OpenAI has integrated a function called ‘Task’ into ‘ChatGPT’ that lets you schedule notifications or repeat requests. This appears to be the early stages of artificial intelligence (AI) agents.
Open AI announced on the 14th (local time) that it might release a beta version of Task for ChatGPT paid users through X (Twitter). ChatGPT Plus, Team, and Pro users world wide will have the option to make use of this feature starting this week.
This feature allows users to set easy notifications on ChatGPT, equivalent to “Notify me when my passport expires in 6 months,” and ChatGPT will send push notifications on all platforms where the motion is enabled. You may also arrange recurring requests, equivalent to “Remind me of my weekend plans every Friday based on my location and weather forecast,” or “Remind me of the news briefing day-after-day at 7 a.m.”
Tasks works with all versions of ChatGPT and sends notifications via desktop, web, and mobile. Nonetheless, settings are currently only possible in the online version of ChatGPT. Also, you can’t set tasks through ‘Advanced Voice Mode’.
Users can select ‘4o with scheduled tasks’ from the ChatGPT drop-down menu to execute the duty and send a message notifying the notification or task they wish to create. Moreover, ChatGPT can suggest specific tasks to users based on existing chats.
Functionality remains to be limited. You can not repeatedly perform searches or purchase items through payment. For instance, you’ll be able to tell ChatGPT if a concert ticket to your favorite artist is on sale, but it will possibly’t notify you the moment a ticket goes on sale or tell it to purchase the ticket for you.
OpenAI said it plans to roll out this feature to mobile apps and free users after beta testing.
The launch of this task appears to be step one in upgrading ChatGPT to an AI agent.
In November last 12 months, it was reported that OpenAI demonstrated an agent function that searches the online and orders products for internal employees. Bloomberg then reported that OpenAI plans to release an operator in January 2025.
Nonetheless, Nick Turley, OpenAI product manager, said in a December interview, “It’s unlikely that a separate product called OpenAI Agent will probably be released,” and added, “Agent functions will progressively be added inside ChatGPT.” In truth, the duty announced on this present day was integrated into ChatGPT.
Due to this fact, OpenAI’s agents are expected to be divided into two types: ▲ChatGPT integrated ‘Task’ that helps with easy tasks, and ▲’Operator’ with advanced functions including coding and payment. In accordance with TechCrunch, OpenAI didn’t reply to this.
Meanwhile, AI agent is a feature that CEO Sam Altman identified as a promising technique to achieve artificial general intelligence (AGI) with the present state model.
“We’re confident that we now know how one can construct AGI because it has been traditionally understood,” he said on his personal blog on the sixth. “We expect the primary AI agents to hitch the workforce in 2025, substantially impacting the corporate’s results.” “I imagine we are able to change it,” he said.
It was predicted that this beta testing period could take several months. On the seventh, The Information quoted an OpenAI worker as saying that the discharge of the AI agent is being delayed as a result of rigorous testing against ‘prompt-injection’ attacks.
In other words, it seems that agents with more privileges than regular chatbots will undergo more testing to forestall abuse.
Reporter Lim Da-jun ydj@aitimes.com
