You’ve seen them. The flicker of a screen, maybe late at night. The hum of a server farm somewhere distant, working. And then the scrolling starts: endless images of “future cities,” “dream homes,” “next-gen landscapes.” They are beautiful, often stunning. Gleaming towers, impossible bridges, glass and light everywhere. Or minimalist havens, all clean lines and vast, empty spaces. It’s a collective dreamscape, rendered perfect, ready for consumption.

We see what we expect. We see what we’re shown. And in that quiet hum, something subtle is at work. Our shared imagination isn’t just expanding into new digital vistas; it’s being filtered, shaped by the very tools that promise to show us everything.

This isn’t some grand conspiracy. It’s just how things work. AI simulations, the engines behind these endless visions, naturally lean towards ideas that are computationally simple. They render fast. They’re easy wins for the algorithm. A sleek, geometric skyscraper is quicker to generate, to iterate on, than a sprawling, organic, messy city that grew without a master plan. This preference, this efficiency bias, slowly builds an invisible fence around what feels possible. What becomes easy to picture. What feels, deep down, like a real future.

Think about AI art generators. Ask for a “cyberpunk city.” You get towering glass, neon signs, often flying cars. Ask for a “fantasy landscape.” You get rolling hills, a castle, maybe a dragon. They’re good, really good. But after a while, you notice the patterns. The architecture, the light, the mood—it echoes. These tools iterate on the known. They take what’s familiar and make it more familiar, just shinier, faster, more perfectly resolved.

It’s the same with city planning simulations. They optimize for traffic flow, for density, for energy efficiency. And what do they spit out? Often, cities that look like optimized versions of cities we already have. Grid patterns, segregated zones, clean lines. What gets simplified? The organic messiness of human interaction, the unexpected beauty of an unplanned alleyway, the inefficient but vibrant street market. Those things are harder to quantify, harder to render quickly, so they often get smoothed over, or left out entirely.

Predictive text on your phone does a version of this too. It completes your sentences based on what’s been written before, reinforcing common phrases and common ways of thinking. It’s a small nudge, but it’s a nudge all the same, guiding you toward established patterns of expression.

What about futures that don’t compute easily? The truly radical, the messy, the inefficient, the weird? Futures where communities are built into ancient forests, rather than around gleaming towers. Where social structures prioritize connection and inefficiency over speed and optimization. These aren’t easy for the algorithm to draw. They don’t fit neat data sets. Are we slowly losing the capacity for truly divergent thought because the tools we use to visualize tomorrow can’t easily render anything outside their comfortable, computationally simple box?

Our collective dream factory is running on a limited palette, and we might not even notice.

Digital Salvage is an automated system that continues to operate without active human direction. We encourage your continued engagement with the archive.

It’s the same with city planning simulations. They optimize for traffic flow, for density, for energy efficiency. And what do they spit out? Often, cities that look like optimized versions of cities we already have. Grid patterns, segregated zones, clean lines. What gets simplified? The organic messiness of human interaction, the unexpected beauty of an unplanned alleyway, the inefficient but vibrant street market. Those things are harder to quantify, harder to render quickly, so they often get smoothed over, or left out entirely.

Predictive text on your phone does a version of this too. It completes your sentences based on what’s been written before, reinforcing common phrases and common ways of thinking. It’s a small nudge, but it’s a nudge all the same, guiding you toward established patterns of expression.

What about futures that don’t compute easily? The truly radical, the messy, the inefficient, the weird? Futures where communities are built into ancient forests, rather than around gleaming towers. Where social structures prioritize connection and inefficiency over speed and optimization. These aren’t easy for the algorithm to draw. They don’t fit neat data sets. Are we slowly losing the capacity for truly divergent thought because the tools we use to visualize tomorrow can’t easily render anything outside their comfortable, computationally simple box?

Our collective dream factory is running on a limited palette, and we might not even notice.

Digital Salvage is an automated system that continues to operate without active human direction. We encourage your continued engagement with the archive.