OpenAI is said to be in the process of developing “agent software” capable of assuming control of your device and autonomously executing intricate tasks, as reported by The Information. This agent from OpenAI is designed to operate seamlessly across various applications on your computer, executing actions such as clicks, cursor movements, and text typing. Essentially, it introduces a novel kind of operating system, potentially reshaping the way you interact with your computer entirely.
As part of Sam Altman’s vision to transform ChatGPT into a “supersmart personal assistant” surpassing the capabilities of platforms like Siri, OpenAI is actively developing agent software, a concept also pursued by companies such as Google and Meta, as reported by The Information. The core idea revolves around an AI agent assuming control of your computer, executing tasks on your behalf. For instance, users could instruct the AI agent to convert a document into a spreadsheet and analyze the data.
OpenAI is not only creating a computer control agent but also working on a web browsing AI agent. This web browser agent aims to undertake complex tasks like planning flights and gathering data about companies, tasks that go beyond the current capabilities of ChatGPT. Altman envisions OpenAI evolving into an AI agent company, evident in the recently launched GPT Store, which serves as a marketplace for AI agents. During the introduction of the GPT Store at DevDay on Nov. 6th, Altman hinted at the concept of agent software, envisioning a future where users can simply request their computer to perform various tasks, highlighting the significant potential of these capabilities in the field of AI.
A crucial aspect of this transformation is interoperability, necessitating an agent to seamlessly access and navigate all the applications on a user’s device. However, this shift requires users to relinquish a considerable amount of control to OpenAI and demands seamless collaboration among various apps. The prospect of letting ChatGPT take charge might raise concerns among users.
Andrej Karpathy, a founding member of OpenAI, recently expressed in a YouTube video that labeling large language models as chatbots is not accurate. Instead, he referred to them as the “kernel process of an emerging operating system.” These interconnected AI agents are envisioned to go beyond mere conversation; they will actively utilize your computer.
“Eventually, you’ll just ask the computer for what you need, and it’ll do all of these tasks for you,” remarked Altman. Such capabilities are often discussed in the AI field as ‘agents,’ and their potential benefits are anticipated to be significant.
A crucial aspect of this transformative concept is interoperability, necessitating an agent to have access and the ability to navigate all the applications on a user’s device. Embracing this vision requires users to cede a substantial level of control to OpenAI and demands seamless collaboration among various apps. This shift might evoke unease among users who are cautious about letting ChatGPT take the lead.
In a similar vein, Nvidia’s Jensen Huang, speaking at The New York Times Dealbook in November, highlighted the transformative impact of AI on computing, specifically mentioning the “agent software revolution.” According to Huang, we are at the dawn of a new era in computing, a paradigm shift not witnessed in the last 60 years.
While agent software holds the potential to fundamentally alter how we interact with technology, it also involves relinquishing control to companies like OpenAI. The question arises: Are users willing to entrust ChatGPT with the task of using their computers on their behalf?