There is an intelligence man, who I know to use the machine as a people, you can sail to the operating system, open the program and to carry out the duties of the complex.
It is not science fiction, but is a result of an update to the platform, Anthropic, one of the rivals to the more severe the ChatGpt.
The company's artificial intelligence has made an update to the chatbotin her Claude, with a feature called, ‘the use of a computer " that allows IT to interact directly with the application in the computer, by opening the scenario automation to be you fight the first time. Claude 3.5. the Sonnet, and this was the name of the update, do not be limited on the language of the processing, but you can imitate the actions of human as well as move and click the cursor of a mouse, typing on the keyboard, and even the language, imitating the behaviour of the user on the computer.
The update has been pushed for developers, who are able to use it to build their platforms on Claude. Through the analysis of images of the screen, and the interpretation of the user interface, while the ChatGpt learn to sail in the operating system, open the program and to carry out the duties of the complex.
As it underlines the Anthropic in his blog, the official, up to date, a user of the human retains control by giving the chatbot, it guides what you need to do, in the absence of which there is no action to be independent of the machine. Anthropic it also acknowledges that the model can be “stuck” in action-based, and encourages developers to start with the task and with low risk. Prior to the announcement, Anthropic explained that the update was subjected to the tests of the safety of the carried out together by the Institute for Security, MADE in the USA, and the Institute for Safety in the United Kingdom.
Discussion about this post