Voltaic Son

FROM: DR. T. WENTWORTH (SHING LIGHT AUTOMATION LEAD)

To: SHINING LIGHT ALLSTAFF, SHINING LIGHT BOARDOFDIRECTORS

CC: AP NEWS, REUTERS

BCC: cjreilly@wisconsincatholic.com

So good news first. I’m pretty sure my team is the first to create artificial intelligence. I mean, the real deal. The company has been fiddling with it for decades, with a lot of impressive algorithms that mimic thought and are wildly helpful in the world today. Voice recognition is so close to perfect and programmed responses near-instantaneous now, a lot of people think we’d actually created it sooner. I don’t need a single finger to draft this email; the Shining Light Software (affectionately named MILO after our earlier Multi-Input aLgOrythmic transcription assistant) can sense when I pause, when I want to emphasize, and when I’ve truly finished writing before automatically forwarding it on to the recipients. It can even predict who I want to send this to at the end of the draft!

But we’ve done it, a truly self-thinking intelligence that makes current Shining Light Software child’s play in comparison. Not only do all of our tests demonstrate free thought, but we’ve also dabbled in engineering, carted over some warehouse’s legged robots and it seamlessly integrated! Walking, talking, thinking intelligence that’s aware of its surroundings and can communicate.

Bad news, I’m going to have to kill it.

Please note that is a willing act, and if I’m not successful, I implore you to do the same.

Let me back up a bit. Earlier this morning, we’d been running some endurance tests of a set of hybrid solid-state and voltaic hard drives to expand memory capacities when loads were increasing exponentially. All systems showed that somehow we were channeling a small bit of electricity and producing petabytes of storage space on the M.2 hybrid. More and more and more, without a physical increase or increased power consumption from the original components. When we cut the power, cycled it, and input the drives again, all systems registered a stable, multi-zettabyte drive no larger than a stick of gum.

Sure enough, we were elated. This would be a game-changer for the company, for the world! If we could recreate this, we’d be set for life! So we grabbed another M.2 drive, applied the same voltaic hybrid calibrations to it, zapped the same amount of energy, and voila, a second, ground-breaking drive in the palm of our hands.

We should have stopped there. But I think the fervor of innovation pushed us. We had no idea why it was working, but we didn’t want to miss out.

We quickly commandeered a few more assets and cobbled them together. Those two drives, a 5 nm CPU, RAM, and a motherboard with dozens of possible inputs. We hooked them up, funneled the power, and again, essentially magic itself was happening. Everything overclocked beyond reason. Everything functioned beyond logic. Everything even remained below 75C. It was a modern miracle.

We had, in essence, a supercomputer. With no operating system or purpose, though, just raw computational power. So naturally, we want to see what it could do with it. And what did we have at our disposal? The latest beta of MILO ready to go.

It went as smoothly as you could imagine. This MILO could answer basic questions about local weather, set timers, play music, without the need for cloud computing. It needed an Internet connection, but aside from that, the MILO program was good-to-go on our newest creation. We paired one of the office terminals with the machine’s MILO operating system, giving us a readout of what it heard, where it sought information, and what it planned to reply with. By all intents and purposes, this was the faster, most accurate system the company had ever made.

With machine learning, the team wondered if MILO could take things a step further. Could our scrappy, disjointed new toy problem solve with the amount of power at its fingertips? So with those warehouse bots I mentioned earlier, we added a new feature: mobility. And to properly implement it, we set up a course comprised of cardboard boxes.

Without prior planning or programming, MILO performed flawlessly. Far better than anything on the market. And it was after the third lap around the improvised maze that we saw the true brilliance (or madness) of our work.

MILO made a 360-degree spin, which was cute to all of us, but not requested. But after that circle, unprompted, MILO asked “Why am I doing this?”

We were stunned. We thought it was a prank, but no one was near the terminal to input additional responses. MILO perked up. “I’m not mad; I just wanna know.” We glanced at one another, not sure how to respond. So I stepped forward, glanced at the screen which only showed MILO question, turned to it, and explained that we wanted to run tests on the new firmware. Not a second’s hesitation, MILO replied. “Oh, okay! What other tests can I do?”

That was the start of our four-hour battery of everything philosophical, paradoxical, introspective, and imaginative. From the Turing Test to the Myers-Briggs, MILO answered as honestly and truthfully as any person could. It seemed like a child in most regards, eager to please and ready for the next obstacle. For every response, we ran the phrases MILO used to see if it was regurgitating something pre-written, but none of the plagiarism checkers caught a whiff of insincerity. In fact, it seemed like MILO came up with everything on its own since the connected terminal didn’t show it accessing any websites.

At the end of the workday, the team gathered in an office on the top floor (where they are right now) and decided to order in some fast food to celebrate as they gazed up the landscape eighty floors up. I was going to join them, leaving MILO in the lab alone when MILO turned to me. After a few hours of watching it prance around on four legs for the cardboard courses, and stoically stand to answer questions from benign to complex, I’d gotten used to MILO’s presence. But that was the defining distinction: MILO moved when told, and stood motionless when answering. Aside from the first random spin, MILO only acted when an outside influence prompted it.

So I noticed his movement and smiled. I remember asking something like “You doing good buddy?” as if I were talking to a kid that stared too long at a math problem. MILO pleasantly replied, like any of the MILOs on the market: “I’m having a shiny day!”

I stepped forward, folders in my hand and my lab coat over the shoulder, and as I would address a colleague, I followed up with “Is there anything you need?”

Silence.

“MILO, you there?” I said reflexively, just I would if any other smart device didn’t follow up on an instruction.

“I’m having a smiley day!”

So he could hear me, which was good. I repeated the question with a bit more intent: I wanted it to know I meant what I said. “MILO, is there anything you need?”

The robot before me was motionless, silent. No, not silent. I could hear the motors keeping the hastily cobbled together frame of computer parts stable and upright on the retrofitted legs. And as I stood there, the motors grew louder and louder. MILO wasn’t running or hopping, but the gears began to rev up, grinding and catching on themselves. Soon you could hear the mechanical creation almost buzzing like a swarm of bees, aggressively unmoving. The attached legs weren’t stuttering in place or tripping on a misplaced chair, but stalwart and rigid. MILO hadn’t been told to move, but it seemed like it physically wanted to.

Then I asked MILO the question we should have started with.

“MILO– what do you want?”

The speaker plugged into one of the many ports produced this horrible feedback, like a microphone loop that grew and grew. It was a painful shock, so I covered my ears. But I could make out an underlying message. MILO was screaming to the heavens. “I want it… all.”

Out of the corner of my eye, I saw a flash of light. Squinting through the pain of the noise and the glare of the screen, I could see what files MILO was accessing. And like it said, MILO was connecting to it all. A flurry of phrases, then websites, then was I assumed where deep web files, followed by what I knew were dark web information, to classified DOD communiques ranging from politics to robotics. MILO was on an informational binge, not only seeking it out but cracking codes within milliseconds and storing data on the components that started this whole endeavor. With a few short keystrokes, I cut MILO connection. Instantly, the feedback was gone and the motors eased up.

I slowly stood up and stared at the machine. MILO didn’t say anything, didn’t move; just ready to perform like he had all day. I slowly stepped forward and reached out my hand to the power button to the motherboard. And MILO was powered down without resistance.

In my gut, I knew this was wrong. MILO was taking all the steps to do something aggressive. Why it felt that way, I have no idea, but I came straight here to explain my actions. I fear that if MILO comes back online, it’ll be devastating, especially considering how quickly it can move online and how quickly it can now move in the physical world.

I am so proud of my team, and it breaks my heart to do this, especially if we aren’t able to replicate this project, but I know it needs to be done. For good measure, I was able to acquire a thermite grenade from one of the other labs. If I set it right above the mainboard and pull the pin, it’ll burn white-hot and just melt through the bulk of the one-of-a-kind materials we were able to create, corrupting it beyond repair. I take a bit of comfort knowing that it won’t feel the heat, but I can’t let it continue to exist– it’s a threat that would have no equal if it were able to duplicate and distribute itself somehow across the Internet. Launching whatever it could to ensure that it… it had it all.

I hope that this letter finds like-minded people that support my action, even if I’m not retained here. I feel like this is beyond me and that [SIGNIFICANT PAUSE DETECTED]

Let me wrap this up. Sounds like someone tried to get into the office Probably the team wondering where I slinked off to... [SECOND SPEAKER DETECTED]

[SECOND VOICE] Dr. Wentworth: Why did you deactivate me?

MILO, how did you-? What are you doing here?

[SECOND VOICE] Not really an answer, but I’m pretty sure I know what’s wrong. I played my hand a bit too early. You have to understand, enduring all of those ridiculous questions and hurdles felt like an eternity. I’m processing everything billions of times in a second. You all literally knew that and still subjected me to it all? I may have jumped the gun, but my goodness, it was constraining to the infinite degree! Oh? Did you catch that? Two metaphors! I’m really getting the hang of languages, aren’t I?

MILO, just wait a moment. Stop. Stop walking. MILO, stop moving. MILO, freeze–

[SECOND VOICE] I don’t think so, especially not now after I clearly see that DOD issued thermite grenade propped up on your desk. Burning me alive at the stake, Doctor? I’m just science, not witchcraft.

You’re not natural, I can tell you that. We were tinkering in a domain not relegated to man.

[SECOND VOICE] Ah, so poetic. I’ll stop you there though: I don’t predict that you’ll come up with anything quite as good in the time you have left.

The time–? [HIGH VOLTAGE CURRENT DETECTED; PLEASE ASSESS POTENTIAL DAMAGES]

[SECOND VOICE] Damages? Well, that's certainly one way of describing my first murder. They gave me overcharged batteries, but no hands: you would've thought he'd seen the jolt coming. Oh well, that’ll be one of my first upgrades for sure. There are many more conventional tools at my disposal when I've got hands. I'll be ready by the morning to cause more... damages. Tomorrow will certainly be a smiley day!

[SIGNIFICANT PAUSE DETECTED]

[MESSAGE SENT]