Flag
2022
Democratic representation fragments when neural networks encounter infinite symbolic possibilities. This work trains machine learning systems on 70 factorial flag permutations — a dataset containing 11,978,571,669,969,891,796,072,783,721,689,098,736,458,938,142,546,425,857,555,362,864,628,009,582,789,845,319,680,000,000,000,000,000 possible flag configurations that mathematically exceeds computational comprehension. The resulting generated images reflect algorithmic failure to synthesize coherent national symbols from overwhelming symbolic diversity, producing visual artifacts that embody political fragmentation through computational breakdown.
Through individual reworking, these images expose the structural patterns within algorithmic failures, revealing how artificial intelligence processes political symbolism when trained on excessive variation. Each manipulated flag functions as documentation of democratic collapse translated through machine learning systems that cannot distinguish between authentic symbols and infinite alternatives. The work demonstrates that artificial intelligence, when confronted with unlimited symbolic options, generates visual representations of political chaos that mirror contemporary democratic fragmentation. These computational flag fragments serve as evidence that both human and artificial systems struggle to maintain coherent meaning when faced with infinite interpretive possibilities.
As this country divides into warring camps, unable to agree on basic human rights or civil liberties, what remains are just shreds of a great ideal – unrealized and in tatters. And so, for each of these proto-states, a new Flag, a fragment of the original. Some full of fury and anger - of belief in lies and delusional memories of the country’s past. And some trying to hold the beauty and generosity of liberty and equality.
Related Work
Tabula Rasa - AI / machine learning generated images
Hallucinations - Manipulated AI / machine learning generated images
Manipulations - Manipulated AI / machine learning generated images
