ChatGPT introduces ‘agent mode’, allows people to think and act like people
लगानी न्यूज
Kathmandu. OpenAI, a pioneer in artificial intelligence (AI) technology, has launched a new ‘Agent Mode’ in ChatGPT, which can automatically complete tasks on behalf of users. On Thursday, the company released this agent mode through a blog post and informed that it will be made available to premium users in a phased manner.
“From now on TAG_OPEN_div_34, ChatGPT can be ordered to look at my calendar and summarize upcoming client meetings with recent news, or plan and purchase Japanese lunch materials for four people,” the company said.
According to OpenAI, chatGPT’s agent mode has acquired the ability to think and act using its virtual computer. TAG_OPEN_div_32 It can carry out complex instructions such as planning, purchasing content, preparing reports, or gathering information by looking at websites.
In a video demo, OpenAI employees requested users to prepare for the wedding ceremony. It included instructions to “find dresses that fit the dress code”, “suggest five options”, and “find hotels that can accommodate before and after the ceremony”.
This feature is available only for users who plan ChatGPT’s Pro, Plus, or Team for the time being. This update works based on chatGPT operators and devices called Deep Research. While the operator browses the Internet, Deep Research collects the facts and makes a detailed analytical report.
OpenAI has acknowledged the risks that come with this new feature. According to the company, it will have limited access to the user’s personal data and such as ‘sending email’ will not be done without the user’s consent. However, it will reject high-risk actions such as bank transfers.
“It’s TAG_OPEN_div_24 cutting-edge and experimental, a glimpse of the future, but I don’t recommend using it for very personal or sensitive information right now,” OpenAI CEO Sam Altman wrote on Social Media X. According to him, it is okay to give calendar access to coordinate group dinners, but such access is not necessary to buy clothes.
While announcing this, other technology companies are also engaged in the development of digital agents. At its developer conference in May, Google announced features that allow AI agents to book restaurants, buy tickets, and so on. Apple also plans to add such capabilities to the new version of Siri, but its update has now been postponed indefinitely.
However, even if an AI agent tries to make things easier, the use so far may lead to hallucinations, biased answers, and unexpected behavior. Recently, elon Musk’s company XAI’s Grok chatbot presented objectionable and anti-Semitic content, after which the discussion of the risks in the AI model started again.
ChatGPT Agent Mode in this context indicates that making AI a digital assistant and taking the way the Internet is run in a new direction. But the company suggests that users should use this technology with caution.
प्रतिक्रिया दिनुहोस्