OpenAI is preparing a frighteningly capable AI. It will partially take over your computer

OpenAI is preparing a frighteningly capable AI. It will partially take over your computer
  • OpenAI is preparing artificial intelligence to take over the browser
  • The project, codenamed Operator, should arrive in January next year
  • The creator of ChatGPT will thus lag behind Google or Anthropic

The organization behind the generative tool ChatGPT is preparing even frighteningly capable artificial intelligence. According to the agency Bloombergwhich refers to two of its resources, will be taken over by an AI labeled Operator control over your web browser. Thanks to this, power will be needed to program or book tickets.

OpenAI management would like to introduce the new tool next year in Januaryin the preview phase of the available for developers within the API. It is expected to compete with Google’s alleged Jarvis project, which is expected to arrive a month earlier with similar features.

The giants in the field of artificial intelligence are evidently dealing in a big way with the concept of so-called AI agentsi.e. software that is able to work with minimal user supervision.

OpenAI is preparing a frighteningly capable AI. It will partially take over your computer ChatGPT home page with text box and suggestions

It hasn’t been that long since the company Anthropic introduced a new experimental trick to the model Claude 3.5 Sonnet allowing artificial intelligence to control the computer as if a person were sitting behind it. He can analyze the content on the screen, move the cursor, click the buttons a fill in the text fields.

“We’re releasing (computer use) early to get developer feedback and expect it to improve rapidly over time,” Anthropic said at the time.

Slowly but surely, we are getting to the stage where artificial intelligence will be able to do it for us perform a substantial part of routine online operations. However, it remains to be seen how reliably models from Google and OpenAI will be able to work.

Looking at the state of today’s language models, which can greatly facilitate work, but they tend to make things upI am personally a little skeptical. I’m probably not the only one who wouldn’t dare trust an artificial intelligence with my payment card in order to buy flights or secure accommodation…

Would you let AI control your browser?

Source: Bloomberg

Source: www.svetandroida.cz