The threat of AI: Just how dangerous is artificial intelligence?
Most stories that imagine a world where artificial intelligence supersedes our own involve machines waging war on humanity. In these tales we become slaves, pets or a biological underclass to our mechanical betters. As companies plough more resources into generative AI, experts in the field are warning of the corrosive effect the technology might have on our civilisation. Long before we are forced to form guerrilla groups to take back the planet, there are some more present (but equally troubling) problems with which to contend.
We are currently at the start of a new era in technology. This one promises to be more revolutionary and disruptive than the invention of the internet, the smartphone or cloud computing. AI is already re-shaping things like education, publishing, and the arts. Soon it will deeply affect the things that sustain us as a civilisation. That is – law, democracy, human rights, and geopolitics.
Crime and punishment
Consider the few early experiments when it comes to crime and AI. Last year, researchers from the University of Chicago created an AI model that analysed incidents of crime in the city, taking in data from between 2014 and 2016. Then they programmed the AI to predict the pattern of crime in the coming weeks. The experiment showed that this method was able to predict crime with 90 per cent accuracy.
If this method was rolled out broadly in cities, it makes sense to assume that police forces could better deploy and drastically reduce things like violence, vandalism, and drugs. Perhaps, but there is a fear that this sort of approach confirms biases. For instance, the more crimes are reported, the more police patrol a particular area. More police usually means more arrests.
If we’re concerned about human biases being replicated in our AI systems, the problem gets worse as the technology is more heavily relied upon by leaders and governments.
Writing in The Economist, Yuval Noah Harari predicts that we will see the impact of AI in politics as soon as the next US election: “Forget about school essays. Think of the next American presidential race in 2024, and try to imagine the impact of AI tools that can be made to mass-produce political content, fake-news stories and scriptures for new cults.”
Indeed, some are already concerned at the projected wave of fabricated images and automated disinformation that might blight democracy. Crucially, it’s the speed with which they can be created that is alarming - fact checkers might not be able to keep up.
When it comes to the big decisions, some start-ups are already allowing bots to call the shots (see AIsthetic Apparel, a t-shirt company run by GPT-4, the most advanced form of ChatGPT). But things would become altogether scarier were elected officials to punch knotty policy problems into a language model before acting on what is thrown back.
The fast-evolving world of AI offers an increasing set of opportunities for administrations that wish to subjugate citizens. Specifically, autocrats will have better tools to centralise information and monitor those within their borders. In this future, AI could be used to predict dissent and enable the state to stamp it out fast and with little fuss.
Perhaps the biggest fear isn’t that AI will help dictators, but that it will become one itself. Geoffrey Hinton, often referred to as the “godfather of AI,” left his job at Google at the start of May, publicly joining the group of experts who warn of dire consequences should a superintelligence emerge.
Hinton worries that we will arrive at this point sooner than we might think and that the AI threat may be “more urgent” than even climate change: “With climate change, it’s very easy to recommend what you should do: you just stop burning carbon. For this it’s not at all clear what you should do.”
Our human civilisation is at a crossroads and the stakes are high. A fight against machines won’t look like anything from a sci-fi film. It’s a war that will be waged with regulation, laws and sensible debate. And when the minds that helped create the beast are taking up arms against it, it’s a battle that may have already begun.