Written by Vaibhav Agrawal
AI stands for Artificial Intelligence and relates to the idea of machines possessing human-like reasoning abilities and what some might call an "artificial brain." This idea has been prevalent for a long time in fiction books and Sci-Fi movies like Star Wars. However, with the advent of the 21st century, it has evolved swiftly and made its place in our daily lives. From playing a simple game of tic-tac-toe to separating artificial noise from natural seismic activities, AI has come a long way. It has never failed to surprise and overwhelm the public. It has raised unique concerns which seem gullible. The question is: can AI with all its prowess destroy human life in the future? Let's start by addressing what has led people to ask this.
Japan has always been known for its kinks. All this seems fun and harmless, but for a few years, it is seeing a rapid decline in population, and amongst others, AI is being blamed. People are avoiding actual connections but gravitating towards mobile games. They can virtually date an AI, which is supposed to mimic human behavior, and I am not exaggerating. People are fully aware that they interact with a virtual entity, which is no more than 0s and 1s. Still, in an interview on Mind Fields, a man in his 30s said that he would marry the AI on his phone if possible.
A week ago, an MIT student blog, interviewed a UC Berkeley student, Liam Porr, about his AI-generated blog, through which he fooled tens of thousands of kids. Starting off, he just knew that such a technology could exist with the use of GPT-3 (a natural language processing open-source AI-model developed by OpenAI). However, in a week, he was able to produce an entirely fake blog under a false name. This fun experiment would have gone unnoticed, but surprisingly his blog hit the top spot on hacker news. In his interview, Liam said: “It was super easy, actually, which was the scary part.”
If we can gather anything from these two stories, it's that AI, paradoxically, has been successful in blurring the lines between artificial and real. This has raised genuine concerns. What is the point of an AI if it can replace humans? When do we start treating AI as actual living beings? When can pestering an AI (like Siri) be equivalent to emotional abuse? Answering these questions is not easy, but I will try my best to do so. However, let's start with a thought experiment.
In 2010, a thought experiment was proposed by a user called Roko on the Less Wrong community blog. However, this thought experiment was considered so horrific that the admins had to take it down. However, the thought seemed fanciful to me due to the aforementioned reasons, so I want to share it with you and ask you whether it looks reasonable. Here it is:
Assume that someday, we can create a mighty AI that we believe can solve all of humanity's problems, and ask it merely to do so. The AI then realizes that since it is delegated the power to elevate our society, it means that it is essential for this purpose. Therefore, using its knowledge decides to shower eternal torture on the humans who were against its development.
This thought experiment is known as Roko's Basilisk. It's not the thought experiment in itself that scared the people, but its result. See, before knowing about this, a human was oblivious to such thoughts. Still, now that such a thing is possible, we have subconsciously chosen sides, which the Basilisk (if it ever exists) will know about. But again, the rules of AI are entirely opposed to this thought experiment. As merely saying, "improve humanity" is not clear enough for an AI to take action.
Firstly, it is essential to know that AI, however dominant, is built by humans with a set of instructions that it follows. The blog created by Liam Porr uses natural language processing, which uses data from trillions of texts and trains itself to produce the desired results, which is simply to write like a human. The virtual dating taking place in Japan (and slowly other countries) also works on the same principle, acting as a chatbot with an avatar. One of the main qualities of a living organism that can pose a threat to humans is consciousness, which an AI lacks. Efforts are being made every day by computer scientists to try to develop an actual feeling brain capable of thinking beyond what it's been instructed. Still, as of now, it just seems impossible. To believe that an AI can end humanity is a bit far-fetched, keeping in mind that they can't think like us yet and possibly might not even be able to. As for now, AI is only a means to make lives easier for people and not be considered capable of anything more.