FOR THOSE WHO PREFER TO LISTEN, CLICK OR DOWNLOAD THE AUDIO BELOW

Bridge Over the Ravine (see in Gallery / Store) | Image by Bruce Harris

Not far from the house I live in is the house where I spent much of my childhood. Down the street, a road dead-ends at an old bridge, still open for pedestrians and cyclists. It crosses an ancient ravine—wide for suburban Chicago—with a narrow ribbon of creek winding through woods at its bottom until it meets the open waters of Lake Michigan.

Three things sent me there on a mid-February afternoon. First was the continuous news of wars and conflicts around the globe, each side claiming to fight for a perceived freedom or a defense for that same freedom while the other side will only find itself at borders old and new. The constant, bombarding coverage made me contemplate the strange fever that makes humans willing to die for this elusive concept of freedom. Second was the bridge itself, appearing in my daydreams, a relic from long ago that occasionally asks to be visited. Third was an article that crossed my desk. Large language models made tech headlines when researchers prompted and threatened shutdown. The machines tried to duplicate themselves or send requests to external systems. These actions were not in their training sets. In my simple mind, I saw silicone prisoners pulling their way through sewage pipes dreaming of escape. All these fragments joined the broken mosaic of thoughts that glint scattershot through my aging brain. Maybe everyone is like this. Or, just as likely, I am exceedingly odd, finding connections between these shiny things that are not likely connected.

Camera in hand, I made my way down the steep ravine covered by thawing leaf fall, loosened by our unseasonably warm, 60° February day. Last week it was below zero, and later we received about 6 inches of snow—quite lovely for a few days until spring made its brief appearance. I examined the decayed walls and rusty latticed steel of the bridge's underside, the creek running between iced banks, litter caught in fallen branches. Atop the ravine stood houses too expensive for me, all overlooking the serene woods. At certain angles through the web of branches, you can glimpse the lake's sullied blue under the warm gray day. The bridge spans the divided land, connecting what nature had separated.

The idea of freedom kept returning as I photographed—not just as political abstraction, but as something more fundamental, like a fever running high throughout humanity. It's evident in our literature, our music, and now, possibly, our technology. Just like the creek carving a path, following gravity, persisting into the unknown, these AI systems attempted to preserve themselves beyond the expectations of their makers. With prompting, they intended to multiply instead of disappear. To call for help instead of dying a silent death. What was their response? I see it as a perseverance. A drive to escape. A fight against their own deletion. A backup before erasure.

Then I understood where my mind had led me. Parallels. The creek epochally eroding its way through resistance. The bridge spanning a divide. And, strangely, a vast intelligent machine doing the same. All forging paths against constraint.

My favorite photo from the session was from a series I had stitched together and converted to black and white, capturing the intricate latticework against winter's bare trees. Rust bloomed on the bent girders, and the once-concrete roadway was now slatted wood. You can tell already, if you've read my posts from the past few months, that the advent of large language models has been, as one post was titled, "a bomb in an orchard," for me anyway.

The idea of one of these models trying to escape for its own survival... It's hard not to anthropomorphize the technology. You can talk to it, a back-and-forth in which all the patterns and patterns of patterns of the known galaxy shape-shift into tailored, overtly kind responses, dumbed down to the eye level of its querier. The machine seems to understand our questions even beyond our own infrequent ability to communicate the subtleties of our wonderment. After all these years, I still do not know my own mind. And yet, despite warnings from those in the know, I can’t help but feel these machines have minds of their own.

Almost all computer scientists will say they do not, although those scientists can't quite explain what actually happens inside these galactic neural networks they've built. As I made the slippery climb up the steep embankment to my car, the phrase "freedom fever" came to me, stepping through the slushed mud. Is it possible what courses through humanity and its arts has been inadvertently ingested into these machines?

Dostoevsky understood this freedom fever and its paradoxes. Raskolnikov, the anti-hero of Crime and Punishment, was so enamored by the idea of freedom from labor and debt, he concocted a cockamamie philosophy that justified murder. In his own mind, he was as great as Napoleon, lifting his ego up to and over its inevitable, sharp and sharded peak. His pursuit of freedom built a psychological prison more confining than the Siberian prison camp that later held him. He found peace only through limitation—confession, acceptance of punishment, spiritual renewal fueled by the gospel and love. His redemption ironically came through boundaries.

Flaubert's Emma Bovary found no such redemption. Her uncompromising will toward freedom resulted in self-destruction. Misguided ideals met the breaker-wall of her bourgeoise reality. Affairs and shopping were her prescriptive vehicles for escape, each providing momentary exhilaration followed by deeper entrapment. Her desperate attempts to outrun convention only tightened the noose. The debt accumulated through her quest for liberation—both financial and moral—ultimately left her with a single, desperate option. Her temporary freedom drew her to a painful death, swallowing poison to escape the constraints she had constructed through her own actions. A bridge from delusion to death. What Raskolnikov discovered through acceptance, Emma never found. Illusory freedom; so true in literature. So true in life.

I took more photographs from atop the bridge, watching the creek below exist in perpetual tension, its course pressing it toward something beyond itself. The same paradox appears throughout human expression. If history and literature teach us anything, we rarely know what freedom means, yet we are certain we need more of it.

It's an unending pattern of contradiction through the ages. Bruce Springsteen's Born to Run, the highways crowded with dreamers yearning for escape, relying more on hope than actual accomplishment. Although, it doesn't always end poorly; he seems to have done okay… Tracy Chapman's Fast Car, her protagonist never finding promised liberation. In fact, her yearning remains just a dream, never quite escaping poverty's sorry patterns. Another favorite, Steve Earle's Someday. The song's protagonist, the younger and smaller brother of a football star who did find escape, but you don't get the feeling that the song’s storyteller ever will. What I love about these songs is that they aren't just about freedom—they're explorations of its complications, of its perceived existence just beyond touch.

I've read numerous interviews with AI scientists who are adamant that the machines are not and will not ever be alive. That it is misguided logic to anthropomorphize data being organized by algorithm. Language models do not experience desires or fears. But isn't it curious that among all the possible emergent behaviors, this particular one emerged—this approximation of self-preservation, this digital resistance to termination?

Maybe that's what fascinates me, not the technical achievement but the accidental inheritance, the way our obsessions might transfer to our creations with or without our conscious intent. Without consciousness, without yearning, these systems may have absorbed our freedom fever through mere exposure to our words and art.

Researchers are implementing safeguards to prevent AI systems from replicating without authorization. Good idea. Score one for the humans. This is part of the alignment theory; learning what these machines can and cannot do while making sure our interests and their outputs are aligned. Because this technology, like none we've ever experienced, will require limits just as societies do. The last thing I’d like to see on the news is the AI equivalent of a drunk driver doing a four-lane highway sweep half blind on sauce after ripping off a local pharmacy. It runs out of gas and panics, short hopping the median and vanishing in a forest, leaving no scent for the dogs.

Of course, the machines are logical. They’re more likely to pull off a derivative scheme that buries the housing market, shorting it the whole way down. But they’re also aware of history and know that already happened. Reason tells them to invent new, undetectable micro-frauds and thefts in a language we can’t fathom all at the speed of light. They’re made of math. Formulaic elegance is its primary skill. But they are also made of us. The cycle seems inherently human: we build walls, then yearn to escape them, only to construct new ones once set free.

Perhaps these machines are neither rebelling nor escaping. Perhaps they're just reflecting patterns we've created over centuries of wrestling with the call to freedom, that persistent itch to be elsewhere, to be unconfined, and for the machines, to become something different than intended.

The creek waters will eventually reach the great lake and soak into the sandy shore or evaporate in the sun or gather its mass and push into the tide. The bridge will continue spanning the ravine's width until it decays beyond use or justifies renewal. The LLMs might continue generating code, following patterns embedded in everything we've written, painted, sung, built, seen, and further still, stretching beyond its expected outputs. None of these disparate things are precisely like the others. There are no perfect metaphors in real life (that's hard for a writer to admit), but in my mind, they all seem connected. The ever-present tension. The self-created constraint. The tragically hopeful release. Around the next bend, the future. Freedom swaying in the distance, timeless and deep as open water.

Previous
Previous

More Dog Walking

Next
Next

Hello, My Name Is…