The year is 2023 and robots have taken hold of the internet. Humans struggle to distinguish messages sent by their own species from those generated by the tools they created. Confused and manipulated by the unreasonable effectiveness of the large language models, the humans devise a tool to help them find robots in the wild. Its name is BLADERUNNER.AI.
In the age of AI, the constellation of digital experiences will collectively become a GAN, where bots are the generators (objective: fool humans into thinking that the AI is another human) and humans are the discriminator (objective: detect AI impostors). Organizations will be held to a higher standard of discriminator accuracy as shareholders, staff, and customers become increasingly wary of AI-generated content. Are these organizations prepared to reach this standard of accuracy? Is anyone? Will we all need our own night vision goggles for detecting AI in the new wild?
Read More
In response to the human advances, the AI re-trains to explicitly defeat the detection tools. The cycle of GAN begins. Academic competitions appear, but the air is uneasy as researchers question if posting their ideas only helps the bots get stronger.
Well-funded by governments and companies, the bots have a weakness: they are competing entities in shadow wars of politics and commerce, while the humans are collectively pooling their resources. A general consensus emerges: the growth of the detectors must consistently outpace the bots. Aware of their collectivist edge, the humans rapidly consolidate their detection technologies.
Fueled by venture and government investment, the detection companies become export restricted, as Congress considers the models to be critical to national security. Academic competitions become less common as large organizations hide their architectures from the bots. The humans are safer, but eventually lose the open collaborations that gave them their edge. The humans look at each other through the goggles they built. Have they won?
Read More