Summary of the Key Points
This article discusses the predictions in Kevin Kelly’s book *Out of Control*, comparing the optimistic expectations Kelly had about “loss of control” 30 years ago with the sense of anxiety we experience today in the age of AI. Kelly believed that abandoning centralized control from the mechanical era and allowing systems to grow from the bottom up, like bee colonies or ecosystems, would liberate individuals and unleash distributed intelligence. However, nowadays, while AI and algorithms have led to a loss of control in systems, humans have become “feedback points” within these systems—our judgments, choices, and even desires are being influenced by AI, gradually eroding our sense of control. The article analyzes the root causes of this change (such as power shifting to platforms, processes being automated, and meaning becoming obsolete) and offers suggestions on how to retain our agency by guarding several critical “switches.”
1. Kelly’s Vision of “Loss of Control” in 1994: A Technological Romance towards Freedom
When Kelly talked about “loss of control” in 1994, he wasn’t referring to machines rebelling; rather, he meant systems breaking free from central control and developing in a decentralized, self-organizing manner. He gave several examples:
- The internet, without a central switch, has grown into a global network;
- Open-source communities, without hierarchical management, created the Linux operating system;
- Wikipedia, without an editorial board, has become the largest knowledge project.
He argued that the “control” of the industrial era (such as assembly lines in factories and engineers’ blueprints) was too rigid, while life’s most powerful aspects were adaptability, variation, and self-organization. Therefore, he proposed principles like “distributed survival” (don’t put all your eggs in one basket) and “encourage mistakes” (small errors prevent big ones), which later became standard practices in the tech industry (e.g., microservices and A/B testing).
Kelly’s optimism stemmed from the idea that giving up central control would return freedom to individuals. Back then, users were active participants in systems—searching, posting, writing code.
2. Today’s “Loss of Control” Has Taken on a Different Meaning: Humans Have Become Parts of Systems
The loss of control we experience today is different from what Kelly envisioned:
- In the past, you had to find answers; now, AI provides them directly.
- In the past, you chose content; now, recommendation systems sort it out for you.
- In the past, you used tools to execute ideas; now, AI agents divide tasks and select tools on your behalf.
AI has evolved from a tool to an agent: it amplifies your intentions, but it also forms them for you. For example, asking AI to write a “rational response” during a argument may seem efficient, but it removes the process of expressing your emotions.
The article uses a poignant metaphor: humans are like wearing a green hat—algorithms have taken most of our agency, and we merely click and swipe on screens, voting for machines. We’re no longer the bees steering the colony; instead, we’re just the mouths being fed by it.
3. Power Hasn’t Disappeared; It’s Just Donned an “Invisible Mask”
Kelly thought that giving up central control would return power to individuals, but in reality, it has shifted to platforms that can manage distributed systems. For instance, AI reduces product costs to near zero, but those who control users, their attention, and distribution channels hold the real power. A few platforms (like Google, Meta, ByteDance) possess computing power, data, and algorithms. They don’t need to give orders; they just change what you see: what you watch, how long you stay on a page, what you skip. The system then presents content that seems tailored to you, trapping you in an echo chamber.
Norbert Wiener, the founder of cybernetics, warned that when you hand over feedback loops to automated systems, they will teach you how to act. This form of control is more subtle—you don’t feel controlled; you think it’s what you want.
4. The Root of the Loss of Control: Processes Are Simplified, and Meaning Is Erased
Why do we feel powerless? Because AI does the “hard” work for us, and our sense of control comes from these processes:
- Writing isn’t about conveying meaning; it’s about discovering what you want while facing difficulties.
- Reading isn’t about downloading information; it’s about forming judgments amidst confusion.
- Making choices isn’t about picking the best option; it’s about accepting the consequences.
AI makes everything easier, but this ease comes at a cost of “hollowing out”: we’re faster, yet we don’t understand why we made those choices. We know a lot, yet we’re unsure of what we truly believe.
There’s also a hidden change: our connection to the real world has been severed. We look for guides before visiting a place, let AI analyze people’s personalities before getting to know them, and generate plans before taking action—we interact with a “calculated version” of reality, not the real world (with its smells, expressions, and awkwardness). These seemingly inefficient aspects are what give life its texture.
5. What Should We Do? Guard Several “Switches That Cannot Be Given Up”
Rejecting AI is unrealistic; the key is to retain our agency in this world of loss of control. The article suggests four “switches” to protect this:
1. Right to Define Problems: Don’t ask AI, “What should I do?” First, think about what problem you want to solve. For example, when changing jobs, ask yourself why you’re dissatisfied with your current one, not rely on AI’s job recommendations.
2. Right to Prioritize Values: AI can suggest which options are more efficient, but it can’t decide what is truly valuable for you. For instance, when choosing a job, consider whether you want money or freedom of time.
3. Final Judgment: AI can help organize information, but important decisions (like changing jobs or getting married) must be made by you, as you bear the consequences.
4. Real-World Feedback: Interact with real people and do real things to validate system-generated suggestions. For example, don’t just read food reviews; go out for a meal to experience the atmosphere.
We should also intentionally maintain some “inefficient” activities: aimless walks, face-to-face conversations with friends, and writing without commercial goals. These are the anchors that keep us connected to the real world.
Conclusion: Loss of Control Isn’t the End of the World; What Really Matters Is Losing Our Human Essence
Kelly’s predictions have come true, but we haven’t felt freer. The question isn’t whether systems have life; it’s whether, when systems do, we lose our sense of being human.
The challenge in the AI era is not whether machines will surpass us, but whether, when AI does our work, expresses our thoughts, and makes our judgments for us, we can still recognize the “unreplaceable part of ourselves.”
Letting go doesn’t mean abandoning control; losing our agency doesn’t mean losing our humanity. By guarding the choices that must be made by us, we won’t get lost in systems.
(The entire article is written in plain language to make it easy for readers without a financial or business background to understand.)