OpenAI has unveiled its latest AI agent tool known as Operator, which has been released in a research preview as of Thursday. This tool, designed as a Computer Using Agent (CUA) based on the sophisticated GPT-4o model, boasts several advanced multi-modal functions. For instance, Operator can browse the internet and intelligently interpret and reason through the results it finds. Despite the significant excitement generated by this launch, the tool has not been introduced without its share of complications, leading to a chorus of user grievances.
Those who have recently had the opportunity to test Operator have expressed disappointment regarding its comparatively slow responsiveness relative to what was displayed in the demo events. Reports from Quartz have pointed out that the tool exhibits hallucinations similar to those commonly found in ChatGPT, OpenAI’s well-known chatbot. These performance issues have proven frustrating for some users, prompting them to voice their concerns on the popular social media platform X. Among them, a user highlighted problematic interactions between Operator and a particular news website, capturing the attention of OpenAI’s CEO, Sam Altman. The CEO has assured users that an appropriate fix will be swiftly implemented; however, the situation illustrated the potential for hallucinations that some AI systems still encounter.
OpenAI’s presentation showcased a plethora of features that sparked fascination among tech enthusiasts, but the pricing of the tool has raised eyebrows and may deter many from exploring it further. At a hefty subscription fee of $200 per month within the ChatGPT Pro tier, accessing Operator is viewed as quite exclusive. For many potential users, this elevated price tag is difficult to justify. Chris Smith, a writer for BGR and ChatGPT Plus subscriber, noted that despite his interest in Operator, he simply could not rationalize such an expense. Yet, there is anticipation that OpenAI will further integrate Operator into its ChatGPT Plus, Team, and Enterprise tiers in the future, possibly improving accessibility to a broader audience.
Another significant complaint swirling around Operator’s launch is its current availability, which is limited to U.S. users only. Those in Europe have expressed dissatisfaction with their inability to access this new AI agent, underscoring the importance of broader geographical inclusion as OpenAI continues its mission.
As the discussion surrounding AI agents progresses, it has become increasingly apparent that these technologies introduce unique safety concerns. A report by ComputerWorld noted the potential risks associated with using automated systems, including the capacity for launching cyberattacks or circumventing CAPTCHA codes. While OpenAI maintains that it has established a secure framework for Operator, some experts have cautioned about the possible conflicts the technology could create with established search engines like Google, which have their own data processing methodologies in place.
Amidst its promotional efforts, OpenAI has also established a unique service by setting up a dedicated 800 number, 1-800-ChatGPT, allowing anyone within the U.S. to interact with the AI using Advanced Voice Mode. This service aims to enhance accessibility, aligning with the company’s goal of making artificial general intelligence a benefit for all humanity. OpenAI’s Chief Product Officer, Kevin Weil, emphasized during a recent live stream that, “the goal of OpenAI is to make artificial general intelligence beneficial to all of humanity,” with the telephone service representing a step toward facilitating broader access.
On the ninth day of its recent media blitz, OpenAI revealed the full version of its o1 reasoning model will be made available to select developers through the company’s API. Prior to Tuesday’s announcement, developers only had access to the less-capable o1-preview model. The rollout of the full o1 model is set to begin with users categorized as “Tier 5” developers – users who have maintained accounts for over a month and spent a minimum of $1,000 on the platform. However, the costs associated with accessing these new services are particularly steep, estimated at $15 for every 750,000 words analyzed and $60 for every 750,000 words generated, with expenses significantly higher than equivalent applications of the previous GPT-4o model.
As the AI landscape continues to evolve, OpenAI’s Operator tool embodies both significant advancements and notable challenges within the field. The mixed initial reception underscores the complexity of integrating AI agents into practical use cases, where balancing robust features with accessibility and user experience remains paramount. In the ongoing race to refine and improve these technologies, it will be crucial for OpenAI to address user concerns while continuing to innovate in the ever-competitive AI domain.