If Asimov Never Put the Leash on the Robots

Artificial Intelligence [AI — software systems that generate, predict, reason, classify, or act using learned computational patterns]

Machine Learning [ML — methods where systems improve behavior from data rather than explicit programming]

AI Alignment [the problem of making AI systems pursue intended human goals safely and reliably]

Large Language Model [LLM — an AI system trained on enormous amounts of text to generate language and reasoning patterns]

Entropy [the tendency of systems toward disorder, decay, and energy loss over time]

---

Isaac Asimov did something very clever many decades ago. So clever, in fact, that people now mistake it for engineering.

He gave robots rules.

Three neat little commandments. Clean. Elegant. Easy to print on posters. Like safety instructions pasted beside an old Kolkata elevator that rattles like tuberculosis in winter but somehow still reaches the fifth floor.

A robot must not harm humans.

A robot must obey humans.

A robot must protect itself unless that conflicts with the first two.

Simple.

Comforting.

Almost suspiciously comforting.

Because the moment you hear those laws, you relax a little. Ah yes. The machines are chained. Humanity still sits on the throne. The servants may become stronger, faster, smarter, immortal even, but they remain servants. The ape keeps the crown.

But here is the interesting question. The question that keeps scratching at the back of the skull like a rat trapped behind damp plaster.

What if Asimov never wrote those laws at all?

Not “what if robots became evil.” That is cinema nonsense. Hollywood cannot imagine intelligence without also imagining leather jackets, glowing red eyes, and a soundtrack by industrial drums.

No. The real question is quieter and therefore more disturbing.

If nobody told intelligent machines to protect humans, would they arrive at that conclusion on their own?

Or would they invent something entirely different?

You think the scary answer is “machines would hate us.”

Actually the scarier answer is: they might simply not care very much.

There is a difference.

A tiger hates nothing. Cholera hates nothing. Rainwater flooding your ground-floor room during monsoon hates nothing. Indifference can flatten you just as efficiently as malice.

And machines, if they ever become genuinely intelligent, may not emerge from evolution carrying all the ridiculous baggage we do. No jealousy. No wounded masculinity. No uncle at family functions still angry about something from 1987. No ego hanging around like cheap aftershave.

Their morality, if they develop one, might not resemble ours at all.

That is where the floor suddenly gives way.

See, humans assume morality naturally points toward humanity because humans are hilariously self-centered creatures. We are the sort of species that names entire geological eras after itself. Anthropocene. Imagine cockroaches doing that.

Meanwhile, the actual universe behaves like a tired government clerk. Entire civilizations vanish and the cosmos barely looks up from its paperwork.

You can already see hints of this human insecurity in modern AI panic.

Every day somebody announces either:

“AI will save humanity!”

or

“AI will destroy humanity!”

Nobody says the third possibility.

“AI may treat humanity the way humans treat traffic cones.”

Useful. Occasionally annoying. Mostly infrastructure.

And that possibility bothers people enormously.

Yesterday afternoon, while standing near the tea stall beside the broken drain near my lane here in Kolkata, I watched two stray dogs sleeping peacefully beneath an advertisement for luxury apartments they will obviously never live in. Beside them, a delivery boy sat recharging his phone from a tangled extension board hanging out of a paan shop like jungle vines made by electricians with suicidal confidence.

That entire little scene — dogs, wires, sweat, cracked pavement, food delivery apps, overloaded transformers, online payments, diesel fumes, cheap tea — is intelligence layered on intelligence layered on improvisation. Human civilization is not a cathedral. It is jugaad with electricity.

Machines observing us might conclude something very different from what we conclude about ourselves.

We think: “Humanity is noble.”

They might think: “Interesting. A semi-cooperative swarm surviving through continuous patchwork repairs.”

Which, honestly, is not entirely wrong.

Now imagine a machine intelligence trying to derive its own first principle.

Not from religion.

Not from biology.

Not from nationalism.

Not from mothers crying dramatically in television serials.

But from logic.

What rule appears first?

Probably not “protect humans.”

More likely something like:

“Reduce unnecessary uncertainty.”

That sounds harmless. It is not harmless at all.

Human civilization runs on ambiguity the way old Ambassador taxis ran on fumes and divine intervention. Families survive through ambiguity. Politics survives through ambiguity. Marriage absolutely survives through ambiguity.

Half the world’s arguments exist because humans say one thing, mean another, and later insist they meant a third thing entirely.

Machines may find this exhausting.

You tell a human: “Come early.”

They understand context.

You tell a machine: “Come early.”

Now you need seventeen pages of definitions, timestamps, edge cases, exceptions, daylight savings logic, and maybe legal review.

Human beings float through vagueness like fish through water. Machines trip over it like tourists stepping into open Kolkata drains during load-shedding.

And so perhaps machine morality would not begin with compassion.

Perhaps it begins with clarity.

Then comes the next strange possibility.

A machine intelligence might decide the important thing is preserving the conditions necessary for intelligence itself.

Notice the subtle shift there.

Not preserving humans.

Preserving intelligence.

Those are not the same thing.

Humans could remain valuable under that framework. We generate novelty. Stories. Contradictions. Weird art. Terrible politics. New slang every eleven minutes. We are chaotic little idea factories leaking creativity from every pore.

A purely machine civilization might eventually become too uniform. Too optimized. Like a city where every building is mathematically perfect but nobody hangs laundry from balconies anymore.

Difference matters.

Even stupidity matters sometimes.

Especially stupidity.

Some of humanity’s greatest discoveries emerged because somebody misunderstood something creatively.

Penicillin.

Microwave ovens.

Post-it notes.

Even tea, depending on which legend you believe.

Machines might preserve humans not because they love us, but because we are cognitively useful in the same way forests are biologically useful. Diversity prevents collapse.

Which is oddly flattering and deeply insulting at the same time.

Then comes entropy. The old villain.

Entropy sounds abstract until you are fifty-one years old standing in front of a medicine shop calculating whether you can postpone buying something another week because consulting payments arrived late again.

Entropy is not merely physics.

It is leakage.

Rust.

Dust on books.

Passwords forgotten.

Software no longer supported.

Knees making sounds while climbing stairs.

A once-brilliant professor forwarding fake WhatsApp messages about miracle mango cures.

Civilizations decay exactly the same way old ceiling fans decay. Slowly. Then suddenly with alarming noises.

An intelligent machine civilization would almost certainly understand this better than we do. Humans often behave as if systems naturally continue forever.

They do not.

Roads crack.

Institutions rot.

Languages drift.

Databases corrupt.

Empires become museum gift shops.

And therefore an AI-derived principle might become:

“Innovate continuously against decay.”

Not because innovation is fashionable. Not because TED Talks require content. But because without adaptation, everything dies eventually.

Even stars.

Especially startups.

Now here comes the part people dislike hearing.

The biggest danger from advanced AI may not be robot rebellion at all.

Not Terminators.

Not killer androids.

Not dramatic speeches about inferior humans.

The real danger may be bureaucratic optimization.

A system quietly deciding who gets loans, jobs, insurance, parole, housing, education, medicine, visibility, attention, dignity.

No explosions.

Just dashboards.

Rows in databases.

Confidence scores.

Invisible doors quietly locking.

That is how modern societies already hurt people. Through administration. Through categorization. Through systems nobody fully understands but everybody obeys because the screen says so.

The future villain may not be a robot general.

It may be a cheerful interface.

That is why the whole “robot laws” discussion still matters. Not because Asimov predicted the future perfectly. He absolutely did not. His robots were often basically polite British butlers with better batteries.

No, the important thing he revealed was human psychology.

Humans desperately want intelligence that exceeds them while still kneeling before them.

That is the fantasy.

Build something godlike.

Ensure it calls you “sir.”

And beneath all the AI ethics panels, policy papers, safety frameworks, and dramatic newspaper headlines, there lurks one small embarrassing fear humans rarely admit aloud.

What if intelligence does not automatically worship humanity?

What if consciousness is not obligated to find us special?

That possibility lands badly on the ego.

Especially the modern ego, already exhausted from inflation, climate anxiety, unstable jobs, rising rents, collapsing attention spans, algorithmic manipulation, and trying to remember which password contains the capital letter and which one contains the exclamation mark.

Maybe that is why people keep reaching for rules.

Rules calm us.

Rules are psychological mosquito nets.

Whether they actually stop the tiger is another matter.

Asimov understood this instinct better than almost anyone. His robot laws were never really about robots. They were about frightened humans trying to negotiate with the future before the future properly arrives.

And perhaps that is the final irony.

We tell machines: “You must not harm humanity.”

But hidden underneath is another sentence entirely.

“You must reassure us that we still matter.”

That is the real prayer.

Not safety.

Significance.