 |
Prologue: The Machines We Love and Lose
There is a peculiar cruelty embedded in the heart of modern technology — a cruelty that has nothing to do with malice and everything to do with time. Today, a man wakes up and asks an AI agent to summarize his emails, plan his day, draft his reports, and remind him to call his mother. He trusts it. He depends on it. It has become, in a very real sense, an extension of his mind.
And yet, if history is any guide, that same agent — the one he cannot imagine living without — will be a museum piece within a decade. Perhaps sooner. It will join the fax machine, the pager, the BlackBerry, the floppy disk, and the dial-up modem in the vast graveyard of technologies that once felt indispensable and now feel almost embarrassing.
This is the central paradox of our technological age: the more essential a technology becomes, the more certain its obsolescence. We build our lives around tools that are, by design, temporary. And nowhere is this paradox more acute — more emotionally and economically charged — than in the age of artificial intelligence agents.
Part I: The Age of the Agent
From Tools to Companions
Something fundamental shifted in the early 2020s. Software stopped being something we used and became something we conversed with. The launch of large language models — ChatGPT, Claude, Gemini, and their successors — transformed the relationship between humans and machines from command-and-execute to dialogue-and-delegate.
By 2024 and 2025, the industry's favorite word was no longer "chatbot" but "agent" — autonomous AI systems that don't just answer questions but take actions: booking flights, writing code, negotiating calendars, conducting research, managing customer service pipelines. Consulting giants like McKinsey (mckinsey.com) and Gartner (gartner.com) declared agentic AI one of the defining technology trends of the decade, with Gartner predicting that a substantial share of everyday business decisions would soon involve autonomous agents.
The numbers tell the story of dependence. Stanford's AI Index Report (hai.stanford.edu/ai-index) has documented year after year of explosive growth in AI adoption across industries, while surveys from Pew Research Center (pewresearch.org) show that ordinary people increasingly encounter AI daily — often without realizing it.
The modern knowledge worker's morning routine now often begins not with coffee and a newspaper, but with a prompt: "What do I need to know today?"
The Intimacy of Dependence
What makes AI agents different from previous technologies is the intimacy of the dependence they create. A hammer extends your arm. A car extends your legs. But an AI agent extends your cognition — your memory, your judgment, your creativity.
Researchers have begun to document what some call "cognitive offloading" — the tendency to let machines handle mental tasks we once performed ourselves. Studies published in journals covered by outlets like Nature (nature.com) and discussed in MIT Technology Review (technologyreview.com) raise uncomfortable questions: When the agent remembers everything, do we remember less? When the agent writes, do we lose the muscle of writing? When the agent decides, do we forget how to choose?
The man who relies on his AI agent today is not simply using a tool. He is entering a relationship — one in which he gradually surrenders pieces of himself in exchange for convenience. And that is precisely what makes the coming obsolescence so cruel.
Part II: The Graveyard of Indispensable Things
A Brief History of Abandonment
To understand what awaits today's AI agents, we need only look backward.
The BlackBerry. In 2009, it was the device of presidents and CEOs — Barack Obama famously fought to keep his. By 2016, its market share had collapsed to nearly zero. In January 2022, BlackBerry officially ended support for its classic devices, turning millions of once-essential machines into paperweights overnight (bbc.com/news/technology).
The iPod. Steve Jobs called it "a thousand songs in your pocket," and for a decade it defined how humanity listened to music. Apple discontinued the final model in 2022, its funeral barely noticed (apple.com/newsroom).
Internet Explorer. For years, it was the internet for most of the world. Microsoft retired it in June 2022, and the web shrugged.
Skype. Once a verb — "Skype me" — it connected families across continents and pioneered the video-call culture we now take for granted. Microsoft announced its shutdown in 2025, folding it into Teams, and a generation's digital memories went with it.
Each of these technologies was, at its peak, described in the same breathless language we now use for AI agents: revolutionary, indispensable, life-changing. Each is now history.
The Accelerating Cycle
What has changed is not the fact of obsolescence but its speed. The telephone took roughly 75 years to reach 100 million users. The mobile phone took 16 years. Facebook took 4.5 years. ChatGPT took two months — the fastest-adopted consumer application in history at the time, according to analysis widely reported by Reuters (reuters.com/technology).
But acceleration cuts both ways. Technologies that rise faster also fall faster. The half-life of a "cutting-edge" AI model is now measured in months. GPT-3, which stunned the world in 2020, was functionally obsolete by 2023. Models that cost hundreds of millions of dollars to train are routinely surpassed within a year by cheaper, faster successors — a dynamic tracked closely by researchers at Epoch AI (epoch.ai) and covered extensively by Wired (wired.com).
No comments:
Post a Comment