On January 14, OpenAI announced the rollout of a new feature for ChatGPT called “Tasks.”
According to Karina Nguyen, a Research Engineer at OpenAI, Tasks allows ChatGPT to “manage tasks asynchronously on your behalf,” which is similar to the product that other companies are calling AI agents—artificial intelligence (AI) systems capable of completing multi-step tasks autonomously.
In a demo shared by OpenAI, Tasks is shown sending daily stock price updates, pushing reminders to exercise via phone notifications, and summarizing the latest news after the user gives the system an editable set of instructions.
Tasks is currently in beta for individuals with a Plus, Pro, or Teams subscription to ChatGPT but will be available to everyone with a ChatGPT account in the future.
Can ChatGPT’s Tasks be considered OpenAI’s AI agent?
Now that Tasks has been released, I find myself asking whether this feature qualifies as an AI agent.
From what we’ve seen in the demo—and from my own experience testing the feature out—Tasks primarily handles very simple operations that are more reminiscent of scheduling a post or setting a recurring alarm than completing a multi-step task. I would prefer to have it off my plate and in the hands of an autonomous agent. I think the introduction of Tasks is a step forward for ChatGPT since it can do a few things for its users without the constant prompting currently needed when using GPT. However, so far, what Tasks can do for its users doesn’t seem groundbreaking or complex.
In addition, Tasks operates exclusively within ChatGPT’s ecosystem and can’t execute tasks outside this walled garden. For instance, it cannot draft and send emails or go to a specified website and begin adding items to an online shopping cart.
These limitations make it challenging to call Tasks an AI agent. I even think OpenAI consciously decided not to use the term “agent” when introducing this feature because they know that Tasks is more of an advanced notification/reminder tool than an actual agent capable of executing multi-step processes autonomously.
For an artificial intelligence system to rightfully be called an AI agent, it needs to be able to handle complex tasks independently, often spanning multiple websites, platforms, or applications. One of the primary reasons the world is so excited about AI agents is because of their potential to streamline multi-step processes, which should free users from micromanaging AI as they typically need to do today.
Other companies have demonstrated agents that can leave their ecosystems, complete external tasks, and return to the user with the result—sometimes requiring just a confirmation click. When you compare those capabilities to what Tasks can do today, it is clear that it lacks this level of functionality that makes AI agents what they are.
OpenAI’s Tasks: Progress toward AI agents, but not quite there yet
Although its scope is still limited, OpenAI’s Tasks feature indicates that progress is being made toward a world with AI agents. Tasks definitely incorporate some of the foundational elements necessary for AI agents but just aren’t advanced enough to be considered an AI agent itself. Until Tasks can operate outside of OpenAI’s walled garden and tackle more complex, multi-step tasks, I don’t think we can call it an AI agent.
In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.
Watch: Micropayments are what are going to allow people to trust AI
title=”YouTube video player” frameborder=”0″ allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen=””>
