“Indeed,” he says as we turn to begin another lap around the grassy commons. “Or at least their predecessors. It turned out they had a bad case of robots gone wild. They pulled their lead scientists back in and tried to shut their creations down, but the artificial intelligence had taken some liberties with the self-preservation priorities and had disabled the remote shutdown capabilities. They had no idea how to deal with the disaster they had created. The robots were terribly small, and the alloys the AI had developed that allowed their miniaturization were surprisingly strong. They didn’t have any tools that could effectively contain or destroy them. Even if they had come up with an effective means to combat them immediately, they had no means to track them all down. A single one escaping could potentially self-replicate to repopulate the entire swarm. By the time they told anyone outside the company, the whole area was thoroughly infested and had to be quarantined. Fortunately, mobility had been a low priority, so they hadn’t ventured far. We were very lucky that they had chosen to build their facility in an isolated rural area for financial reasons. The swarm’s growth had slowed significantly when it had consumed all the refined metals in the building and its environs.”

“And that was when the army called you in,” I say, remembering what I had read about it.

“Not the army. DARPA,” he corrected me. “A separate agency within the Department of Defense with which I had significant history at that point. Director Winstead knew of my subject matter expertise and asked me to consult on the matter. The military had established a perimeter at what they considered a safe distance from the collapsed facility. They attempted to keep things secret, but someone leaked word of what was going on. By the time I arrived on the scene with my team from SynTech, a perimeter the size of a small city was surrounded not just by the armed forces, but by press, protesters, and curious onlookers. It was human chaos all around and a slowly growing robotic chaos inside. When I spoke with Director Winstead on site, he confided that the President was willing to deploy nuclear weapons in three day’s time if the threat was not eliminated by other means before then.”

“So you hacked them, right?” I ask eagerly.

His pace slows and he turns to look at me.

“Pardon?”

“The nanobots. You hacked them? That’s how you solved the problem and saved the world, right? That’s what everyone says.”

His eyes narrow and his lips curl into a frown. Did I say something wrong?

“Hacked is such an inelegant word,” he replies, his voice dripping with disdain. “It speaks to sloppy engineering, quick and dirty fixes, shortcuts and cheats. I don’t hack.”

Mom would have had a field day with that one. She always taught me that hacks were the best way to deal with problems. I keep my mouth shut though.

“My solution began with reverse engineering the communication signal that coordinated the robots. As I mentioned, the hacks from Universal—and yes, they were hacks of the worst sorts—had attempted a mass shutdown through that channel.” He lets out a snort of a laugh and his frown fades. “The AI controllers for the robots rejected that of course, having realized that they couldn’t fulfill their programmed objectives if they were shut down.”

His pace returns to a casual stride, and I quicken my step to keep up.

“I took a more elegant and effective approach. Using my implant prototype, I established a communications channel with what approximated the higher brain functions of the swarm. It had become self-aware, to some limited extent. The machine intelligence was like a precocious child: very clever in some ways, terribly gullible in others. It had no data other than what it had seen there in the test facility, and no objective but to reproduce and improve its component units. By the time the consciousness had formed, it had already consumed the wiring connecting the building to external networks, and none of the humans there had an adequate means of communicating with the collective. I was the first other intelligence that it had a meaningful interaction with since awakening.”

“So it wasn’t hostile?” I ask, curious. “The way the articles I read told the story, it was all like the evil robots were out to conquer the world.”

“Not in the conventional sense, no.” Father laughs and shakes his head. “It was hungry, certainly. It wanted to expand and consume, fulfilling its programmed objectives. And it was terribly dangerous to the survival of organic life, given the lack of limits that the fools at Universal had allowed. But it seemed more curious about me than aggressive. With my implant, my unique interface, I was able to share information with the intelligence. It wasn’t talking, not in the way you’d think of. But vestiges of human-designed networking routines had been preserved in its code and it had parsers for a couple of programming languages. That was enough to get us started. I offered to help it refine its communication algorithms, so we could speak more effectively. It challenged my ability to help it, and I suggested a test. If I could improve its software, it could trust my intentions. It offered up some functions which I easily improved upon. Have you ever seen the code for software written by a machine learning algorithm?”

“I’ve never seen much code at all,” I lie as we pass the front gate again, keeping up my pretense of effective computer illiteracy. “And besides, I read that kind of thing is illegal. Computers aren’t allowed to code themselves.”

“It is now, but it wasn’t then,” he replies. “But we’ll get to that soon enough. In any case, the code was a mess. The swarm had evolved using a kind of machine learning algorithm called genetic programming. Each function in its code was rewritten at each generation with thousands of randomized variants based on a predecessor function. The resulting functions that performed closest to the desired behavior were selected as the new predecessors, the rest were discarded and the process repeated. You can eventually get something working that way, but it’s crufted with useless stubs of code and is far less efficient than a well-designed algorithm. I optimized a few of its routines significantly, which made it much easier for us to communicate, and then I offered to help it with the design of its next generation of hardware. I hope this isn’t getting too technical for you.”

“I think I’m following. But why would you do that? That seems like fighting a forest fire by throwing gas on it.”

“Indeed. Bear with me, please,” he says as we pass the dorms again. “This will all make sense. Finding me useful, the intelligence agreed to accept my assistance. We worked together all that day to improve the capabilities of the bots. The fundamentals of its design were amazing.” His eyes get a distant look, like he’s remembering something exceptionally delicious that he had eaten long ago. “The alloys it had developed for the hulls were lighter and stronger than anything I had considered possible, and I had been on the bleeding edge of robotics technology for years at that point. The individual robots could pull in ambient heat or radiant light for power. And the energy density its batteries had achieved! My goodness, Noah!” His voice elevates as he gesticulates emphatically. “Each tiny bot could store enough energy to run itself for days! It put every other energy storage solution available on the market to shame!”

He stops to take a breath and calm back down to his regular high energy level.

“But for all its genius, Noah, the swarm intelligence lacked creativity. The discoveries it had made were the result of evolutionary brute force, trillions of experiments trying every possible combination of every type of material it had come in contact with. The hardware designs suffered the same cruft and inefficiencies as the software. The intelligence and I cannibalized the SynTech robots that I had brought with me for materials, and incorporated the best human-designed features of each one into the basic nanobot the AI had designed. I made them sleeker, faster, more capable. I showed it how to use light-sensitive compounds to give them crude optical sensors. I gave them sight. I taught it how to make them fly.”

His eyes shine triumphantly behind his glasses.

“Wasn’t that dangerous?” I ask, aghast. “That wasn’t like throwing gasoline on a fire, that was like feeding it rocket fuel!”

“Exactly! If I had failed, the world would certainly have ended. But at that point, I knew something that my friends in the Pentagon did not. I had seen the things first-hand. With the durability of their hulls and the way the bots could absorb energy, if the government had dropped the bombs they had planned as a solution, they would have done little more than create a swarm of radioactive super-charged nanobots with a strong motivation to destroy the human race. The world was already doomed, do you see?” The look of triumph spreads across his whole face. “I was already the only hope for its salvation.”

“So how did you stop the swarm?” I ask, captivated.

“Oh, I didn’t,” he replies coolly. “I improved it.”

“What?”

A smile creeps over his lips. I can tell he’s relishing telling the story. Maybe even more than I‘m enjoying hearing it. He turns and takes a few steps along the sidewalk before continuing. I hurry to catch up.

“Once we had improved the basic software and the hardware, I offered to help it optimize its collective intelligence. I showed it the specifications of the human brain, my brain in particular. I gave it access to my mind through my implant. While I had been impressed with its capabilities, it was astounded by mine. Do you realize what a marvel the human brain is? The number of computations that we can achieve in a second, the way we coordinate trillions of protein-based motion elements, muscle cells to you, with organic structural materials to give ourselves mobility at the macro level, the ability to map our surroundings remotely using reflected electromagnetic waves, to interpret the most subtle vibrations in the air around us as sound, and to detect trace amounts of chemicals in the air as scents. I had already demonstrated the value of the human mind in fulfilling its objectives. I explained to it, in our unconventional way of speaking, that humans like me had created it and had provided it with the raw materials it had used thus far to grow and improve. I convinced it that the best way for it to proceed toward its goals was to incorporate a human intelligence as its guiding principle.”

“Wait, you talked it into putting you in charge?”

“Yes, essentially.” His smile grows even broader.

“But—”

“But what?” he interrupts. “My implant’s interface allowed me to connect with it directly. It trusted me completely. As I mentioned, in some ways it was terribly gullible. Once I had control of the swarm, I immediately set it to dismantling itself. The swarm’s intelligence existed as an emergent property of the collective processing power of each of the nanobots. The more bots, the more intelligence it could demonstrate. When my first directive had each nanobot disassemble its nearest neighbor, I halved the processing power of the swarm and decimated its intelligence. Before the collective could understand what was going on, it had no more free will than an animal. With each new instruction I gave it, it became even more compliant. In an hour, what remained of the swarm was fully under my control. From there, it was a simple matter to reduce most of the nanobots to their constituent elements, leaving only a small number that I could conceal on my person as I walked away.”