OpenAI has launched the ChatGPT Agent, a new AI tool designed to assist users by managing tasks like scheduling and research. This agent combines features from OpenAI’s previous AI agents, the Operator and the Deep Research agent, delivering a user-friendly experience within the ChatGPT interface. However, the ChatGPT Agent requires user approval for significant actions to avoid errors, a design choice aimed at safety but limiting automation efficacy. Despite its advanced capabilities, user feedback highlights issues like slow performance and erroneous outputs, exemplified by a failed trip-planning request. Presently, the ChatGPT Agent is available for Pro users, with a monthly cap of 400 prompts, and plans for extension to Plus and Team subscribers. Despite the potential benefits, its reliability remains questionable, prompting discussions about AI autonomy and safety in practical applications. This launch comes amid ongoing scrutiny of AI’s trustworthiness and efficiency.
Source link
