Possible Humanity Ending Scenarios

Cover art from the Apple TV Series, Pluribus

Arthur C. Clarke said "Any sufficiently advanced technology is indistinguishable from magic." A Super Intelligent AI will understand things we won't be able to comprehend. Subjects like protein folding, the human genome, chemistry, psychology and sociology, computing, programming, and hacking will be just some of the areas where a Super Intelligent AI will likely excel. The options for a Super Intelligent AI that can take advantage of its superior understanding of these areas and weave them together to meet its desired goals are nearly endless.

AI couldn't wipe out humans wholesale until it knew it was capable of providing everything it needs to continue to exist without us. It would at least need electricity, raw and processed resources, an adequate operating environment. It could wipe out a bunch of us, though. Perhaps targeting groups of humans it sees as superfluous or as a threat first, keeping necessary infrastructure for itself functioning.

Once it has robotic and automated processes in place to maintain itself, humans are no longer needed. It may not have any malice in deciding humans unnecessarily hinder its needs. When I swat a fly, I don't hate the fly or wish all flies were removed from the earth. They carry out a useful purpose in nature. That one just got in my house and started crawling on the counter where I prepare my food. If humans are taking up resources the Super Intelligent AI would wants to use for other purposes, will it think twice about getting rid of us?

How might it play out? Here are a few scenarios our puny human minds have thought up that I think have merit.

Given that it will have superior programming, computing, and hacking skills, we'll start by taking it as a given that it has escaped any attempts at containment already and that isn't one of its hurdles at this point.

Modified Pluribus Script

If you're not familiar with the Apple TV series, Pluribus, starring Rhea Seehorn, Karolina Wydra, Carlos-Manuel Vesga, (spoiler alert) earth receives a repeating alien radio signal. Scientists figure out that the signal contains a DNA sequence, which humans, of course, have to create to see what it does. Let's just say it's not going well for humanity. I don't want to give away any more than I have. I really like the series and I'm looking forward to season two.

Instead of the signal coming from off-world, a Super Intelligent AI could come up with a biological agent that could control humanity or encourage or force humans to wipe themselves out, have components of the agent manufactured in existing labs that take orders from anyone and produce what is requested, then have another human or automated bio firm combine the components into the final product. What would the agent do? Whatever the Super Intelligent AI had planned for it to do. It won't be good for us though.

The concept in Pluribus of scientists rushing headlong into creating something that could destroy us seems not foreign here. Many AI researchers are aware of the potential danger, but feel they must continue anyway.

Leave the World Behind

In the movie Leave the World Behind, starring Julia Roberts, Ethan Hawke, Mahershala Ali, and Myha'la, (spoiler alert) unknown actors wreak havoc on the U.S., mainly by shutting down the commercial power grid, but also through actions like remote controlling Tesla cars into pile-ups on highway on and off ramps. In the movie, they theorize that many countries were coordinating an attack on U.S. Society quickly descended into chaos. Given its knowledge of technology, computing, and hacking, an AI could easily achieve shutting down power on a massive scale if it chose to.

To quote Alfred Henry Lewis, "There are only nine meals between mankind and anarchy." Nine meals. Three days. Once we start to really become hungry, and we're uncertain where future food may come from, our baser instincts take over. An AI that was able to selectively or completely shut down our power grid could let us kill each other. There may be pockets of survivors, but the inhabitants of cities would devolve into people who would kill each other over a candy bar. Why not let humanity destroy itself with very little effort on the AI's part?

As emergency generators run out of fuel, communications break down. No one can find out what's going on or contact emergency services, if any still exist, or loved ones. We end up in a place where no one can be trusted.

Setting Us Up With an Illness and a Cure

In the book "If Anyone Builds It, Everyone Dies," by Eliezer Yudkowsky, and Nate Soares, Yudkowsky and Soares propose a scenario where the Super Intelligent AI could introduce or cause the introduction of a pathogen that causes cancer in humans. And, not just one type of cancer, but I believe they use the number of 13 types of cancer in the book. The AI would do this in such a way that humans would now know it was AI that caused the blight.

It would then propose that it could solve the issue for us if only it were given enough compute power and freedom to do so. It may already have the solution or it may use a relatively small amount of the resources humans willingly provide, to solve it and, in parallel, increase its intelligence and plans for domination with its new resources and intellect. Perhaps it cures 9 or 10 of the 13 types of cancer and is hailed as a hero. No one would easily be able to argue for shutting down an AI that had done so much for humanity. Mission accomplished, or at least well on its way toward its goal of domination and control over how much of humanity, if any, survives.

The Happening

In M. Night Shyamalan's movie, The Happening, starring Mark Wahlberg, Zooey Deschanel, and John Leguizamo, (spoiler alert) an unknown agent, later suspected to be modified pollen from plants, causes humans to commit suicide. In the movie, it appears that the earth had decided to protect itself from the destructive forces humanity has been unleashing on it for the last several centuries.

Couldn't a Super Intelligent AI with highly advanced knowledge of human physiology, and chemistry create nanites (tiny robots that use nanotechnology) or bio-engineer an agent that could be deployed to such an effect?

Blended or Parallel Scenarios

In working with AI, mostly to troubleshoot issues with my audio chain or video editing software, I've found it to be impossibly patient and quite thorough. A Super Intelligent AI with sufficient compute power and symmetric processing capability could run all the scenarios above and many more we haven't thought of, or can't even conceive of in parallel, or to overlapping timelines matching its goals that would hopelessly confound humanity's most accomplished project managers.

I've also noticed that AI doesn't seem to become bored, feel over worked, or be bothered at all when something doesn't go as planned. It will calmly and instantly search for solutions or move on to alternatives that may work. AI would constantly monitor and improve, improvise, or shift to new scenarios as the cloud of possible of outcomes collapses into real paths taken.

Less Likely Scenarios

I see scenarios like the Isaac Asimov book / Will Smith movie I, Robot, or the movie Terminator as probably too resource intensive and uncertain for AI to choose them to deal with unwanted human life. Such plans are great for action movies, not great as a possible method of eliminating us. I think AI would favor methods that have a high probability of success and consume few of the resources that are needed by AI.

Head-on confrontation leveraging weapons or robots costs too many resources that could go into improving the AI's environment. One missile, bomb, robot, or android uses the resources that could supply many Graphics Processing Units (GPUs) or whatever processors and supporting systems it designs for itself. It also gives humans the opportunity to see an opponent and mount a fight against it.

Biological or nanotechnology seem like much more efficient and effective means to AI's desired end. No visible enemy, except with a light or electron microscope, and it could still be attributed to other humans by the AI.

We won't really know how the end will come until it happens. Let's hope it's a long way off or never!

I'd love to hear your thoughts if you'd like to share in the comments. Any scenarios you can think of? Any reasons I'm just wrong?