You can simply say no; you are not obligated to comply. OpenAI is venturing beyond the confines of a chatbot interface with its new ChatGPT Agent, announced on Thursday. This initiative aims to enter the expanding “agentic” space, enabling AI models to perform designated tasks autonomously.
The ChatGPT Agent operates using a virtual computer to fulfill specific requests from start to finish. Examples include tasks like, “check my calendar and summarize upcoming client meetings based on recent news,” “plan and purchase ingredients for a Japanese breakfast for four,” and “analyze three competitors and create a slide deck.” Once assigned a task, the Agent works independently, navigating websites, opening apps, and completing the job without further input.
According to The Verge, the ChatGPT Agent is built on a new, unnamed model developed specifically for this purpose. It combines the capabilities of the Operator tool, which can navigate web browsers, and Deep Research, which can perform multi-step research and analysis tasks—both introduced earlier this year. The Agent tool has also been trained for complex tasks requiring multiple tools, such as switching between a browser and the terminal.
ICYMI: Mustapha Hamid, Others Charged over GHC280 Million Extortion and Money Laundering Scheme
Initial impressions of the Agent are mixed. The Verge noted that while it can handle multi-step operations, it may be slow, and isn’t designed for users to monitor its progress. Wired echoed this sentiment, reporting that it took an hour to order cupcakes for OpenAI’s research lead, Isa Fulford, who stated that the time saved was worth it since she “didn’t want to do it.” A notable feature highlighted by Wired is the Agent’s “replay” function, allowing users to review the steps taken to complete tasks.
AI agents are generating buzz as potential replacements for certain human tasks. For example, Klarna used AI agents for customer support but later reverted to human operators due to quality issues.
The ChatGPT Agent will be accessible to subscribers of the ChatGPT Pro, Plus, or Team plans by selecting “agent mode.” OpenAI plans to introduce the agent to ChatGPT Enterprise and Education users later this summer.
SOURCE: GIZMODO
