OpenAI has unveiled a new capability of ChatGPT called Deep Research for in-depth data exploration and analysis, complete with step-by-step explanations and supporting references. Deep Research is tailored for individuals in specialized fields like finance, science, engineering, and law, where traditional deep data search methods consume considerable time to verify data credibility systematically. Additionally, Deep Research can be applied to conducting comparative product data searches with intricate details, such as cars and electronics.
Operating on the customized o3 model, Deep Research has been tested with Humanity’s Last Exam, an academically profound question, scoring as high as 25.3% compared to o1 at 9% and o3 at 10%. To access, users can select the Deep Research button within the chatbox, where questions are deliberated and answered through a more profound approach. An illustrative example below analyzes the app market odds for iOS in different countries. GPT-4o’s responses are quicker, broader, and less detailed. However, Deep Research provides answers with accompanying tables, figures, source references, and summaries, albeit requiring more time to deliver (OpenAI suggests a timeframe of 5-30 minutes).
ChatGPT Deep Research is now available for ChatGPT Pro customers through the website, with a daily limit of 100 queries. Plus and Team customers will have access subsequently, followed by the Enterprise group. OpenAI notes that the Deep Research function currently has limitations and may provide inaccurate data or references, and is unable to distinguish and filter out gossip-related information.
Meanwhile, Google introduced deep data search feature called Gemini, alongside Gemini 2.0 last December.
TLDR: OpenAI introduces ChatGPT’s Deep Research for in-depth data exploration and analysis, catering to specialized fields. It provides thorough responses with references but may take longer. Google also unveiled a similar feature called Gemini.
Leave a Comment