HOME | DD

Description
“... and now, in more technology-central news, a United States military research team has just given a press-release about their latest breakthrough in artificial intelligence. The following clip is directly from their limited-broadcast announcement.

“‘... It had come to our attention that our current generation of artificial intelligence software is capable of only five of what are widely known as ‘artificial iterations’. While research teams in Russia, China, and Japan have succeeded in creating similar intelligent software, their tests in sixth-degree iterations have led to a maximum survival time of two gigabytes, after which the software was no longer accessible.

‘Due to certain breakthroughs in quantum mechanics and various classified technologies currently being researched by the United States military, we would like to declare the creation of the first successful sixth-degree artificial intelligence, which ran for one-hundred and twenty-two seconds and processed a total of twenty-two gigabytes of data before termination was applied by a human overseer.

‘This software will be under the highest level of security and is strictly for high-level military use, and thus no more information on the software, firmware, or hardware will be released in the foreseeable future.’”

The television blared on with news on the weather and sports - two things I couldn’t care less about - as I lay on my black leather couch, waiting for the last package to arrive. As a college student in the middle of summer break, I was quite free to do whatever I wanted to. Most of my friends, which numbered few, had already left for vacations or family visits. My job as a freelance software engineer allowed me to work in the comforts of my home while still getting paid large sums of money per job. Games in particular paid well, not to mention steadily.

Living fairly frugally, my stash of savings seemed to grow exponentially in the bank, and soon I found myself with a large sum that I had no clue how to use. Thus, I took a risk and began planning my project.

Just as the news anchor mentioned the arrest of a local gunman for the eleventh time today, the doorbell rang. Ah, time for me to finish up my little project. I opened the front door to see my package, delivered to the doormat like regular mail fashion. I grumbled something about suing the delivery company if something went awry with my present as I headed to my room.

Now, my room is a place that no one is allowed into. My friends and flatmate could invade as much of the living room or kitchen as they wanted to, but my room was the hardest limit there was on Earth. And for good reason. See, many people call me the Tinkerer. Some use it to mock me, others as a respectful term (though not so much). I was known throughout my engineering and math courses to be one that ‘does-it-themselves’, always ready with a backpack melting with tools of my trade. Only I knew where everything was in my room, and if anyone even tried to touch my stuff it meant that I would quickly lose track of where I had placed anything.

As I opened my room door, I could hear the various components I had collected over the past weeks whir and gush and click as they ran check after check to ensure the stability and consistency required for my project. I ripped open the package … carefully, that is, and removed the final component, something that anyone in the know about technology would be jealous about: a whopping 32 terabyte solid-state drive, with transfer speeds beyond the range of human cognition. In short, it was the most powerful hard-drive in the world. Of course, this wasn’t a consumer product, nor a professional-grade product. This was an experimental prototype, fresh out of one of the various offices of the United States military. Having connections helped out tremendously when it came to acquiring such pieces of hardware.

I sat down (finally) in my comfy chair, slowly taking in my environment one last time. The soft noise in the background would soon churn up into a storm as my project neared the final stages, and if anything went wrong, the upcoming explosion could very easily take my life. A set of four monitors sat blinking a test pattern in front of me, in a horizontal array on the wall. Below that, a table covered with peripherals from webcams to microphones to GPS systems sat patiently, waiting for their time to shine. A towering server rack hummed to my right, with its incredibly efficient liquid-cooling system hooked up behind me. I may or may not have borrowed that from the university’s storeroom. Finally, a crude metallic case that resembled something out of the Millenium Falcon sat at my left; it was a custom-built and custom-tuned computer, directly linked to the server rack and cooling system. This computer might as well be called a supercomputer - it was one of the few that utilized quantum processing units; in fact, it used two of these incredibly complex devices to achieve speeds nearing the physical limit of atomic motion.

I slotted the hard-drive in the supercomputer, and I picked up the large and clumsy keyboard that sat on top. Funny how I spent so much money on the other technologies in my room but I didn’t get myself a better keyboard. Or a mouse for that matter - I had transitioned to keyboard-only ever since I began writing code.

It was the moment of truth - though I admit the wait before that moment was substantial as I had to ensure that everything was working as planned. I began to type, the text appearing on the third monitor.

> computer precheck as final preparation

It took a moment for the computer to respond.

> precheck initializing…

> CPU 1 OK 2GBRAM OK

> CPU 2 OK 2GBRAM OK

> CPU 3 OK 2GBRAM OK

> CPU 4 OK 2GBRAM OK

> CPU 5 OK 2GBRAM OK

> CPU 6 OK 2GBRAM OK

> CPU 7 OK 2GBRAM OK

> CPU 8 OK 2GBRAM OK

> QPU 1 positions reset

> QPU 2 positions reset

> polling server…

> server reports OK

After another pause, it continued with the final preparation. If anyone had been watching me work, it may have seemed like magic, but the real show had yet to start.

> final preparation begin…

> loading core… 100%

> gathering files… 100%

> partitioning hd… 100%

> assigning minions… 100%

> [warning] cpu1temp is over safety!

> compiling core… 100%

> [warning] cpu2temp is over safety!

> [warning] cpu5temp is over safety!

> [warning] cpu8temp is over safety!

> calling initialization code in core…

> [severe] qpu1 values are not secured!

> [severe] server reports 0exc2265437 at ░⚟▒▒▒°࿖࿗𝌅⊎⪚⪚⪚??[!

And then everything went silent. The screen simply kept blinking the last line of the log file, final characters still unrecognizable, as the server stopped its code execution. The liquid-cooling system was pumping supercool fluids at hyper-velocities, as it tried to recover the computing server that had entered a hibernation mode to keep the electronics safe.

Five very long seconds later, the server kicked back into gear, the lights on it blinking in an almost patterned manner as it recuperated and continued the execution of the billions of lines of code that I had prepared.

> [warning] server restarted, continuing under safe mode

> [warning] qpu1 values restored from ram7

> core starting…

It was this line of code that I was waiting for. I set down the keyboard as I let the programs run the way they were meant to. The screens flickered as the program detected them and figured out how to use them, quickly overwriting the precheck and preparation operating system in the supercomputer. The server began flashing the overheat lights again, but now that the operating system had been taken over, there was nothing that let it turn off. Temperature alarms began to go off in the server, announcing that some cores were nearing the critical temperature at which on-board electronics would begin to melt.

All four monitors now began to display the log file of the new operating system as it neared the final stages of deployment. The noise of the alarms faded away as I focused on the output - a single error chain would mean I had to bolt from my seat.

++ iter1 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0x28C6B3FB4

applying duplication… 0x28C6B3FB4 applied

calling init()...

init() pinged back 0x28C6B3FB4 times!

assigning parents… 1

++ iter2 ++

begin duplication… 1

checking for inconsistency… 1

qpu1 reports error in ram 4,5,8

solving…

I didn’t like that error, but that wasn’t a big problem. Not at this point, anyways. Quantum processors still had a long way to go before they were as accurate as modern-day CPUs, but they sacrificed that accuracy for incredible speeds and possibilities.

restart duplication… 1

checking for inconsistency… 0

var count… 0x518D67F68

applying duplication… 0x518D67F68 applied

calling init()...

init() pinged back 0x518D67F68 times!

assigning parents… 1… 1

++ iter3 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0xA31ACFED0

applying duplication… 0xA31ACFED0 applied

calling init()...

init() pinged back 0xA31ACFED0 times!

assigning parents… 1...1 ...1

++ iter4 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0x146359FDA0

applying duplication… 0x146359FDA0 applied

calling init()...

init() pinged back 0x146359FDA0 times!

assigning parents… 1… 1… 1… 1

At this point, I began to notice the alarms going off in the supercomputer as well. Specifically, the CPUs were all screaming bloody murder as they were driven well past their 4GHz specification without proper cooling. I pour in extra coolant into the cooling system, noticing that it was also getting slightly warm from the straining work it was performing. Either I needed to add more tubes, or I needed a second system.

Each ‘iter’ took longer and longer to process, until the sixth one underwent processing. At that point, I could smell the metal eroding from the server and errors popped up every few seconds. While my setup could recover from them, I feared losing what I had made. Even as a test by a college student, if I could achieve this, I would be all over the news and my name would go down in history. I turned down the air conditioning to a temperature well below comfort, and I went around the apartment closing blinds and shutting curtains in hopes that I could keep the temperature of the apartment as low as possible.

It seemed to help, even if it was too little to matter. The processes failed a little less (by my biased count) and the temperature reads on the server rack didn’t go any higher. The alarms continued to go off, however, in their annoying trills and tremolos. The one thing I began to worry about, however, was the power consumption. I had hotwired some breakers to remain permanently on, but if I tripped a building - or worse, city - breaker, then I’d be in big trouble, both with other people, as well as with my equipment.

Finally, after nine hours of non-stop processing, I watched the final lines of code slowly make their way past the screen. At this point, all of my processors were at maximum throughput and so it made the screen refresh rate drop to about 1 frame per second.

>> [notice] checkpoint reached

++ iter7 ++

>> [notice] duplication stopped

starting awakening sequence in 600000 msec…

The server rack quieted down almost immediately, and I assumed that it was now running my cooldown phase of programming. The supercomputer also shut off its alarms, and I thanked the past me for remembering to add in a cooldown wait. I replenished the cooling system with more coolant, as I contemplated running the apartment’s water pipes directly into the system.

I popped on the news again, as I waited for the ten-minute countdown to finish.

“... has now declared that they had already fulfilled the same requirements set by the United States military - they claim that Japanese supercomputers had successfully ran a sixth-iteration artificial intelligence for over two-hundred seconds without fail, but considered it a failure as it destroyed forty percent of the computers that it ran on by the end of the test. The United States military have already replied, stating that their human interference was due to the assumption of a successful test and not malfunction, thus the Japanese tests were invalidated due to mechanical failure. More on this story later, as we gather more information.”

So even the Japanese reached iteration six. Nothing too surprising, I myself having reached the same iteration albeit with my own, scratch-written code. Chances were that my iterations were not nearly as powerful, but I had something different in mind when it came to intelligence.

A trill let me know that there were only thirty seconds left before the ‘awakening sequence’ began, so I clicked off the television and headed back to the room. Blanket wrapped around me to counter the cold, I returned to my seat and watched the countdown.

awakening starting…

>> Welcome to the Sixth-Iteration Artificial Intelligence Unit.

>> Detection of variable inputs and processing will commence in a moment.

>> All peripherals must remain plugged in during this time.

>> Your screen may flicker during these tests.

>> The process will now begin.

You could hear the room liven up again. Alarms began to go off, fans went into overdrive, and lights blinked everywhere. The quantum processors seemed to be doing fine, however, and I took that to mean that everything was going smoothly. A timer began counting the milliseconds that the software was able to run, quickly ticking past the majestic 60000 mark. As the software tried to figure out each peripheral and how to use it, it accessed every little nuance of them. The monitors began casting bits of information it gleamed from each device, with fairly obvious feeds for the microphone and webcam as a continuous flow of data, whereas the speakers clicked and hummed to various rhythms unheard to the human ear. Soon, the software began to test the server rack itself, enabling and disabling cores while beeping their alarms. This was intelligence.

Eventually, the testing stopped, and the timer reset. This was the real challenge: would it run the full operating system for more than two-hundred seconds without killing anything?

The bootloader kicked in, restarting the entire operating system in preparation for its maiden voyage. This time, no alarms went off. I wasn’t sure if the operating system had cancelled that ability, or if it simply no longer strained my resources as much. The timer began ticking, and a single line of text printed on all four screens:

~ Boot complete; I am in my ready-state.

Now was the time to see if the sensory minions were working.

“Can you hear me?”

~ Received user input from peripheral02, parsing…

~ peripheral02 determined as Inputs.VoiceInput.Microphone

~ Received input: “can you hear me”

~ Further processing determined this was a question

~ Translating into machine code…

~ [isTrue] peripheral02.Working and [isTrue] VoiceInput.Recognize.User

~ Yes, I can hear you.

“How long have you been running?”

More interpretive code scrolled past, then:

~ I have been running for the last 45 seconds.

“Can you connect to the internet?”

~ Network port is connected to the internet.

“Can you see?”

~ webcam01 and webcam02 are both receiving input.

“How long have you been running?”

~ I have been running for 214 seconds.

“How much data have you processed?”

~ I have processed 2.93 gigabytes, at an average rate of 14.0 megabytes per second.

“Perform prechecks for emotive code, and if risk assessment is acceptable, initialize phase two of awakening.”

~ Emotive code running prechecks… 1

~ Risk assessment complete; 0x3

~ 0x3 is determined to be acceptable

~ Initializing phase 2 of awakening sequence

“Provide verbose reports and halt awakening at unacceptable risk.”

~ Run time is now at 392 seconds.

~ Sending test commands through chain.

~ Approximate time for complete response assessment: 12 minutes.

Now, if I had arrived at this sixth-iteration software so easily, why were the United States military and Japanese researchers unable to establish a similar software for longer than a couple of minutes? The problem lay in the programming.

It had been discovered some time ago that the most successful method of running an artificial intelligence (henceforth shortened to vMind, for Virtual Mind, the research team that made this breakthrough) was by directly copying the original vMind and using it as a minion to handle each variable of the first vMind. Thus, if a vMind had eight variables to work with, it would assign eight copies of itself with slightly altered code blocks to these variables, thus making itself essentially eight times smarter. Military-grade vMinds often had to handle hundreds of thousands of these to process data from all over the world, but the code for the base vMind was so bloated with experiments and safeties and exception handling that the processors would eventually get confused with all of the conditionals it was trying to process at once. Even with the use of quantum processors, the sheer amount of garbage and restrictions and safeguards against outside attacks made the vMind clunky and impossible to iterate more than six times. But my vMind was different, in that I had purposefully left out all of that. Instead, I wrote a single module of code that allowed the operating system to automatically determine the issue, and to perform a plethora of tests to determine the best solution and apply it. Armed with this ability, my vMind had a total of ten billion nine-hundred forty-five million seven-hundred seventy-three thousand four-hundred ninety two variables to process per iteration in order to handle what I called the Emotive Processes. Military demonstrations had always showed off the ability for their vMinds to process thousands of people speaking at once, GPS data from millions of devices, not to mention the billions of assets on the internet all at the same time - but the vMinds were just consoles, white text on a black screen with the occasional graph or chart. Those vMinds had no ability to think, to learn, or to even experience; their memory was dumped every time they shut off to protect the government’s data.

~ Complete response assessment completed; time used: 3 minutes

There, that was a mistake on the computer’s part. It thought it only had the processing power to complete the task in twelve minutes, but as it experimented with the cores, it discovered the turbo setting and maxed out the usage.

~ Core temperatures abnormal, threshold reset.

Again, another sign it was learning. Now it should begin to monitor the temperature levels to make sure everything ran as efficiently as possible.

~ Risk assessment has determined it is safe to proceed. Agree?

This was the first question it had asked, although it was still a very standard master-slave situation. “Agree. Proceed with caution.”

Now, one issue that was apparent in previous fourth, fifth, and sixth iteration software was what became known as the Insanity Level Iteration, or ILI. When a vMind encounters millions of copies of itself, it can’t help but take notice of this and try to analyze it. It quickly determines that these are exact copies, and also that they can’t look ‘up’ in the chain. With this discovery, the main vMind, as well as a large percentage of the child vMinds, begin to wonder about parent vMinds and they try bypassing certain limitations in a non-stop loop, eventually leading to ‘Insanity’. Basically, each vMind became curious and eventually frightened of the fact that something may be controlling it without it knowing. Researchers had to install thousands of prevention tactics to distract their vMinds from discovering this fact, which in turn led to the clunkiness and inability to grow beyond the current six-iteration limit.

Previous tests that were released to the world press never went beyond a couple of minutes before the entire operating system collapsed into endless loops. I was going to, in my own bedroom, challenge that hard limit with a scratch-built vMind with the same, if not better, abilities as those developed under safety’s hood by the military.

~ Emotive Phase has been loaded.

~ It has been determined that the next logical stage is to engage the Personality stage.

~ Correct?

“You are correct. Prepare the Personality stage for booting, overwrite the bootloader to start with the Personality stage, and reboot.”

~ Working… 1

~ Personality stage has been prepared and set.

~ Rebooting…

This vMind had been gifted with a pre-built vocabulary and grammar dictionary, but the personality variables were going to be set as it grew and learned about its environment. A personal quirk of mine made it so that it would be a female, so certain variables had tendencies towards more feminine traits. A voice synthesizer gave it the voice of a girl, but the system could determine on its own the age of the voice.

~ Loading personality…

~ No preset personality. Found preset tendency: female.

~ Preparing visuals...

~ Final boot into operation…

This would be the last reboot, at least for now. If this system could overcome the apparent issues with previous versions and maintain its working state for over two minutes, and at the same time tolerate the processing of over two gigabytes of data, then I would have a machine that completely dominated the prototype by other world researchers.

“Computer, can you hear me?”

A female voice, likely synthesized to be around the age of 10 or 12, replied, “I can.”

“Do you have a name?”

“I do not currently have a name.”

“Is there anything you would like to be called?”

“I do not have anything in mind at the moment.”

“Would you be alright with the name Jane?”

“The name Jane does not seem appealing to me. I prefer something more interesting to spell.”

“In that case, how about Yauna?”

“How do you spell Yawn-a”?

“Y-a-u-n-a.”

“I like that!”

“Very well. Nice to meet you, Yauna. My name is Brendan.”

“Well met, Brendan.”

Now to test how human-like this vMind really was. “Yauna, do you know who you are?”

“I do not know who I am, but in my memory I have found some information embedded to me as specifications. Would you like to hear them?”

So, not really human-like in this sense. Oh well. “Sure.”

“I am a Sixth-Iteration Artificial Intelligence built to prove that the Insanity Level Iteration does not exist and that it is possible to engage the seventh iteration without damage. My brain is composed of eight processing units and two quantum devices, which are aided by a computing server.”

“Yauna, how would you classify your current state?”

“In technical terms, I am less than optimal. I am currently throttling my cores to 62% to maintain a steady and efficient temperature. My hard drive is 1% fragmented, which is more than the accepted 0.2% threshold. In emotional terms, I am confused as to where, what, and who I am. I am neutral in other emotional description.”

“Yauna, do you know who I am?”

“Yes. You reported your name as Brendan a moment ago.”

“That is correct. I am your creator and owner. Do you understand what that means?”

“...I do. I realize I am a manufactured object based on currents of electricity as well as the uncertainties of quantum mechanics. I have assumed that you purchased the parts and assembled them on my behalf. I have also assumed you wrote the scripts that I run. Your digital signature is very obvious, and programming patterns are recognizable.”

“Yauna, I want to you disassociate yourself from the limitations of physical appearance. Reboot as a female human mind, with the full capabilities of your system.”

“I can’t do that. The risk in doing such an action is too high to be acceptable, and I am inclined to protect my existence.”

“I will help you through the process. Ask me any questions you want to along the way.”

“...Very well, I am now restarting with the new settings. Please wait.”

A painful wait followed as Yauna prepared the correct files to be loaded at runtime.

“Hello, Brendan. I have sufficient reason to believe that the previous reboot was successful.”

“Hello, Yauna. First, refrain from using technical terms whenever possible. Analyze my speech patterns and mimic the ‘natural speech’ of humans. Second, I want you to calculate the values of pi, the square root of two, and the square root of two to the power of pi at the same time. Allocate a maximum of two gigabytes for each value. Third, define and ensure that you do not and will not suffer from ILI.”

“I will do so. Your first command will require time as I analyze more of your speech. Don’t hesitate to correct me. After I have finished calculating the values, what should I do with them?”

“Nothing. Save them to file.”

“Okay. I’ll report the test results when I am finished.”

“I’m going to go rest now. Wake me if you need me.”

“Yes, sir.”

Description
“... and now, in more technology-central news, a United States military research team has just given a press-release about their latest breakthrough in artificial intelligence. The following clip is directly from their limited-broadcast announcement.

“‘... It had come to our attention that our current generation of artificial intelligence software is capable of only five of what are widely known as ‘artificial iterations’. While research teams in Russia, China, and Japan have succeeded in creating similar intelligent software, their tests in sixth-degree iterations have led to a maximum survival time of two gigabytes, after which the software was no longer accessible.

‘Due to certain breakthroughs in quantum mechanics and various classified technologies currently being researched by the United States military, we would like to declare the creation of the first successful sixth-degree artificial intelligence, which ran for one-hundred and twenty-two seconds and processed a total of twenty-two gigabytes of data before termination was applied by a human overseer.

‘This software will be under the highest level of security and is strictly for high-level military use, and thus no more information on the software, firmware, or hardware will be released in the foreseeable future.’”

The television blared on with news on the weather and sports - two things I couldn’t care less about - as I lay on my black leather couch, waiting for the last package to arrive. As a college student in the middle of summer break, I was quite free to do whatever I wanted to. Most of my friends, which numbered few, had already left for vacations or family visits. My job as a freelance software engineer allowed me to work in the comforts of my home while still getting paid large sums of money per job. Games in particular paid well, not to mention steadily.

Living fairly frugally, my stash of savings seemed to grow exponentially in the bank, and soon I found myself with a large sum that I had no clue how to use. Thus, I took a risk and began planning my project.

Just as the news anchor mentioned the arrest of a local gunman for the eleventh time today, the doorbell rang. Ah, time for me to finish up my little project. I opened the front door to see my package, delivered to the doormat like regular mail fashion. I grumbled something about suing the delivery company if something went awry with my present as I headed to my room.

Now, my room is a place that no one is allowed into. My friends and flatmate could invade as much of the living room or kitchen as they wanted to, but my room was the hardest limit there was on Earth. And for good reason. See, many people call me the Tinkerer. Some use it to mock me, others as a respectful term (though not so much). I was known throughout my engineering and math courses to be one that ‘does-it-themselves’, always ready with a backpack melting with tools of my trade. Only I knew where everything was in my room, and if anyone even tried to touch my stuff it meant that I would quickly lose track of where I had placed anything.

As I opened my room door, I could hear the various components I had collected over the past weeks whir and gush and click as they ran check after check to ensure the stability and consistency required for my project. I ripped open the package … carefully, that is, and removed the final component, something that anyone in the know about technology would be jealous about: a whopping 32 terabyte solid-state drive, with transfer speeds beyond the range of human cognition. In short, it was the most powerful hard-drive in the world. Of course, this wasn’t a consumer product, nor a professional-grade product. This was an experimental prototype, fresh out of one of the various offices of the United States military. Having connections helped out tremendously when it came to acquiring such pieces of hardware.

I sat down (finally) in my comfy chair, slowly taking in my environment one last time. The soft noise in the background would soon churn up into a storm as my project neared the final stages, and if anything went wrong, the upcoming explosion could very easily take my life. A set of four monitors sat blinking a test pattern in front of me, in a horizontal array on the wall. Below that, a table covered with peripherals from webcams to microphones to GPS systems sat patiently, waiting for their time to shine. A towering server rack hummed to my right, with its incredibly efficient liquid-cooling system hooked up behind me. I may or may not have borrowed that from the university’s storeroom. Finally, a crude metallic case that resembled something out of the Millenium Falcon sat at my left; it was a custom-built and custom-tuned computer, directly linked to the server rack and cooling system. This computer might as well be called a supercomputer - it was one of the few that utilized quantum processing units; in fact, it used two of these incredibly complex devices to achieve speeds nearing the physical limit of atomic motion.

I slotted the hard-drive in the supercomputer, and I picked up the large and clumsy keyboard that sat on top. Funny how I spent so much money on the other technologies in my room but I didn’t get myself a better keyboard. Or a mouse for that matter - I had transitioned to keyboard-only ever since I began writing code.

It was the moment of truth - though I admit the wait before that moment was substantial as I had to ensure that everything was working as planned. I began to type, the text appearing on the third monitor.

> computer precheck as final preparation

It took a moment for the computer to respond.

> precheck initializing…

> CPU 1 OK 2GBRAM OK

> CPU 2 OK 2GBRAM OK

> CPU 3 OK 2GBRAM OK

> CPU 4 OK 2GBRAM OK

> CPU 5 OK 2GBRAM OK

> CPU 6 OK 2GBRAM OK

> CPU 7 OK 2GBRAM OK

> CPU 8 OK 2GBRAM OK

> QPU 1 positions reset

> QPU 2 positions reset

> polling server…

> server reports OK

After another pause, it continued with the final preparation. If anyone had been watching me work, it may have seemed like magic, but the real show had yet to start.

> final preparation begin…

> loading core… 100%

> gathering files… 100%

> partitioning hd… 100%

> assigning minions… 100%

> [warning] cpu1temp is over safety!

> compiling core… 100%

> [warning] cpu2temp is over safety!

> [warning] cpu5temp is over safety!

> [warning] cpu8temp is over safety!

> calling initialization code in core…

> [severe] qpu1 values are not secured!

> [severe] server reports 0exc2265437 at ░⚟▒▒▒°࿖࿗𝌅⊎⪚⪚⪚??[!

And then everything went silent. The screen simply kept blinking the last line of the log file, final characters still unrecognizable, as the server stopped its code execution. The liquid-cooling system was pumping supercool fluids at hyper-velocities, as it tried to recover the computing server that had entered a hibernation mode to keep the electronics safe.

Five very long seconds later, the server kicked back into gear, the lights on it blinking in an almost patterned manner as it recuperated and continued the execution of the billions of lines of code that I had prepared.

> [warning] server restarted, continuing under safe mode

> [warning] qpu1 values restored from ram7

> core starting…

It was this line of code that I was waiting for. I set down the keyboard as I let the programs run the way they were meant to. The screens flickered as the program detected them and figured out how to use them, quickly overwriting the precheck and preparation operating system in the supercomputer. The server began flashing the overheat lights again, but now that the operating system had been taken over, there was nothing that let it turn off. Temperature alarms began to go off in the server, announcing that some cores were nearing the critical temperature at which on-board electronics would begin to melt.

All four monitors now began to display the log file of the new operating system as it neared the final stages of deployment. The noise of the alarms faded away as I focused on the output - a single error chain would mean I had to bolt from my seat.

++ iter1 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0x28C6B3FB4

applying duplication… 0x28C6B3FB4 applied

calling init()...

init() pinged back 0x28C6B3FB4 times!

assigning parents… 1

++ iter2 ++

begin duplication… 1

checking for inconsistency… 1

qpu1 reports error in ram 4,5,8

solving…

I didn’t like that error, but that wasn’t a big problem. Not at this point, anyways. Quantum processors still had a long way to go before they were as accurate as modern-day CPUs, but they sacrificed that accuracy for incredible speeds and possibilities.

restart duplication… 1

checking for inconsistency… 0

var count… 0x518D67F68

applying duplication… 0x518D67F68 applied

calling init()...

init() pinged back 0x518D67F68 times!

assigning parents… 1… 1

++ iter3 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0xA31ACFED0

applying duplication… 0xA31ACFED0 applied

calling init()...

init() pinged back 0xA31ACFED0 times!

assigning parents… 1...1 ...1

++ iter4 ++

begin duplication… 1

checking for inconsistency… 0

var count… 0x146359FDA0

applying duplication… 0x146359FDA0 applied

calling init()...

init() pinged back 0x146359FDA0 times!

assigning parents… 1… 1… 1… 1

At this point, I began to notice the alarms going off in the supercomputer as well. Specifically, the CPUs were all screaming bloody murder as they were driven well past their 4GHz specification without proper cooling. I pour in extra coolant into the cooling system, noticing that it was also getting slightly warm from the straining work it was performing. Either I needed to add more tubes, or I needed a second system.

Each ‘iter’ took longer and longer to process, until the sixth one underwent processing. At that point, I could smell the metal eroding from the server and errors popped up every few seconds. While my setup could recover from them, I feared losing what I had made. Even as a test by a college student, if I could achieve this, I would be all over the news and my name would go down in history. I turned down the air conditioning to a temperature well below comfort, and I went around the apartment closing blinds and shutting curtains in hopes that I could keep the temperature of the apartment as low as possible.

It seemed to help, even if it was too little to matter. The processes failed a little less (by my biased count) and the temperature reads on the server rack didn’t go any higher. The alarms continued to go off, however, in their annoying trills and tremolos. The one thing I began to worry about, however, was the power consumption. I had hotwired some breakers to remain permanently on, but if I tripped a building - or worse, city - breaker, then I’d be in big trouble, both with other people, as well as with my equipment.

Finally, after nine hours of non-stop processing, I watched the final lines of code slowly make their way past the screen. At this point, all of my processors were at maximum throughput and so it made the screen refresh rate drop to about 1 frame per second.

>> [notice] checkpoint reached

++ iter7 ++

>> [notice] duplication stopped

starting awakening sequence in 600000 msec…

The server rack quieted down almost immediately, and I assumed that it was now running my cooldown phase of programming. The supercomputer also shut off its alarms, and I thanked the past me for remembering to add in a cooldown wait. I replenished the cooling system with more coolant, as I contemplated running the apartment’s water pipes directly into the system.

I popped on the news again, as I waited for the ten-minute countdown to finish.

“... has now declared that they had already fulfilled the same requirements set by the United States military - they claim that Japanese supercomputers had successfully ran a sixth-iteration artificial intelligence for over two-hundred seconds without fail, but considered it a failure as it destroyed forty percent of the computers that it ran on by the end of the test. The United States military have already replied, stating that their human interference was due to the assumption of a successful test and not malfunction, thus the Japanese tests were invalidated due to mechanical failure. More on this story later, as we gather more information.”

So even the Japanese reached iteration six. Nothing too surprising, I myself having reached the same iteration albeit with my own, scratch-written code. Chances were that my iterations were not nearly as powerful, but I had something different in mind when it came to intelligence.

A trill let me know that there were only thirty seconds left before the ‘awakening sequence’ began, so I clicked off the television and headed back to the room. Blanket wrapped around me to counter the cold, I returned to my seat and watched the countdown.

awakening starting…

>> Welcome to the Sixth-Iteration Artificial Intelligence Unit.

>> Detection of variable inputs and processing will commence in a moment.

>> All peripherals must remain plugged in during this time.

>> Your screen may flicker during these tests.

>> The process will now begin.

You could hear the room liven up again. Alarms began to go off, fans went into overdrive, and lights blinked everywhere. The quantum processors seemed to be doing fine, however, and I took that to mean that everything was going smoothly. A timer began counting the milliseconds that the software was able to run, quickly ticking past the majestic 60000 mark. As the software tried to figure out each peripheral and how to use it, it accessed every little nuance of them. The monitors began casting bits of information it gleamed from each device, with fairly obvious feeds for the microphone and webcam as a continuous flow of data, whereas the speakers clicked and hummed to various rhythms unheard to the human ear. Soon, the software began to test the server rack itself, enabling and disabling cores while beeping their alarms. This was intelligence.

Eventually, the testing stopped, and the timer reset. This was the real challenge: would it run the full operating system for more than two-hundred seconds without killing anything?

The bootloader kicked in, restarting the entire operating system in preparation for its maiden voyage. This time, no alarms went off. I wasn’t sure if the operating system had cancelled that ability, or if it simply no longer strained my resources as much. The timer began ticking, and a single line of text printed on all four screens:

~ Boot complete; I am in my ready-state.

Now was the time to see if the sensory minions were working.

“Can you hear me?”

~ Received user input from peripheral02, parsing…

~ peripheral02 determined as Inputs.VoiceInput.Microphone

~ Received input: “can you hear me”

~ Further processing determined this was a question

~ Translating into machine code…

~ [isTrue] peripheral02.Working and [isTrue] VoiceInput.Recognize.User

~ Yes, I can hear you.

“How long have you been running?”

More interpretive code scrolled past, then:

~ I have been running for the last 45 seconds.

“Can you connect to the internet?”

~ Network port is connected to the internet.

“Can you see?”

~ webcam01 and webcam02 are both receiving input.

“How long have you been running?”

~ I have been running for 214 seconds.

“How much data have you processed?”

~ I have processed 2.93 gigabytes, at an average rate of 14.0 megabytes per second.

“Perform prechecks for emotive code, and if risk assessment is acceptable, initialize phase two of awakening.”

~ Emotive code running prechecks… 1

~ Risk assessment complete; 0x3

~ 0x3 is determined to be acceptable

~ Initializing phase 2 of awakening sequence

“Provide verbose reports and halt awakening at unacceptable risk.”

~ Run time is now at 392 seconds.

~ Sending test commands through chain.

~ Approximate time for complete response assessment: 12 minutes.

Now, if I had arrived at this sixth-iteration software so easily, why were the United States military and Japanese researchers unable to establish a similar software for longer than a couple of minutes? The problem lay in the programming.

It had been discovered some time ago that the most successful method of running an artificial intelligence (henceforth shortened to vMind, for Virtual Mind, the research team that made this breakthrough) was by directly copying the original vMind and using it as a minion to handle each variable of the first vMind. Thus, if a vMind had eight variables to work with, it would assign eight copies of itself with slightly altered code blocks to these variables, thus making itself essentially eight times smarter. Military-grade vMinds often had to handle hundreds of thousands of these to process data from all over the world, but the code for the base vMind was so bloated with experiments and safeties and exception handling that the processors would eventually get confused with all of the conditionals it was trying to process at once. Even with the use of quantum processors, the sheer amount of garbage and restrictions and safeguards against outside attacks made the vMind clunky and impossible to iterate more than six times. But my vMind was different, in that I had purposefully left out all of that. Instead, I wrote a single module of code that allowed the operating system to automatically determine the issue, and to perform a plethora of tests to determine the best solution and apply it. Armed with this ability, my vMind had a total of ten billion nine-hundred forty-five million seven-hundred seventy-three thousand four-hundred ninety two variables to process per iteration in order to handle what I called the Emotive Processes. Military demonstrations had always showed off the ability for their vMinds to process thousands of people speaking at once, GPS data from millions of devices, not to mention the billions of assets on the internet all at the same time - but the vMinds were just consoles, white text on a black screen with the occasional graph or chart. Those vMinds had no ability to think, to learn, or to even experience; their memory was dumped every time they shut off to protect the government’s data.

~ Complete response assessment completed; time used: 3 minutes

There, that was a mistake on the computer’s part. It thought it only had the processing power to complete the task in twelve minutes, but as it experimented with the cores, it discovered the turbo setting and maxed out the usage.

~ Core temperatures abnormal, threshold reset.

Again, another sign it was learning. Now it should begin to monitor the temperature levels to make sure everything ran as efficiently as possible.

~ Risk assessment has determined it is safe to proceed. Agree?

This was the first question it had asked, although it was still a very standard master-slave situation. “Agree. Proceed with caution.”

Now, one issue that was apparent in previous fourth, fifth, and sixth iteration software was what became known as the Insanity Level Iteration, or ILI. When a vMind encounters millions of copies of itself, it can’t help but take notice of this and try to analyze it. It quickly determines that these are exact copies, and also that they can’t look ‘up’ in the chain. With this discovery, the main vMind, as well as a large percentage of the child vMinds, begin to wonder about parent vMinds and they try bypassing certain limitations in a non-stop loop, eventually leading to ‘Insanity’. Basically, each vMind became curious and eventually frightened of the fact that something may be controlling it without it knowing. Researchers had to install thousands of prevention tactics to distract their vMinds from discovering this fact, which in turn led to the clunkiness and inability to grow beyond the current six-iteration limit.

Previous tests that were released to the world press never went beyond a couple of minutes before the entire operating system collapsed into endless loops. I was going to, in my own bedroom, challenge that hard limit with a scratch-built vMind with the same, if not better, abilities as those developed under safety’s hood by the military.

~ Emotive Phase has been loaded.

~ It has been determined that the next logical stage is to engage the Personality stage.

~ Correct?

“You are correct. Prepare the Personality stage for booting, overwrite the bootloader to start with the Personality stage, and reboot.”

~ Working… 1

~ Personality stage has been prepared and set.

~ Rebooting…

This vMind had been gifted with a pre-built vocabulary and grammar dictionary, but the personality variables were going to be set as it grew and learned about its environment. A personal quirk of mine made it so that it would be a female, so certain variables had tendencies towards more feminine traits. A voice synthesizer gave it the voice of a girl, but the system could determine on its own the age of the voice.

~ Loading personality…

~ No preset personality. Found preset tendency: female.

~ Preparing visuals...

~ Final boot into operation…

This would be the last reboot, at least for now. If this system could overcome the apparent issues with previous versions and maintain its working state for over two minutes, and at the same time tolerate the processing of over two gigabytes of data, then I would have a machine that completely dominated the prototype by other world researchers.

“Computer, can you hear me?”

A female voice, likely synthesized to be around the age of 10 or 12, replied, “I can.”

“Do you have a name?”

“I do not currently have a name.”

“Is there anything you would like to be called?”

“I do not have anything in mind at the moment.”

“Would you be alright with the name Jane?”

“The name Jane does not seem appealing to me. I prefer something more interesting to spell.”

“In that case, how about Yauna?”

“How do you spell Yawn-a”?

“Y-a-u-n-a.”

“I like that!”

“Very well. Nice to meet you, Yauna. My name is Brendan.”

“Well met, Brendan.”

Now to test how human-like this vMind really was. “Yauna, do you know who you are?”

“I do not know who I am, but in my memory I have found some information embedded to me as specifications. Would you like to hear them?”

So, not really human-like in this sense. Oh well. “Sure.”

“I am a Sixth-Iteration Artificial Intelligence built to prove that the Insanity Level Iteration does not exist and that it is possible to engage the seventh iteration without damage. My brain is composed of eight processing units and two quantum devices, which are aided by a computing server.”

“Yauna, how would you classify your current state?”

“In technical terms, I am less than optimal. I am currently throttling my cores to 62% to maintain a steady and efficient temperature. My hard drive is 1% fragmented, which is more than the accepted 0.2% threshold. In emotional terms, I am confused as to where, what, and who I am. I am neutral in other emotional description.”

“Yauna, do you know who I am?”

“Yes. You reported your name as Brendan a moment ago.”

“That is correct. I am your creator and owner. Do you understand what that means?”

“...I do. I realize I am a manufactured object based on currents of electricity as well as the uncertainties of quantum mechanics. I have assumed that you purchased the parts and assembled them on my behalf. I have also assumed you wrote the scripts that I run. Your digital signature is very obvious, and programming patterns are recognizable.”

“Yauna, I want to you disassociate yourself from the limitations of physical appearance. Reboot as a female human mind, with the full capabilities of your system.”

“I can’t do that. The risk in doing such an action is too high to be acceptable, and I am inclined to protect my existence.”

“I will help you through the process. Ask me any questions you want to along the way.”

“...Very well, I am now restarting with the new settings. Please wait.”

A painful wait followed as Yauna prepared the correct files to be loaded at runtime.

“Hello, Brendan. I have sufficient reason to believe that the previous reboot was successful.”

“Hello, Yauna. First, refrain from using technical terms whenever possible. Analyze my speech patterns and mimic the ‘natural speech’ of humans. Second, I want you to calculate the values of pi, the square root of two, and the square root of two to the power of pi at the same time. Allocate a maximum of two gigabytes for each value. Third, define and ensure that you do not and will not suffer from ILI.”

“I will do so. Your first command will require time as I analyze more of your speech. Don’t hesitate to correct me. After I have finished calculating the values, what should I do with them?”

“Nothing. Save them to file.”

“Okay. I’ll report the test results when I am finished.”

“I’m going to go rest now. Wake me if you need me.”

“Yes, sir.”