ChatGPT deep research tool can create reports of hundreds of online sources


There are no two ways in this, in Openai there is a newly acquired sense of urgency. Two days after release O3-Mini To the world, the company made an unexpected announcement on Sunday evening, rChecking deep studiesThe new function allows ChatGPT to find, analyze and synthesize hundreds of websites and online sources to create reports “at the analyst level”.

In addition to ordinary text questions, users can upload files, including PDF files and electric tables, when serving ChatGPT in this way. Then the chat bot will be required “from 5 to 30 minutes” to collect the answer, the side panel documenting the progress and quotes of the agent when it works. “This reaches in dozens of minutes, which will require a person for many hours,” Openai says about a new function.

“Our final aspiration is a model that can reveal and discover new knowledge,” said Mark Jong, the Openai research director, while identifying a direct broadcast of the company. “This is a core for our (artificial general intelligence) roadmap.”

As for the restrictions, Openai says that ChatGPT can sometimes hallucinate facts or draw incorrect conclusions during deep studies, although “at a particularly lower” speed than other modern models. In addition, an agent can sometimes try to distinguish authoritative information and rumors with all his might. Users can also notice some formatting errors. “We expect all these problems will quickly improve with great use and time,” the company notes.

If all this sounds familiar, this is because Advanced Suite Google includes its own Deep research featureWhich not only shares the same name, but also generally offers the same set of opportunities. One significant difference between them is that Google offers access to twins advanced after $ 20 a month One premium plan aiThe field is on the contrary, you will need $ 200 a month ChatGPT for Planning to start using the DEEP Research Openai version today.

“Deep studies in CatgPT are currently very intensively calculated,” the company calculation, adding that this will limit Pro users for 100 requests per month. “The longer the study of the request is required, the more conclusion it requires.”

Openai says that he is working on a version of deep research based on a smaller, more economical model. In turn, this will allow the company to offer “significantly higher rates of the bet.” At the same time, Openai hopes to get a tool in the hands of PLUS users “about a month” after a round of security testing. As in most other recent issues of the company, European users will need to wait before they can test this tool for themselves, and deep research is not yet available for people in the UK, Switzerland and a wider European economic zone.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *