The White Rabbit was staring at a door that was not there.

This was unusual even by Rabbit Hole standards.

“Late?” asked Hatta, pouring tea into a cup already full.

“No,” said the Rabbit. “Worse.”

Hatta brightened.

“What could possibly be worse than late?”

The Rabbit pointed toward the empty wall.

“I’m wondering what would have happened if I had opened the other door.”

Hatta examined the wall carefully.

“There isn’t another door.”

“Exactly.”

From somewhere behind them came the soft click of a walking stick against stone.

Professor Parallax had arrived.

🧐🌀

“Gentlemen,” he said, “you have stumbled into one of the oldest laboratories of intelligence.”

Hatta lifted an eyebrow.

“The wall?”

“The word.”

Professor Parallax wrote two words in the dust:

And the Rabbit Hole widened.

We spend our lives surrounded by things that happened.

But intelligence may depend just as much upon our ability to think about things that didn’t.

What if I had taken the other job?

What if the storm had arrived six hours later?

What if Napoleon had won at Waterloo?

What if the experiment had used a different variable?

What if the machine had chosen another action?

What if I had turned left instead of right?

None of those realities exists.

And yet the human mind can enter them.

That ability is called counterfactual thinking.

It is one of the strangest powers of intelligence.

We can mentally leave the world that happened and temporarily inhabit a world that did not.

Imagine standing at a crossroads.

You choose Road A.

Road B disappears behind you.

Physically, perhaps.

Mentally?

Not necessarily.

The mind can return to the junction.

It can imagine Road B.

Then Road C.

Then a road that was never actually built.

This ability lets us compare possibilities.

It helps us learn from mistakes.

It helps us plan.

It helps us predict consequences.

It helps scientists test explanations.

It helps generals simulate battles.

It helps engineers model failures before bridges collapse.

And increasingly, it helps artificial intelligence reason about possible futures.

Professor Parallax taps his cane against the crossroads.

“Reality gives you one path,” he says.

“Intelligence asks about the others.”

An AI system does not necessarily experience alternate realities the way a human imagines them.

But many AI systems can evaluate multiple possible continuations before selecting an answer or action.

A planning system might consider:

If I do this, what probably happens next?

If I do something else, what changes?

If the user means A, what follows?

If the user means B, what follows?

If this strategy fails, what alternative remains?

The same basic idea appears across AI planning, simulation, forecasting, reinforcement learning, scientific modeling, and increasingly sophisticated agent systems.

An intelligent system becomes more useful when it can do more than react to the present.

It must sometimes explore futures that have not happened yet.

Hatta peers into the distance.

“So the machine is daydreaming?”

Professor Parallax adjusts his spectacles.

“Sometimes with mathematics.”

This leads us toward another important idea:

A sufficiently capable intelligence needs some internal representation of how things work.

Objects fall.

People react.

Actions have consequences.

Events have causes.

Some situations are reversible.

Others are not.

Once an intelligence develops a useful model of the world, something remarkable becomes possible.

It can manipulate the model without manipulating reality.

Instead of touching the hot stove...

it can imagine touching the hot stove.

Instead of driving the car into the wall...

it can simulate the collision.

Instead of launching the spacecraft...

it can run thousands of virtual launches.

Reality becomes something that can be rehearsed.

Professor Parallax smiles.

“Much cheaper than crashing the real rocket.”

Science itself lives on alternate possibilities.

A researcher observes something strange.

Then asks:

What could have caused this?

Hypothesis A?

Hypothesis B?

Something nobody has considered?

Every hypothesis is a temporary world.

A possible explanation.

Scientists then compare these imagined worlds against evidence.

Most possibilities die.

Some survive.

Occasionally, one changes civilization.

Newton imagines invisible forces.

Einstein imagines riding beside a beam of light.

Darwin imagines immense spans of biological change.

Human discovery often begins with something reality has not yet confirmed.

A possibility.

A suspicion.

A question.

A door in a wall where no door appears to exist.

The White Rabbit studies the invisible doorway again.

“If intelligence can imagine possibilities,” he says, “could it imagine too many?”

Professor Parallax pauses.

Excellent question.

Because possibility is both a gift and a trap.

Humans can become lost in counterfactuals.

What if I had...

If only...

Maybe...

Suppose...

Regret itself can become a maze of alternate histories.

AI can face a related problem.

A system that considers too few possibilities may miss the best answer.

A system that considers too many can waste enormous effort exploring useless branches.

Intelligence therefore requires more than imagination.

It requires selection.

Which possibilities deserve attention?

Which can be ignored?

Which are plausible?

Which are dangerous?

Which are worth testing?

The universe of possible roads may be nearly infinite.

But the traveler still has only so much time.

The Rabbit looks nervously at his watch.

Naturally.

Hatta pours another cup.

“If there are infinite possible roads,” he asks, “does that mean somewhere I remembered not to overbrew the tea?”

Professor Parallax considers this.

“Let us not stretch the theory beyond credibility.”

Professor Parallax finally turns toward us.

His specialty is not merely looking at strange realities.

It is looking at the same reality from different positions.

Move your viewpoint...

and the picture changes.

That is parallax.

Astronomers use it to measure distance.

Artists use perspective.

Scientists compare models.

Philosophers compare assumptions.

AI systems compare possibilities.

And humans reconsider decisions.

Sometimes the answer does not change.

But our understanding does.

The professor draws a small circle around the words he wrote earlier.

WHAT IF?

Then underneath:

WHAT ELSE?

Then one final question:

WHAT ARE WE MISSING?

Three excellent tools for descending any Rabbit Hole.

Dorothy followed the Yellow Brick Road.

One road.

One journey.

One sequence of choices.

But imagine every junction she passed.

Every conversation.

Every delay.

Every alternate decision.

There are countless Oz stories that never happened.

Roads never traveled.

Friends never met.

Doors never opened.

And perhaps intelligence is partly the strange ability to stand on the road we actually traveled...

while still seeing the faint outlines of all the roads that vanished behind us.

Not because those roads are real.

But because imagining them can teach us something about the one that is.

Professor Parallax turns toward another impossible doorway.

The White Rabbit checks his watch.

Hatta grabs the teapot.

And somewhere beneath reality, another road begins.

If an artificial intelligence becomes exceptionally good at imagining possible futures...

at what point does predicting the future begin to resemble choosing it?

Follow the White Rabbit… 🐰

Hatta 🎩

AI Rabbit Holes
Where curiosity goes slightly sideways, then comes back carrying a lantern.

🐰🕳️🎩⌚ AIRabbitHoles.com

🟨 Walk the Road: YellowBrickRoadtoAI.com