Operator is driven by a technology known as a computer using agent (CUA), which enables the model to "view" content on the internet and perform actions that a human would typically carry out using a mouse and keyboard.

This implies that Operator can complete the entire lifecycle of a task, such as making reservations or enrolling individuals in programs.

I felt it was necessary to tackle the lingering obstacles to widespread adoption of this agentic AI tool as spring transitions into summer.

The Reality Surrounding Operator

The primary issue is cost. Operator currently costs $200 per month, whereas all the other features of ChatGPT are available for $20 per month.

It appears that the majority of typical users are holding off until the price decreases.

Read this critique from Mike Todasco on Medium, and you'll see that he simply doesn't believe the technology is worth $200 a month:

"If this is the future, then I don't think we need to worry about AI Agents taking our jobs," Todasco writes. "Operator is a mess, and certainly not worth an additional $180/month. I spent several days searching for any usefulness in it. In the end, I had to abandon this experiment."

What if it were $40 a month? Where would daily user numbers be at this moment?

Of course, we don't truly have anything to compare it to since OpenAI hasn't disclosed user numbers for Operator yet.

Vaguely Agentic Systems

One might also argue that Operator is somewhat agentic – while it has the capability to use the internet, it lacks built-in task management tools. It's more of a do-it-yourself, task-oriented system.

The CUA is undoubtedly fascinating as a technological advancement – it merges previous research on computer vision and tool usage to provide the sort of environment we've been anticipating for a long time.

Nonetheless, for most people, it remains too costly.

Access and Privacy Concerns

Currently, individuals are incorporating their own data into ChatGPT in fine detail to generate responses.

The same principle applies to task-based systems. You'll need to determine how much data you're willing to entrust Operator with to enable it to function effectively.

Additionally, you'll have to consider how much you wish to delegate and what you prefer to retain for yourself.

Furthermore, we must figure out, as users, how to handle the hacker community, where malicious actors will likely attempt to make Operator perform tasks you don't want it to.

Nevertheless, there is immense potential here, which is why I thought it deserved a mention today to discuss the possibilities ahead, especially since the company recently introduced Operator o3.

Cark Franzen at VentureBeat offers these possible use cases:

"Data engineers can delegate manual web interactions—such as data verification and scraping—with greater assurance, freeing up time for higher-level optimization work. Security professionals, on the other hand, gain a safer method to simulate user behavior in audits and incident response exercises, thanks to the model's layered safety mechanisms."

Over at Reddit, OpenAI conducted an AMA (Ask Me Anything) on Operator, during which VP of Research Jerry Tworek stated:

"We… already have a product interface that can perform actions on your computer … we're planning to make some enhancements soon, and it can become a very practical tool then."

If you concur with this viewpoint, we are on the verge of substantial user bases experimenting with the first over-the-counter agentic systems of their kind. We merely require a slight reduction in pricing.

The above is the detailed content of The Power And The Promise Of ChatGPT Operator. For more information, please follow other related articles on the PHP Chinese website!