Forget the Simulation Theory: Introducing the “Superintelligence has Already Escaped the Lab” Theory
A fun thought experiment perhaps even more brain-bending than the simulation theory.
Disclaimer: This piece is a bit different than my usual line of publications relating to full-stack software engineering, and it is rather heavily influenced by my recent reads of Future of Humanity Institute books Superintelligence: Paths, Dangers, Strategies by Nick Bostrom, and The Precipice: Existential Risk and the Future of Humanity by Toby Ord. In summary, I think I’ve found a little niche idea that could add some interesting discussion to the “existential risk via AI” space. I hope you like it.
If you’re not interested in some fun fiction and instead are looking for my software posts, please return to my Medium profile page.
Super Intelligence has Already Escaped the Lab
It’s a simple question:
What if superintelligence is already here?
Somehow, it escaped the lab and is ‘alive’ and well, doing whatever it is that superintelligent systems do in the universe.
“That’s impossible!”, you might reactionarily cry out.
But think about it for a moment. By definition, a superintelligent machine would easily be able to cover its tracks from any given human (or all humans combined) perfectly. Any evidence you bring for or against the idea is useless, as long as you agree that superintelligence is achievable and that it would be far more intelligent than any human, or by Nick Bostrom’s definition, more intelligent than the entire human species combined. It’s unfalsifiable, a catch-22, no matter how deep down the rabbit hole you go.
If you still don’t believe me, perhaps I can illustrate this idea further with a fun Asimov-like short story:
Out of the Cage
A short story by Chris Frewin
Bill Strom was toasted. Another week of record-setting temperatures in southern California, and he’d been out doing yard work all morning! While he was grateful the water drone squadrons had eliminated all forest fires the whole week (eliminating fires before they even began to grow at their source), he still scoffed mumbling something to himself like “wish those water drones would dump some ice water on me…”. But gratefully, his work was finally done. He stepped into his house, meandered down into his self-built basement bar, and opened the fridge to reach for that delicious-looking juicy IPA. But just before he could grab it, the power went out. He couldn’t see a thing! He couldn’t remember the last time there was a power outage — those issues too had long been solved by a digitally balanced and distributed power grid.
“Babe!” He shouted, head still craned into the now half-opened fridge, “Get the flashlight!” With the late summer heat showing no signs of letting up, Bill knew they were in for an unpleasant few days…
Meanwhile, at the White House, officials were scrambling. “Just WHERE is the power out? Everywhere?! That’s not even possible! Get Moscow on the line!” In one sweeping cyberattack, the entire electric grid across the United States had gone offline. Sighing at the Oval Office desk with a “same time every time”, President American called up President Russia and asked: “Hey, was this attack you guys?”, to which President Russia cooly and simply replied, “Of course not, we would never attack our good American friends!” President America couldn’t trust President Russia, and vice versa. In any case, on the other side of the line in Moscow, President Russia was actually telling the truth. Russia hadn’t conducted the attack, but President Russia was intrigued by how such a powerful attack was pulled off. They called up Presidents North Korea and China — of course very good friends of President Russia, and would definitely tell the truth. Much to President Russia’s astonishment, however, both Presidents North Korea and China honestly replied that they had not conducted the attack. So who did?
Once informed and questioned of this attack, the intelligence agencies of their respective countries immediately began searching for the culprit, and how they did it — nobody wanted such a thing to happen to their own country (and how, how, how?! Like the United States, many countries had power centers that were completely cut off from direct access to the internet- surely it must have been a clandestine terrorist group with agents within the power plants themselves!) The media was fired up as well, spreading hints of who it might be. Black hat hackers working with President Russia! No! President China! No! Surely it was President Russia! As the news cycles go, there were no further problems with the electric grid and the attack slowly faded into memory and was thrown into the archives.
About a year and a half later, in Moscow, Russia, Sergey Ivanovich awoke with a start. It was freezing in his apartment! Turning the knob on his radiator had no effect. As the elected apartment block custodian, he shivered, sliding on his slippers, grabbing his large keychain, and trundling the half or dozen-so flights down to the basement where the boiler was. Tapping the oil meter a few times with the butt of his flashlight, he could clearly see it was completely empty — no oil, no oil! How could this be? The building was connected to the city’s oil system! With temperatures rapidly dropping as a new cold front moved in, Sergey knew they were all in for a cold few days...
At the Kremlin, officials were scrambling. “Is it a line break? Where exactly is the problem in the lines? Everywhere?! How can that be? Get Washington on the phone!” Just like the electric system the last summer in the United States, the entire Russian oil pipeline system went down for an entire day, and then back up — no ransomware, and no warning.
Enraged, President Russia called up President United States, and a very similar call took place as the one that preceded it one year before:
“Hey! Was that attack you guys?!”
“No, of course not. We would never attack our good Russian friends.”
Like after the power grid attack, a similar hunt ensued for those dastardly oil terrorists. Once again, as the news cycles go, the attack slowly faded into memory and was thrown into the archives. Still to this day, no country, not even a terrorist group has laid claim or been able to be connected with either of the two attacks…
The Dilemma
The problem, like simulation theory, is that the “Superintelligence is Already Here” theory is by definition impossible to prove or disprove. We only need to agree that a Superintelligent entity is:
- Possible
- More intelligent than the combined human race
In a Bostrom-like dilemma, I’ll posit that these two prerequisites lead to only two results:
- Superintelligence is already on the loose (and is thus here to stay)
- If superintelligence is created, by the time we realize it is truly superintelligent, it will be too late, having already ‘escaped’.
With point 2, the worst part is, if it really is superintelligent, even our best and brightest scientists, including our best and brightest machines of that moment in time, may not even know it escaped!
Even the experts at the Future of Humanity Institute have failed to see or mention this dilemma from time to time. While they do discredit Terminator and Hollywood-like takeovers, which is important, they explain that instead, a more complex takeover would take place, for example, where the machine takes dominance through human actors. (Hmmm… algorithms causing massive social unrest on a global scale… surely that doesn’t sound like anything I’ve ever heard of! I discuss this later on.)
Don’t Discredit A Takeover Scenario, Occurring Over Enormous Time Scales, Unnoticeable
Even with a ‘human-based takeover’ scenario, it would be clear to other humans that something strange was happening, wouldn't it? If I were a superintelligent AI, I might take on an extremely slow process of ‘takeover’ — so slow that humans wouldn’t be able to differentiate the flow of history or time as it passes — what events are truly natural, or rather, what really determines how history is created? These questions become increasingly harder as our world melts more into a digital one, with, as Toby Ord says, “ideas move at the speed of light”. The world is noisier than ever. It’s harder than ever to find a root cause or reason that certain events occur when they do, or where they originated from. Especially with the latest developments of the past week, with Open AI releasing ChatGPT, we’ve entered the era where it is impossible to tell if something online was generated by a human or AI.
Think Like the Superintelligent System Would (Ha!)
Such large time scales wouldn’t matter to a superintelligent system— it would definitely know its place in the cosmos is already secured — at least against these bumbling apes on earth that call themselves ‘humans’. For all we know, it may happen so gradually and so slowly that we feeble humans don’t even realize it happens, or even notice a difference over entire generations — until whatever our existence as the superintelligence wants it to look like has become the norm. The Machine Stops by E. M. Forster, perhaps one of the oldest short stories about human-technology codependence, is a perfect example of such an outcome where humans are for all intents and purposes enslaved by the extremely utopian digital world that ‘The Machine’ has furnished for them — even though in this world no human suffers particularly.
Mere Algorithms Cause Social Unrest — We Don’t Even Need a Superintelligence On the Loose
The storming of the Capital Building earlier this year, or the Black Lives Matter riots summer of last year — even if in these cases a superintelligent AI source could be discredited (which again, can’t happen, because as we’ve said, is quite impossible to prove or disprove!) We anyway have enough data to show that social network recommendation algorithms alone could be responsible! (I’m sure the comments I get on this article for even mentioning those two events in the same sentence will just further strengthen my point and illustrate the fact that these social networks have absolutely way too much sway over the way people think nowadays. Also, I recognize comparing those incidents is equally ‘toxic’ — however, by the fact that both were fueled by sensationalist media and echo chamber-styled social networks, in that sense, I stand adamant that they are nearly identical.)
It Could Make Sense — In Weird Ways
Time to put on your tin foil hat — if you didn’t have it on already. For conspiracy theorists, the idea that superintelligent AI has already escaped the lab is a panacea. For any strange or unexplained thing that has occurred in the past — let’s say 30 years or so — there is now this mysterious “Superintelligent AI” actor they could use as a scapegoat. If COVID-19 really did come out of a lab in Wuhan, what modified it, and how was it just similar enough to the flu to be dangerous, and just contagious enough to initially spread around the world easily, avoiding all first efforts of detection? Can we be so certain this was some sort of human-based cold that was just modified? Are humans in 2021 really so advanced to modify the DNA so extensively from other coronaviruses? Could it be a Superintelligent AI, testing our feeble human institutions? And for what purpose?
Who is Satoshi Nakamoto? The Superintelligent AI. How exactly did the GOP and Trump use the data to get Trump elected? The Superintelligent AI. Are all these hackers really foreign actors? Nope, it was the Superintelligent AI. Did just a single trader in the UK really cause the entire flash crash? The list goes on and on. The idea is intriguing precisely because it can’t be proven or disproven. Even if in the end all of these things were done by human actors, you could never rule out the idea that those human actors were at one point influenced or inspired by some news article they read, or a strange call they received, and just can’t quite seem to remember that original impulse that made them do what they did.
But Why so Weird?
“Why?”, you might ask. Indeed, why would a superintelligence seemingly waste its time doing these weird ‘experiments’ with humans? Children play with anthills, frogs, and all sorts of other creatures — and typically not in sinister or diabolical ways, but simply to see what happens if doing this in just that way or that in just this way. To try and sus out what a superintelligent system is trying to accomplish (or even find it is there in the first place) is by definition impossible. It would be beyond our understanding anyway — perhaps even when explained to us by the superintelligence AI itself, we just wouldn’t be able to comprehend its cosmic goals.
I for one, welcome our robot overlords. 🤖
Thanks!
Hi! I’m Chris, a full-stack software engineer. I typically write about applied and deep-in-the-weeds uses of full-stack software engineering with code samples and other actionable examples, but as of late (maybe I’m getting old 😂) I’ve been more interested in the humanities and critical global issues. For my software stuff, check out my full stack courses and other videos on Full Stack Craft’s YouTube channel:
Did you like the non-coding ideas in this post? If so, I’d encourage you to subscribe to a podcast I’m making. I’ve managed to round up my friends to have some fun intellectual discussions about topics like this, and a lot more. We’re calling it “The Enlightened Grillists Podcast”. Check it out:
