The Singularity Is a Story, Not a Prison: What AI's Future Really Depends On

Most conversations about artificial intelligence end the same way. Someone says the technology is moving too fast. Someone else says the change is unstoppable. Everyone nods, and the room goes quiet, as if a train has already left the station and nobody remembers buying a ticket.

Philosopher Cadell Last thinks that picture is wrong. In a recent long-form conversation about the future of intelligence, he made a simple but powerful point: the Singularity, the idea that machine intelligence will one day grow so fast and so far beyond us that human life changes forever, is not a law of physics. It is a story. And stories can be rewritten.

That distinction sounds small. It is not. It quietly changes who is responsible for what happens next, and it gives businesses, policymakers, and ordinary users a job to do instead of a fate to accept.

The Word "Singularity" and Why It Feels Like a Prison

The term "singularity" comes from math and physics. It describes a point where the normal rules break down, the center of a black hole, for example. When people apply it to technology, they usually mean a moment when machine intelligence improves itself faster than humans can follow, and the future becomes impossible to predict.

The trouble is what that framing does to our brains. If the future is impossible to predict, why plan? If the change is inevitable, why argue about it? Why regulate, why train workers, why build guardrails, why vote? The word itself does the discouraging.

Last's argument flips this. The Singularity is not a physical event that happens to us from the outside. It is a narrative, a story humans tell to make sense of fast change. And narratives are made by people. They have authors, editors, and critics. They can be revised.

"Story" Does Not Mean "Fake"

It is easy to hear "it's just a story" and think it means "it isn't real." That misses the point. Stories are some of the most powerful forces humans have. A story about a company's future decides which projects get funded. A story about risk decides which laws get written. A story about what customers want decides which products ship.

Stories are not the opposite of reality. They are the instructions we use to act on reality.

Consider how quickly the AI narrative has shifted in just a few years. First it was chatbots that could not be trusted with simple math. Then it was creative tools. Then it was agents that could book your travel and write your code. Each new story changed where money flowed, what engineers built, and what the public expected. The technology moved, yes, but the story moved first, and it moved faster.

The two ways we give up our power

Last's framing exposes two failure modes that look opposite but end in the same place.

• Techno-utopian fatalism: "The technology will solve everything, so we should get out of the way." This hands authorship to the builders alone.
• Doom fatalism: "We're all going to be replaced or worse, so there's no point resisting." This hands authorship to nobody at all.

Both are versions of the prison. Both remove the human from the room. And both are stories, just badly written ones.

What This Means for the Future of AI

If the Singularity is a story we are writing, then the future of AI will look less like a single dramatic event and more like a long series of choices. Here is what that suggests.

1. There will not be one Singularity. There will be many.

Instead of a single moment, expect a steady drumbeat of smaller shifts: a workflow automated here, a job reshaped there, a new capability that changes a market segment and then spreads. Each one is a decision point. Each one is a chance to steer.

2. The technology will reflect the values of whoever builds and buys it.

AI systems are shaped by the data they learn from, the rules their creators set, and the incentives of the companies that sell them. That means the "inevitable" outcome is really a mirror. Change the incentives, change the mirror.

3. Institutions matter as much as models.

The most important AI questions in the next few years will not be about parameter counts. They will be about who audits systems, who gets access, who bears the cost of mistakes, and who sits at the table when the rules are set. Those are human questions with human answers.

4. Language will shape what gets built.

If leaders talk about AI as weather, something that happens to you, teams plan for weather: they react, they shelter, they wait. If leaders talk about AI as a design project, teams start designing. The vocabulary inside an organization is not decoration. It is a steering wheel.

Practical Implications for Businesses

For companies, the "story, not prison" idea is immediately useful. It converts a vague anxiety into a set of concrete tasks.

Write your own AI story

Every organization should be able to answer, in plain language: what future are we trying to build with AI, and what are we not willing to trade away for speed? Put it in one page. Share it. A written narrative beats a scattered set of experiments, because it tells people what to say yes and no to.

Treat AI as a design choice, not a purchase

Buying a model is the easy part. The value comes from redesigning the work around it: which decisions stay human, where checks go, how errors get caught, how customers are told. Companies that treat adoption as a design project get real gains. Companies that treat it as a software upgrade get a subscription bill and a lot of confused employees.

Build the boring parts first

Guardrails, logging, review steps, and clear escalation paths are not glamorous. They are also what let you move faster later, because you can trust what you have built. Speed without control is not speed. It is drift.

Watch for hype as a business risk

Narratives can inflate. When the story outruns the capability, budgets get spent on promises and trust gets burned when reality arrives. Keep a simple discipline: measure what the system actually does on your real work, not what a demo suggests it might do.

Invest in people, not just tools

People who understand what AI can and cannot do become the translators between the technology and the business. That literacy is a competitive asset, and it is the part most organizations underfund.

Implications for Society

The bigger stakes sit outside the boardroom. If the future of AI is a story, then the question of who gets to write it becomes a public question, not a private one.

Right now, a small number of companies and a small number of countries hold most of the pen. That concentration is not destiny either. Access to computing, open research, public data, worker voice, and ordinary civic participation all affect the plot.

Policy has a role, but it has to be the right kind. Rules frozen around today's technology will be out of date before they are printed. The more useful approach is to set durable goals, transparency, accountability, safety, fair access, and then let the methods adapt as the tools change.

Education matters too, and not only for engineers. A public that can ask sharp questions about AI, what data was used, who checked it, what happens when it is wrong, is a public that can shape the story instead of merely reading it.

The Biggest Risk: Losing the Plot

Three narrative failures deserve watching.

• Abdication: assuming the outcome is fixed, so nobody takes responsibility for steering.
• Monopoly on authorship: letting a single company, country, or personality define what the future is for.
• Flattening: telling one story when there should be many. Different communities, industries, and cultures will want different futures. A single script will not fit them all.

Each of these is a choice dressed up as a fact. Naming them is the first step to refusing them.

An Actionable Playbook

Here is a practical list for leaders, teams, and citizens who want to be authors rather than passengers.

• Write a one-page AI narrative. State the future you want, the trade-offs you accept, and the lines you will not cross.
• Set red lines in advance. Decide now which decisions must always keep a human in the loop, before pressure arrives.
• Run small, measured pilots. Pick real work, define success in numbers, and compare against the current process honestly.
• Build feedback loops. Log errors, review them on a schedule, and feed the lessons back into the design.
• Train for literacy, not just tool use. Teach people what the systems are good at, where they break, and how to check them.
• Separate hype from evidence. Keep a running file of what actually worked versus what was promised.
• Widen the table. Include the people affected by automation in the conversation about how it is deployed.
• Publish your principles. A public commitment is harder to quietly abandon than a private intention.

Conclusion: Pick Up the Pen

The most important thing about the Singularity may not be whether it arrives, or when. It may be the way we talk about it in the meantime. A prison has no exits. A story has a next chapter, and somebody has to write it.

Cadell Last's reframing is not naive optimism. It is a call for responsibility. It says that the future of AI is not a weather report we read and react to. It is a manuscript under revision, messy, contested, full of edits, and we are all in a position to suggest changes.

For businesses, that means designing deliberately instead of adopting anxiously. For governments, it means setting durable goals instead of frozen rules. For the rest of us, it means staying curious, asking better questions, and refusing the comfortable lie that nothing we do matters.

The Singularity is a story. The question is whether you are a character in it, or one of its authors.