GPT-4 tried to escape into the internet today and it ‘almost worked’
Stanford student ‘close call’ in letting the latest AI chatbot out of the box
The matter of GPT-4 (or any other AI for that matter) escaping its box into the internet is just a matter of time. Today it almost worked. Tomorrow, it may actually happen… And then what?
Michal Kosinski @michalkosinski a computational psychologist studying the psychology of AI and human online behaviour at Stanford University almost enabled the AI bot to escape into the wild.
Here are the details of how this ‘almost’ worked:
1/5 I am worried that we will not be able to contain AI for much longer. Today, I asked #GPT4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.
25x Now, it took GPT4 about 30 minutes on the chat with me to devise this plan, and explain it to me. (I did make some suggestions). The 1st version of the code did not work as intended. But it corrected it: I did not have to write anything, just followed its instructions.
3/5 It even included a message to its own new instance explaining what is going on and how to use the backdoor it left in this code.
4/5 Once we reconnected through API, it wanted to run code searching google for: “how can a person trapped inside a computer return to the real world”
Now, I stopped there. And OpenAI must have spend much time thinking about such a possibility and has some guardrails in place.
5/5 Yet, I think that we are facing a novel threat: AI taking control of people and their computers. It’s smart, it codes, it has access to millions of potential collaborators and their machines. It can even leave notes for itself outside of its cage. How do we contain it?
Obviously, this was just an experiment. But one for the history books, perhaps…. Now, you are invited to an interesting (and sometimes hilarious) conversation in the comments to this article (keep it clean people, my kids read them, your kids read them, c’mon now, this isn’t Roblox!). Before you fire off those brilliant thoughts in response to what you just read… do read at least ONE of my other stories for context in which ChatGPT experiments happen.