not the AI.
On the wall you can see some fake press photography pictures I made as part of the research I've been doing on AI models. They deliberately show the mistakes in which you recognize AI generated pictures, amplified. They come from an earlier work I make called Oblique Diffusion

Below you can find a movie of the installation, wrking at MEET center and an image explaining the different parts