OpenAI is pushing ChatGPT closer to becoming a full-fledged AI assistant with its new "Tasks" feature. While it's not quite as hands-on as Siri or Google Assistant yet, it can now handle scheduled reminders and recurring requests.
The company is rolling out Tasks right away for paying subscribers - that includes ChatGPT Plus, Team, and Pro users worldwide. Free users will have to wait, though, as OpenAI hasn't announced when (or if) they'll get access. For now, the company wants to gather feedback from paid users first.
Video: OpenAI
To use Tasks, users need to select the "4o with scheduled tasks" option in ChatGPT's model menu. From there, they can tell the AI what they need and when they want it done, whether it's a one-time reminder or something that repeats. OpenAI shared some examples of what users can ask for:
- A heads-up when their passport is six months from expiring
- Daily weather forecasts at 7 AM
- Weekly weekend activity suggestions based on location and weather
- Morning affirmations at 7 AM
- Daily knock-knock jokes at 6 PM for bedtime stories
The system can even suggest tasks based on your chat history, though you'll always have the final say on whether to set them up. Users can manage their tasks either in their regular chat window or through a dedicated task manager, which is currently only available on the web version.
Video: OpenAI
Up to 10 active tasks at the same time
During the beta phase, users can only have 10 active tasks at once. When a task is completed, ChatGPT sends notifications through desktop, browser, or smartphone alerts.
While the system will eventually be able to perform scheduled internet searches, it can't yet do continuous background searches or make purchases. This means we're still far from having an AI that can independently plan and book your vacation.
Behind the scenes, OpenAI is developing more sophisticated agent systems, including one called "Operator" that's designed to write code and book travel arrangements on its own.
With Tasks, ChatGPT is stepping into territory traditionally held by voice assistants like Google Assistant, Siri, and Alexa. These services are all trying to evolve from rigid, rule-based systems into more flexible AI companions. However, the challenge lies in making language models reliable enough for everyday use.
