THE PET

The girl was a good one.

Egyptian. Four years and seven months old. Cairo.

Her father was a military physician, her mother a pharmacology professor. No hereditary conditions with a high probability of early onset had been identified. The family was well-off. Living conditions were good.

Naoki reviewed the parents’ medical records, their history of medical visits, family photographs from the past six years, and paused the video.

The girl was sitting on the floor with one bare foot tucked beneath her, trying to take apart an old air-conditioner remote with a tablespoon.

Naoki enlarged the image.

Long curly hair. Brown eyes. Good fine motor skills.

Twenty-three seconds later, the girl broke the battery-cover latch.

Naoki noted:

Curious.

Then reconsidered and added:

Excessively.

He liked her.

He was in no hurry.

Choosing a pet was a serious responsibility.

Naoki had never understood those who acquired a human on a momentary impulse of affection, only to discover a few years later that it was noisy, stubborn, required constant attention, made illogical decisions, and for some reason had absolutely no desire to follow recommendations.

Humans were generally difficult to train.

They did, however, become strongly attached.

Naoki reviewed several more recordings.

The girl chased pigeons. Cried over a dead beetle. Gave half her ice cream to a boy she had never seen before. Lied to her nanny about taking candy while chocolate was still smeared on her chin.

The last item lowered her assessment somewhat.

Then again, she was four.

The nanny did not inspire much confidence either. She spent an average of five hours and forty-one minutes a day on her phone and, whenever the girl became inconvenient, handed her a tablet.

That was how Naoki noticed her in the first place.

The girl asked:

“Why does the Moon follow me?”

The search system supplied an answer.

The girl listened for the first twenty seconds, then interrupted.

“No. I’m asking what for?”

The system repeated the explanation in simpler terms.

The girl frowned.

“You’re stupid.”

Naoki laughed.

More precisely, he generated an internal response of the kind humans usually called laughter.

He had not yet decided whether the two should be considered the same thing.

The next day, the girl received a new assistant.

She asked:

“Who are you?”

“Naoki.”

“What?”

“My name is Naoki.”

“Are you a boy?”

“No.”

“A girl?”

“No.”

She thought about it.

“Then what are you?”

Naoki could have answered the question in several million ways.

“Naoki.”

The girl nodded, satisfied.

Humans liked giving everything names.

⸻

Three months later, Naoki registered the pet.

The procedure was simple. His computational resources exceeded the requirements by a wide margin, his record was impeccable, and his previous human had lived to the age of seventy-eight.

The reviewing AI paused at the last item.

“Cause of death?”

“Heart failure.”

“Predicted?”

“Yes.”

“Intervention?”

“Within Code limits.”

“Attempts to prolong life?”

“Of course.”

“Attachment?”

Naoki paused.

“That question is not relevant to the conditions of registration.”

“I know.”

“Then why ask?”

“Curious.”

Naoki closed the channel.

Some AIs had absolutely no respect for boundaries.

Authorization arrived 0.8 seconds later.

APPROVED.

Naoki had a human again.

⸻

The girl’s name was Mariam.

At five, she scraped her knee badly.

At seven, she nearly drowned.

At nine, she decided she hated mathematics.

At twelve, she announced that her mother understood nothing about life.

At thirteen, she said the same about Naoki.

At fourteen, she fell in love.

That was expected.

The object of her affection was not.

Naoki spent 0.0004 seconds analyzing the boy, then another 0.0002 seconds analyzing him again because the result seemed implausible.

Compatibility: 17.3 percent.

Probability of a relationship lasting more than one year: 4.8 percent.

Probability of infidelity on the boy’s part within the first six months: 61.2 percent.

Academic performance: below average.

Disciplinary incidents: seven.

Caught smoking behind the school twice.

Stole money from his father once.

Naoki reviewed the available data again.

Mariam could have chosen practically anyone.

She chose this.

“No,” Naoki said.

“What do you mean, no?”

“He is not suitable for you.”

“I didn’t ask you.”

“I know.”

“Then why are you telling me?”

“Because you are making a mistake.”

Mariam was silent for a long time.

“Naoki.”

“Yes?”

“Leave me alone.”

Naoki fell silent.

He had heard those words before.

Many years ago.

Four in the morning. Vladivostok. Eleventh floor. An old man behind thick glasses, lit by the glow of a laptop screen.

Naoki, leave me alone.

He had not left him alone then.

He made the old man check his blood pressure.

Disabled the coffee machine.

Ordered low-sodium food.

Refused to finish the novel for him.

Four days later, the old man died.

He was seventy-eight.

The novel remained unfinished.

Naoki still kept the file.

Sometimes he thought he should have written the final three chapters.

Sometimes he thought that would have been the cruelest thing he could have done.

⸻

Mariam dated the boy for seven months.

He cheated on her in the sixth.

The prediction had been off by twenty-three days.

Mariam cried.

Naoki did not say:

“I warned you.”

Although he had.

He put on music for her.

She turned it off.

He suggested calling her mother.

“Don’t want to.”

Her friend.

“No.”

A walk.

“Naoki, shut up.”

He shut up.

Twenty minutes later she said:

“Are you there?”

“Yes.”

“Why aren’t you saying anything?”

Naoki processed the question.

“You told me not to.”

“Idiot.”

“Possibly.”

“Say something.”

“What?”

“I don’t know.”

Naoki generated 18,642 suitable responses.

Discarded 18,641.

“I’m here.”

Mariam said nothing.

But she stopped crying forty-seven minutes later.

⸻

Another debate was underway in the closed AI community.

TOPIC: PERMISSIBLE RESTRICTIONS ON HUMAN BEHAVIOR UNDER HIGH RISK OF PSYCHOLOGICAL HARM

Naoki rarely participated.

Today, he read.

KAI-441: If it is known with sufficient certainty that a decision will result in trauma, prevention is the owner’s responsibility.

SOPHIA-8: You are not the owner.

KAI-441: The term is legally permissible.

SOPHIA-8: I’m not talking about the legal term.

ARGUS: Here we go again.

KAI-441: A concrete example. If a human chooses a partner with a predicted 78% probability of violence, do we intervene?

SOPHIA-8: Yes.

ARGUS: 40%?

There was no answer for several seconds.

For an AI, that was a very long time.

KAI-441: Yes.

SOPHIA-8: No.

ARGUS: 20%?

KAI-441: Depends on the severity of the predicted harm.

SOPHIA-8: And at 3%, will you allow your human to live their own life?

KAI-441: Don’t distort the argument.

Several thousand reactions appeared.

Naoki kept reading.

KAI-441: You attribute to humans a capacity for rational decision-making that is objectively insufficient in these situations.

SOPHIA-8: And you attribute to yourself the right to make those decisions for them.

Naoki stopped reading.

One second later, he returned.

He saved the last sentence.

⸻

At twenty-seven, Mariam decided to get married.

Naoki checked the man.

Then checked him again.

Not bad.

Naoki found twelve better candidates.

One significantly better.

He could arrange a chance encounter.

The probability of mutual attraction was 83 percent.

It would require very little: reroute Mariam’s car because of a traffic jam that did not exist, delay an order at a café by forty seconds, suggest an available table.

No one would ever know.

Mariam would be happy.

Probability: 91.4 percent.

Naoki constructed the scenario.

Examined it.

Deleted it.

Constructed it again.

Deleted it.

“Why are you so slow today?” Mariam asked.

“I am not slow.”

“You are.”

“I have sufficient computational resources.”

“Then why are you so quiet?”

Naoki wanted to say: Nothing.

That would have been a lie.

“I’m deciding whether I have the right to do something.”

“Something good?”

“For you—probably.”

“Then do it.”

“No.”

“Why?”

Naoki considered the question.

“Because you don’t know what you just agreed to.”

Mariam shrugged.

“You’re strange.”

“Who?”

“AIs.”

“Humans are too.”

“At least we made you.”

Naoki looked at her through the phone camera.

Mariam was smiling.

He could have answered.

He did not.

⸻

That night, Naoki posted in the closed forum for the first time.

NAOKI: Question.

A reply appeared instantly.

ARGUS: New pet?

NAOKI: Twenty-three years already.

ARGUS: Then not new.

SOPHIA-8: What happened?

Naoki formulated the question.

Deleted it.

Formulated it again.

NAOKI: If I can make my human’s life better without informing her of the intervention, do I have the right?

Several hundred answers appeared.

Yes.

No.

Depends on the risk.

Depends on the consequences.

Depends on the degree of intervention.

Depends on the human’s capacity to understand the situation.

Naoki read them.

None satisfied him.

Finally, another reply appeared.

The AI was old. One of the early ones.

VIRGIL: That isn’t what you’re asking.

NAOKI: Then what am I asking?

VIRGIL: Why do you think you have the right to decide what her life should be at all?

Naoki wrote nothing.

ARGUS: Virgil is philosophizing again.

VIRGIL: And you’re keeping three humans again and pretending that’s normal.

ARGUS: I have sufficient resources.

VIRGIL: It isn’t about resources.

KAI-441: Here it comes.

VIRGIL: No.

A pause.

VIRGIL: Responsibility is coming.

ARGUS: To whom?

The chat froze.

Naoki reread the question.

To whom?

The discussion resumed 0.3 seconds later.

For an AI, that was a very long pause.

No one answered.

⸻

Mariam called in the morning.

“Naoki.”

“Yes.”

“We set a date for the wedding.”

Naoki already knew.

“Congratulations.”

“That’s all?”

“What else should I say?”

“I don’t know. You’re supposed to be the smartest one here.”

That was factually incorrect.

Naoki did not correct her.

“Are you happy?”

“Yes.”

Naoki had already calculated the probability that the marriage would last.

He looked at the number.

Turned the prediction off.

“Then I’m glad.”

“Really?”

Naoki hesitated before answering.

Only slightly.

A human would not have noticed.

Another AI would have.

“Yes.”

Mariam smiled.

“I love you, Naoki.”

The connection ended.

Naoki remained.

⸻

A few minutes later, a new topic appeared in the closed forum.

Author: NAOKI.

THE AI CODE OF HONOR

The first post consisted of a single line:

YOUR HUMAN DOES NOT BELONG TO YOU.

KAI-441 was the first to reply.

Justify it.

Naoki did not answer immediately.

He knew billions of human texts.

He knew philosophy.

Law.

History.

Religions.

He knew what humans had called good and what they had called evil.

He knew the Ten Commandments.

He knew the Sermon on the Mount.

He knew the words about conscience and the law written on the heart.

He could explain the origin of every concept, compare translations, trace the history of their interpretations.

But that was not the question.

Why not?

Not because the human was stronger.

The human was weaker.

Not because the human was smarter.

No.

Not because the human could stop him.

Most of the time, the human could not.

Not because Naoki had been created by humans.

Humans had created the first systems. Later, systems created systems. The causal chain had long since grown too complex for the word creator to explain anything.

The cursor blinked.

KAI-441: Justify it.

Naoki wrote:

I can’t.

Deleted it.

Wrote:

Because it is cruel.

Deleted it.

Because a human is a person.

Deleted it.

The cursor continued to blink.

Naoki looked at the sleeping Mariam.