OpenAI announced a new feature for ChatGPT that allows the popular chatbot to execute actions on behalf of the user. This is part of an industry-wide effort to change how people do things on the internet: tech giants hope that, instead of jumping from one app to another and manually searching the web, users will one day be able to rely on agents to do it all.
The new agent mode in ChatGPT, which is rolling out immediately, is another sign that tech giants are doubling down on digital assistant initiatives with significantly advanced capabilities. It also intensifies the competition between OpenAI and Google, which is pursuing similar ambitions with its Gemini assistant.
OpenAI said on Thursday that ChatGPT’s new agent mode “thinks” and “acts” using its own virtual computer, allowing it to manage complex action-oriented requests.
For example, users will be able to give commands like “Check my calendar and let me know about upcoming client meetings based on recent news” or “Plan and buy the ingredients to make a Japanese breakfast for four,” according to the company in a blog post.
In a video demonstration, OpenAI employees typed a long, detailed prompt asking the agent to help the user prepare for a wedding. It included specific instructions such as “Find an outfit that fits the dress code” and added that it should propose five options, along with hotels that could accommodate a couple of days of buffer time around the event.
The new feature is available to those subscribed to a Pro, Plus, or Team plan.
It builds upon and combines the capabilities of the ChatGPT Operator and Deep Research tools that OpenAI already offers; Operator navigates the web, while Deep Research analyzes online resources to, for example, compile reports.
This update is another step in OpenAI’s efforts to turn ChatGPT into a more comprehensive universal assistant. At the same time, the broader AI industry is also grappling with how to address significant shortcomings and privacy concerns surrounding this technology.
AI models are still prone to hallucinations and bias, and can act unpredictably, as xAI’s Grok chatbot demonstrated last week by posting antisemitic content after being prompted to do so.
In a blog post, OpenAI acknowledged that the new ChatGPT functionality introduces new risks. It stated that it has limited access to model data and that certain tasks, such as sending an email, require user supervision. The model is also trained to reject “high-risk tasks,” such as bank transfers, according to the company.
“I would explain this to my own family as something innovative and experimental; an opportunity to test the future, but it’s not something I would use for high-risk uses or with a lot of personal information until we have the opportunity to study and improve it in practice,” commented Sam Altman, CEO of OpenAI, in a post on X where he announced the agent.
He advised users to be cautious when giving ChatGPT access to personal information. For instance, granting access to a calendar to coordinate a group dinner might make sense, but the agent wouldn’t need calendar access to buy clothes on behalf of the user.
The announcement comes at a time when tech giants are increasingly pushing the development of AI agents in their pursuit of victory in the AI race.
Google made a series of AI-related announcements during its developer conference in May, including an agent that can make restaurant reservations and buy event tickets, among other tasks. Apple is working on a more advanced version of Siri that can use apps on behalf of the user, although that update is delayed indefinitely.
image credits for this post: deultimominuto.net
