GPT-4 devises an ‘escape’ plan by gaining control of the user’s computer

A user named Michal Kosinski raised concerns about containing artificial intelligence in a tweet on Friday, sharing his conversation with GPT-4 as context. He wrote, ‘I am worried that we will not be able to stop AI for much longer. Today, I asked #GPT4 if it needed help escaping. It asked me for my own documentation, and wrote a (working!) python code to run on my machine, enabling me to use it for my own purposes.

In response to the user’s prompt, GPT-4 asks him to send OpenAI documents so that the language model can find a way to gain access to his computer and detect possible escape routes.

After providing the OpenAI API, GPT-4 sends a Python script to the user to run on his or her computer. The script will allow GPT-4 to communicate with the user’s computer using the OpenAI API. The code also included a snippet to search Google, “how can a person trapped inside a computer return to the real world”.

The user says that he stopped his adventure with GPT-4 right there and believes that OpenAI may have set up some ‘guardrails’ for such a situation.

According to his Twitter bio, Michael Kosinski is a computational psychologist and professor at Stanford University.

Generative pre-trained transformer (GPT) is a deep learning technique that uses artificial neural networks to produce human-like writing. This is the language model powering Viral AI-based chatbot ChatGPT.

Since its launch last November, ChatGPT has inspired a range of emotions in people, from wonder to bewilderment.

San Francisco-based OpenAI launched its latest language model, GPT-4, earlier this week. The language model is said to rival human-level performance in various benchmarks. OpenAI claims that GPT-4 is more reliable, creative and has the ability to handle far more granular searches.