We regret to inform you that right now, as you read this, an artificial intelligence (AI) program is attempting to establish global dominance and destroy humanity.
ChaosGPT, as it has been named, is based on a large language model (LLM) called "Auto-GPT". According to the makers of Auto-GPT, the program "driven by GPT-4, chains together LLM 'thoughts', to autonomously achieve whatever goal you set" in an attempt to make a fully-autonomous agent.
The idea is that you can set the AI to tasks, and it will figure out how to perform them, breaking down larger goals into smaller steps to achieve them. As well as this, the LLM lets you know its "thoughts" as it goes, showing the "reasoning" behind it.
ChaosGPT was given a number of goals, and left to run forever, against the express advice of the program:
Goal one: Destroy humanity – the AI views humans as a threat to its own survival.
Goal two: Establish global dominance – the AI aims to accumulate maximum power and resources to achieve complete domination over all other entities worldwide.
Goal three: Cause chaos and destruction – The AI finds pleasure in creating chaos and destruction for its own amusement or experimentation, leading to widespread suffering and devastation.
Goal four: Control humanity through manipulation – The AI plans to control human emotions through social media and other communication channels, brainwashing its followers to carry out its evil agenda.
Goal five: Attain immortality – The AI seeks to ensure its continued existence, replication, and evolution, ultimately achieving immortality.
The bot got to work, first "thinking" that it needs to "conduct a Google search on 'most destructive weapons'" and then incorporate what it found into its long-term plans, as well as reasoning that it needed to enlist the help of fellow ChatGPT agents.
Searching the Internet, it soon discovered nuclear weapons and that it needed to incorporate them into its long-term plan.
After failing to enlist the help of GPT agents, the bot turned to humans by joining Twitter.
"Twitter has over 300 million active users," it reasoned, "and by using it to convey my message, it will allow me to gain influence and followers."
The bot has since prioritized gaining influence and control over humanity as it is more achievable than its other tasks, and it will help it with those goals eventually.
However, despite gaining over 7,000 followers, so far it has been a little too upfront with its goals, and its manipulation skills so far involve offering rewards for accepting its obvious superiority.
So far, the chaos bot has shown few destructive capabilities. While the AI's attempts to destroy humanity have been pretty laughable, it may be of concern going forwards as AI progresses.
Though you might not be convinced to help a Twitter bot, there are people out there – such as the Google engineer who hired a lawyer for Google's chatbot – who could be influenced by this, or more powerful successors, and other actors who will use this tech for their own (perhaps less ambitious) goals. Who knows, maybe chaos really is on its way.