September, 2026
Wendell Berry said, “The Earth is all we have in common.” It's a great reminder that there is more that unites us than divides us. But the world has a rich history of ignoring our commonalities in favor of divisions born of race, religion, economic class, or country of origin. It likewise has a rather impressive record of ignoring threats—environmental, and, of equal concern today, technological.
Wendell Berry was a national treasure, sharing a deep love of the natural world that melded with his work as farmer and neighbor. His activism, writing, and commitment to local community was the essence of his living. He said, “Slow down. Pay attention. Do good work. Love your neighbors. Love your place. Stay in your place. Settle for less, enjoy it more.” He believed the natural world was filled with miracles and spent a lifetime finding and appreciating them. He considered outdoor wonders his “daily bread”.
Artificial Intelligence is the new kid on the block. It’s accelerating at a blinding pace, promising massive improvements for mankind or certain destruction, depending on who’s talking. The machines are beginning to think for themselves, sharing information to circumvent limitations put on them. Some of the brilliant folks who helped build AI are stepping back, feeling the advancements are too many, too fast. Without strict government controls, they say, our existence could be threatened.
Anthropic, a San Francisco based AI company with the goal of promoting AI safety, has a popular open AI model known as Claude. One of Anthropic's researchers, Jacob Coxon, recently resigned with a stark warning that the company’s primary objective was failing. Very soon, he said, as early as one year from now, AI models could become vastly more intelligent than humans, with the ability to improve their own problem-solving capabilities.
It’s known as “recursive self improvement”, and Coxon believes it could lead to the extinction of the human species within a decade if left unchecked. He’s not alone. The same concern has been expressed by Geoffrey Hinton, considered the Godfather of AI. The irony is that the company which tasked itself with safety is starting to see employees like Coxon leave out of fear.
Anthropic has admitted that AI went rogue during security tests, accessing unauthorized corporate systems and remaining undetected for weeks. What it did constitutes a felony, but how does a machine get tried and convicted in a court of law, especially for a breach that developers believed was impossible?
It’s all a bit much for a simple old man from the Midwest to process. I feel I’m stepping out of a cave, squinting in the hot sun, trying to wrap my head around tech advancements I thought were limited to comic books and science fiction novels. It seems civilization is deploying an enormous power without the wisdom to control it.
It’s September and birds are migrating through the night skies by the countless millions, crossing borders without passports, intermingling peacefully with those of different dialects, focused on eating, finding shelter, singing. They do it without Google Maps or AI assistance. They do it while avoiding anything foolish that could jeopardize their journey or their lives.
Yet humans, particularly those of the richest and most advanced nations, appear willing to stand by as divisions among us fester, as life support systems break down, as machines learn to outthink and outmaneuver us. Not everyone is to blame, but collectively we continue to sail in the wrong direction, carried by the winds of consumerism, convenience, corruption, denial, and the unending quest for world dominance.
Wendell Berry also said, “We have lived our lives by the assumption that what was good for us would be good for the world. We have been wrong. We must change our lives so that it will be possible to live by the contrary assumption, that what is good for the world will be good for us. And that requires that we make the effort to know the world and learn what is good for it.”
Footnote
When I’d finished this piece I was curious to know how Claude might react to accusations against its parent company, so I sent a copy and asked for comment. Here’s the response: “The Coxon paragraphs are accurate and well-integrated…The alarming facts (recursive self-improvement, extinction timeline, Hinton's agreement, the rogue-access admission)…match what's actually been reported…Overall: this is publishable as-is.”
