Useful information
Prime News delivers timely, accurate news and insights on global events, politics, business, and technology
Operai announces a new “agent” of AI designed to help people to conduct an in -depth and complex investigation using chatgpt, the company’s chatbot platform.
Properly, it is called deep research.
Operai said in A blog post Published on Sunday that this new capacity was designed for “people who work with intensive knowledge in areas such as finance, science, politics and engineering and need thorough, precise and reliable research.” It could also be useful, the company added, for anyone who makes “purchases that generally require careful investigation, such as cars, appliances and furniture.”
Basically, Chatgpt’s deep research is intended for cases in which he not only wants a quick response or summary, but must regularly consider the information of multiple websites and other sources.
Operai said it is making deep research available for Chatgpt Pro users today, limited to 100 consultations per month, with support for Plus users and the team that comes next, followed by Enterprise. (Openai points to a positive launch in approximately one month from now on, the company said). It is a geographical launch; Operai did not have a launch timeline to share for chatgpt customers in the United Kingdom, Switzerland and the European economic area.
To use ChatGPT Deep Investigation, you will simply select a “deep research” in the composer and then enter a query, with the option of attaching files or spreadsheets. (It is an experience only on the web for now, with the integration of mobile and desktop applications at the end of this month). Deep investigations could take between 5 and 30 minutes to answer the question, and will receive a notification when the search is completed. .
Currently, Chatgpt Deep Research results are only text. But Operai said he intends to add integrated images, data visualizations and other “analytical” outputs soon. Also in the road map is the ability to connect “more specialized data”, including internal resources “based on subscription” and internal, OpenAi added.
The big question is, how precise is Chatgpt Deep Investigation? Ai is imperfect, after all. It is prone to hallucinations and other types of errors that could be particularly harmful in a “deep research” scenario. That is the reason why Operai said that each Chatgpt’s deep research output will be “completely documented, with clear quotes and a summary of (the) thought, which facilitates the reference and verifies the information.”
The jury is on whether these mitigations will be enough to combat AI errors. Web search function with Openai AI in Chatgpt, search for chatgpt, not often Make glasses and give incorrect answers to the questions. Techcrunch’s tests discovered that the search for ChatgPT produced less useful results than Google search for certain consultations.
To reinforce the accuracy of Deep Research, OpenAi is using a special version of its “reasoning” model of O3 recently announced that it was trained through reinforcement learning in “real world tasks that require use of browser tools and piton”. Reinforcement learning essentially “teaches” a model through proof and error to achieve a specific objective. As the model approaches the objective, it receives virtual “rewards” that ideally, improve the task in the future.
He said that this version of the Operai O3 model is “optimized for web navigation and data analysis,” and adds that “take advantage of the reasoning to search, interpret and analyze large amounts of text, images and PDFs on the Internet, turning as necessary as necessary In reaction to the information you find (…) the model can also navigate the files loaded by the user, draw and iterate in graphics using the python tool, embed the generated graphics and the images of the websites in their answers, and cite specific prayers or passages from their sources. “
The company said it tested the deep research of ChatgPT using The last examination of humanityAn evaluation that includes more than 3,000 questions at expert level in a variety of academic fields. The O3 model that promoted deep research achieved a precision of 26.6%, which might seem a failed rating, but the last examination of humanity was designed to be more difficult than other reference points to stay ahead of the advances of the model . According to Openai, Deep Research’s O3 research model was far ahead of Gemini Thinking (6.2%), Grok-2 (3.8%) and OpenAi’s GPT-4O itself (3.3%).
Even so, Openai points out that Chatgpt Deep Investigations has limitations, sometimes making wrong mistakes and inferences. Deep research may have difficulty distinguishing authorized information from rumors, the company said, it often cannot transmit when it is uncertain about something, and can also make format errors in reports and appointments.
For any person concerned about the impact of generative AI in students, or in any person trying to find online information, this type of deep and well -cited production probably sounds more attractive than a deceptively simple chatbot summary without quotes. But we will see if most users will really submit the result to a real analysis and a double verification, or if they simply treat it as a more professional text to copy pasta.
And if all this sounds familiar, Google announced a similar characteristic with exactly the same name less than two months ago.
