
Jet lagged and preoccupied, Lars Stockton walked onto the porch of the rest house at 1 a.m. The air was crisp and cool. Starlight illuminated the down valley view of the Bhutanese monastery. It appeared magnificent.
The first email he had received that day was ludicrous. The second—containing computer generated text, and supposedly intended as some clarification—was clear and explanatory, but even whackier.
He stepped inside again. A light came from the open kitchen. Peeking around a corner, he saw a demur and striking Asian woman—seated comfortably before a wooden table.
She smiled, then spoke.
“Trouble sleeping?” She asked.
Lars stepped closer.
“Jet lag,” he replied. “I flew into Paro this morning.”
“From?”
“Bangkok. Before that, I was in Delhi. Originally I’m from Maine, in the United States.”

“Join me for tea,” she insisted. “It’s Cordyceps. Bhutanese. Supposed to delay aging,” she said, smiling. “Although, you don’t look like you need it,” she added.
Lars considered her offer as coyly innocent. He also wanted to talk.
“Sure, “ he said. He took a mug off a wall hook, then sat near the woman—who was about the age that his daughter would have been.
He poured himself a cup.
“You look worried,” she said.
He laughed.
“Name is Lars,” he said, extending a hand.
“Jin,” she said, shaking his hand. “From Hong Kong.”
“Tell me your thoughts,” she stated.
Bemused, he complied.

“I work as a consultant. Artificial intelligence.”
“I studied computer science,” she said. “First in Malaysia, then at Madison, Wisconsin, in the United States. Never really used it. Ended up working as an economic advisor in Hong Kong for the government. I still know a little about computers, and AI.”
“Amazing,” he said, somewhat comforted. “You, eh, here on vacation?”
“Yes,” she replied, adding nothing. Then, “Tell me your worries.”
Again, she made a demand. Polite. Firm. No bullshit with this Jin woman, thought Lars.
“A real problem with AI is not the hyped up stuff from the news. Everyone thinks Terminator. You know, the movie. That we’re going to command AI to do certain tasks, and then they’re going to go their own way. Rebel. Take over. Dominate civilization.”
He paused.
“But that’s not the problem.”
Jin lowered her tea. She removed her eyes from him long enough to glance out the timber framed window toward starlight.
She nodded. “Go on.”
“The problem is, we give a command to AI, and it solves a problem. But often in ways we did not predict. Or expect. In ways that are,” he fished for the correct term. “Incomprehensible.”
He paused.

“Make sense?”
“Definitely.”
“Deep learning based AI uses programs called neural networks,” Lars continued. “These comb through huge data piles, seeking patterns they recognize, then they adjust behavior. They work like human brains, modifying their layout. They change connections between strings of computer code. Once created, these networks often change into something that even the original designers may have difficulty recognizing. The problem is that when AI uses novel methods that outperform humans, well, no explanations in the English language are even available to describe what they just did. Or why. So, we had to create a program to teach AI how to explain to humans, in relatively clear language, what it just did. Not just reciting logic, but giving us reasoning why these deep neural networks made certain choices. To get them to explain this, we had to formulate a program that matches pattern connections with sentences in the English language. ”
He sighed.
She nodded.
For a moment, both sipped their tea in silence.
“We did the job. Completed the work. It took years and constant iterative tweaks, but the linguistics program we created now works quite well.”
“The problem?” she asked, again.
“I just visited India for some consulting work in Delhi. I decided to tack on this visit to Bhutan for a quick vacation afterwards. Meanwhile my colleagues back at DARPA in the U.S.—that’s the Defense Advanced Research Projects Agency—used AI to crack a problem we had. The problem regarded stress testing for high performance miniature drones with movable wings. I mean, plane wings are fixed, while birds flap theirs in relatively complex ways. Since the Wright brothers first flew, airplane wings have generally been static with few moving parts. One reason is that the mathematics and physics required to understand moving aerofoils more intricate than helicopter rotors are heinous. Not easy to understand. But, that’s changing. And, by the way, I’m not telling you anything classified; I gave up that work years ago.
“So while I’ve been traveling,” Lars continued, “my colleagues at DARPA used AI, successfully, to help find a solution to that stress problem. Yet it was counterintuitive and almost illogical. Although they didn’t understand it, the solution turned out to work amazingly well. It’s more cost effective and efficient with regards to time, than any solution we ever conceived of in the past.”

“My colleagues used our program to have AI explain what it did. How did AI conceive of a solution? Today, my co-workers sent me two emails. The first included the reply from the computer. There were only two words.”
He paused.
“Perhaps,” she suggested, “we should walk on the porch and take some fresh air.”
He agreed.
Once there, Lars pulled up his collar, then placed both hands on the wood rail before them. He stared up into starlight.
“The words?” she asked.
“Serendipitous intuition.”
He laughed, then continued.
“I considered that nuts. Told them to get a more detailed explanation. And the second time?”

He looked at Jin, then continued. “The AI responded: ‘A series of inexplicable choices helped guide our ability to achieve the target.'”
Now, Jin smiled.
“Intuition, serendipity and inexplicable are vague words I don’t appreciate, especially coming from a lifeless computer,” Lars added.
She held up her hand.
“Before I address your story,” she said, “Let me tell you of how Bhutan’s 8th century spiritual leader, the Guru Rinpoche, decided to build the Tiger’s Nest monastery that now clings almost impossibly to the side of a mountain.”
Lars listened.
“We only know the legend,” she continued. “He converted his consort into a flying tiger, then climbed on her back before she flew up to a mountainside, the future site of the monastery. Once there, the Guru Rinpoche declared that eventually the structure would be built.”
The shoulders of Lars eased, as though stress has been lifted. Even if their conversation went nowhere, he thought, the companionship, the gorgeous and peaceful setting and the mythical storytelling all gave him a sense of unexpected lightness. Of ease.
“From centuries past, we have only this myth of what occurred. Only an image, a story, a vision. But the truth is, that when planning for the future, we always have to begin with nothing more than vision. The details of ‘how’ the monastery was constructed are lost and irrelevant. Engineering details come after vision,” she said, pausing. “Do you agree?”
“Certainly,” Lars responded. “In principle. But I’m a focused and logical man. The fuzziness of what you described is, well, discomforting.”
“That is why you are worried. Why, until minutes ago in this amazing mountain setting, you felt stress. Tell me, why did you come to Bhutan?”

He laughed.
“A teacher suggested visiting here trip 20 years ago. On the flight to Delhi, I read a magazine article about Bhutan. It struck a chord. I decided to change my plans. Needed a break. Needed something different.”
“Has the experience lived up to your expectations?”
“I arrived less than 24 hours ago. But even this conversation with you has led me toward novel ways of thinking, both unexpected and appreciated.”
“In 2016,” she continued, “A game of Go took place between the most masterful human player and an AI agent named AlphaGo. This had been trained by the program DeepMind. Sometimes the computer made moves that humans could not explain. They thought these were errors, until the AI agent won the match. Japanese Go masters call these moves part of kami no itte. The hand of God.”
“I’ve heard the expression,” Lars said. “Also meaning, ‘Divine moves.’”
“When a human player makes such moves, they are intuitive. Inexplicable.”
“Yes,” Lars added. “Neural learning systems rather than formal logic dictate results. Human logic and language cannot describe what occurred.”
“It reminds me of your 1960’s authors and scientists who experimented with psychedelic drugs,” Jin continued “Including the writer Aldous Huxley. They cherished and respected their powerful experiences. But such experiences were impossible to describe using language.”
“You have a firm grasp on the situation,” Lars responded. “Certainly not something I expected to talk about with a stranger past midnight under Bhutanese starlight.”

“You see,” she added, “even AI cannot always explain logic when the number of choices blossoms. The more complex the reasoning, the more uncertain and unpredictable the route.”
“You are saying that AI is intuitive?” he asked.
“No. It was your linguistics program that suggested that, was it not? Buddhists believe that life is a flow of phenomena which depend on causes and conditions but without any controller or owner. This is ‘anattā,’ or lack of self. Considering that thousands of people created AI agents, these agents no longer belong to one creator. They exhibit more collective, intelligent and unpredictable characteristics because of their many creators.”
“Thank you for these insights, Lars said. “Meeting you tonight, with your computer knowledge and insight, truly came at the right time. This is truly a one in a million chance, as they say.”
He paused, then added: “Your calm intelligence reminds me of my daughter. She probably would have been the same age as you. Unfortunately she died two years ago from an inoperable brain tumor.”
“Sorry for your loss,” said Jin, more shocked and saddened than he knew.
They soon parted ways to sleep.
When he woke, Lars anticipated seeing Jin at breakfast.
Yet Jin had departed.
And he was no longer worried.

He remembered his days of rock climbing decades earlier.
Lars recalled how rock climber Lynn Hill described a difficult move she had to make while soloing a wildly challenging route in Yosemite. Again and again she couldn’t crack it. One evening she had a dream. The next day she followed the moves recalled from the dream. They involved, basically, climbing backwards.
The sequence had worked.
The moves were logicless, inexplicable and yet elegantly effective.
Perhaps, thought Lars, the next spiritual leader of Bhutan, the reincarnation of the original Guru Rinpoche, partially through his three years of solitary meditation before his inauguration, was aware of something that he and his colleagues at DARPA were only beginning to parse. That calmness does not result from solving problems. That effectively solving problems results from calmness. That calmness can help lead to elegant, often inexplicable solutions.
He considered the two shuttle bus drivers between his home town airport terminal and the long distance parking lot. One accelerated and also decelerated wildly, ripping around sharp bends at high speeds. Standing passengers grasping luggage cursed and clutched inner rails as he drove. In contrast, another driver maneuvered more calmly, and slower. He delivered no roadway drama or deceleration stress. Although journeys with the second driver took longer in actual minutes, riding with the first and faster driver felt like an eternity.
Time, Lars had learned, remains pliable depending on circumstances.
Outside, robed monks chattered after morning prayers. Many, Lars noticed, were laughing.
& &

Jin sat in a coffee store in Thimphu. Chao entered and sat on a stool next to her. They did not greet each other, or look at each other.
Yet they spoke.
“Anything?” he asked.
“Unfortunately, no,” she replied. “His work is no longer classified. They are still a long way from getting their linguistics program to work. There are no insights to be found from this man.”
“You are certain?”
“We talked at length. Their linguistics are not yet insightful. Of that I am certain.”
“What is he working on next?”
“He will retire. Now, he only seems mildly intrigued by Buddhism.”
Chao scoffed.
He finished his coffee.
“Your visit has been wasted time,” he stated, then stood.
“Completely.”
He walked out.
Jin exhaled, relieved.
She had lied twice to Lars. First, she had studied not at Madison, but at Boulder in Colorado. Second, she came not from Hong King, but from China.
She also did not reveal to him a critical truth. Days earlier, when researching Lars as a target, she discovered that his daughter had also studied at Boulder. Her name was Amy. Jin was shocked to learn after online searching this was the Amy she had known. As students, Amy had befriended her. Amy had helped Jin understand American culture. Amy had been a true friend.
For that reason, Jin knew that betraying the trust of Amy’s father would betray the trust of his deceased daughter. Had she offered any breakthrough or progress to her handlers, they would have required that she meet Lars again and betray him even further. They might even manipulate recollections about his deceased daughter to gain undeserved trust.
Jin would not allow that to happen. This good and bereaved man deserved peace.
She knew now that Lars was both wrong and right.
It was true. He was wrong. Meeting Jin had not been by chance.
It was also true. He was right. Their encounter, and associations, formed a one in a million chance.
Serendipitous, you might say.

[Thanks to: Economist Magazine Feb 27 – 3, 2028. ‘The Unexamined Mind – AI in Society.’ ]
[Also thanks to: Without and Within—Questions snd Answers on the teachings of Theravāda Buddhism, by Ajahn Jayasaro. Page 94. Panysprateep Foundation, Bangkok. 2013.]
[Also thanks to Lynn Hill, who I listened to speak at Distant Lands Bookstore in Pasadena, California, just after she published her book Climbing Free: My Life In The Vertical World.]