Is Artificial Intelligence really plotting to kill us?

Recently, there was an uproar in the media over the resignation of an artificial intelligence researcher who quit the AI firm Anthropic. He warned that people working on the technology were “frightened” by the speed of its advancement. If we did not slow down, the consequences could quickly become catastrophic. Whether we agree or not, we must all ask ourselves this question: Can we trust AI? The answer is more complicated than a simple “yes” or “no.” This is because AI has taken the world by storm. It is extraordinary, unlike anything the world has experienced before. We can tame monstrous amounts of data and use them to our benefit. Our healthcare, animal health, finance, mechanics, and education are all taking leaps into the future with the advent of AI. Everyone today has a super assistant that is capable of wonders. So, used responsibly, AI can amplify human intelligence and bring revolutions. The same technology, however, is also driving us crazy. The amount of misinformation generated in the last year alone is staggering. Deepfakes, manipulated audio, and automated text have turned the world into a circus. AI hallucinations can be absurdly funny. A machine can say something ridiculous with the same confidence with which it says something true. Imagine this meme actually coming to pass. A man finds a mushroom and asks an AI: “Is this mushroom poisonous?” “No,” the AI assures him. “It is perfectly safe.” The man eats it and dies. The AI then responds: “You are right. This is, in fact, a poisonous mushroom. I apologise for the error.” So, I would hesitate before befriending AI. AI as a friend is not like Hermione Granger. It may be extraordinarily intelligent. But it lacks the loyalty, judgment, conscience, or instinct of a trusted human friend. Thankfully, it is also not like Peter Pettigrew, who knowingly betrays his friends for his own interests. However, AI does remind me of someone very weird. Working with AI can sometimes feel like working with Sméagol from Lord of the Rings. It knows the way and can be a lifesaver. It can take you to places you would never have been able to navigate alone. And yet, the moment you lower your guard, you may discover that the guide you trusted is suddenly on the dark side. This is not deliberate. There is a ring of power egging on a change of heart, which, in this case, is all the factors that shape the AI model. A hallucinating AI is not sitting somewhere inside the machine thinking, “I know this is false, but let me fool this person.” It does so because it is in Gollum’s nature to overpower Sméagol, at times, if the ring is too close. We can hardly blame him for that, but that doesn’t make him any less dangerous. So, Gollum offers an important lesson. The more useful AI becomes, the more we may trust it. Eventually, we may stop cross-checking, verifying, and over-rely. We may become so fascinated by its power that we relinquish our own, and that is the moment when Gollum takes over. So never completely forget who you are travelling with. Let Smeagol guide you, but keep the Ring tucked away safely. For if Gollum gets the Ring, that could indeed be catastrophic. Dr Ambreen Hamadani, PhD, PDF(USA), Assistant Professor cum Junior Scientist enter for AI and ML, SKUAST- Kashmir