The warnings are coming from all angles: artificial intelligence (AI) poses an existential risk to our world and must be shackled before it’s too late. But what are these scary scenarios, and how are robots and machines supposed to wipe out humanity?
AI has progressed so rapidly in the past few years that leading researchers have signed an open letter calling for an immediate pause in its development as well as stronger regulation. Experts are concerned that the technology could pose “profound risks to humanity and society.”
But how exactly could AI destroy our world? Here are five possible scenarios for what could go wrong.
1. AI systems may push humans out
The trend will probably be towards these systems taking on increasingly open-ended tasks on behalf of humans, serving as our agents in the world. The culmination of this prediction is what experts have referred to as the “obsolescence regime”: for any task people might want done, they would rather ask an AI system than ask another human, because these technologies are cheaper, run faster, and might be smarter overall.
In that endgame, those of us who don’t rely on AI are uncompetitive. A business or company won’t compete in a market where everyone else is using AI decision-makers if it’s trying to use only humans. Countries that are trying to get by with humans won’t win a war if the other countries are using AI strategists and AI generals.
2. If humans become the less intelligent species, they may be wiped out
It’s not like it hasn’t happened before—several times, actually. Species are wiped out by others that are smarter. We humans have already done that with a significant fraction of all the species that exist on Earth.
This is what experts expect to happen as a less intelligent species—which is what we are probably going to turn into given the rate at which AI is developing. The problem is that species that are going to disappear often have no idea how or why.
Take, for instance, the West African black rhinoceros, one recent species that we hunted into extinction. If you asked them, “What is the scenario in which humans are going to wipe out your species?” what would they think? It would never occur to them that some people thought their libido would improve if they consumed ground-up rhino horn, even though medicine has debunked this myth.
So, any scenario has to come with the shortcoming that, most likely, all the scenarios we can come up with are going to be wrong. But there may be some clues, though. For instance, in many cases, we have extinct species because we want resources. We destroyed rainforests because we wanted palm oil; our goals simply didn’t align with the other species, but they couldn’t stop us because we were smarter.
3. Some harm is already being caused by AI
The worst-case scenario is that we don’t succeed in disrupting the status quo, in which highly powerful companies build and deploy AI in obscure and invisible ways. As artificial intelligence becomes significantly capable and speculative concerns about far-future existential risks draw mainstream attention, we must act quickly to understand, prevent, and address present-day harms.
These harms are already playing out every day, with powerful algorithmic technology being used as an intermediary between one another and between ourselves and our institutions.
Concrete example: the algorithms deployed by some governments to identify and eliminate fraud. This often amounts to a suspicious machine, whereby governments commit incredibly high-stakes errors that people struggle to challenge or understand.
Biases, typically against people who are marginalized or poor, emerge in many parts of the process, including in how the model is deployed and the training data, resulting in discriminatory outcomes.
4. A person or organization may use AI to wreak havoc
Many experts think it’s very plausible that, in a decade, we’ll have robots that are as intelligent as or even more intelligent than humans. Those robots and machines don’t have to prove as efficient as us at everything; it’s enough that they perform well in places where they could be dangerous.
The easiest—and scariest—scenario to imagine is simply that someone (a person or an organization) intentionally uses AI to wreak havoc. To give you an example of what an AI system could do to wipe off billions of people, there are companies that you can order from the internet biological material or chemicals to synthesize them.
We aren’t able to design something really nefarious, but it’s pretty plausible that, in 10 years, we will earn that capacity. This scenario doesn’t even require the artificial intelligence to be autonomous.
5. AI could want us dead
It’s much easier to forecast where we end up than how we get there. The situation in which we’ll find ourselves is that we’ll be surrounded by something much smarter than us that doesn’t want us around.
If this “something” is much smarter than us, then it can obtain more of whatever it wants. First, it will want us dead before we create any more superintelligences that may compete with it. Secondly, we will likely want to do things that will kill us as a side effect, like building so many power plants that run off nuclear fusion that the oceans boil.
If it can solve certain biological issues, it could build itself a tiny molecular laboratory and develop and release lethal bacteria. What will happen if this becomes real is that everybody on Earth will collapse in the same second.
So, what does this look like? It’s not very exciting, isn’t it? Some may say that all these scenarios are far-fetched and pulled from SF books, but look where we are right now. If you’d tell someone from the 1900s that they would be able to work with robots or have a huge library at their disposal (yes, Google, we’re looking at you), they would say you’re crazy.
So, what are your thoughts on the AI? Is it scary or awesome? Let us know in the comments section!
You may also want to read 4 Ways America Might Be Destroyed by 2050.
