Today, well-known popular science blogger 'Three Dads Laboratory' announced via Weibo that Tencent's Yuanbao app has integrated DeepSeek technology and embedded advertising content in its answers. Screenshots shared by the blogger show that in the suggestions and reminders provided by the Yuanbao app to avoid pitfalls, users are explicitly recommended to compare prices and make appointments through legitimate platforms like 58 Daojia, as well as to inquire prices through Kuaihuishou or local legitimate recyclers. These suggestions have been criticized for potentially containing embedded advertisements.

Recently, OpenAI made significant updates after launching its flagship AI models o3 and o3-mini, altering the way o3-mini responds. Now, this model not only answers users' questions but also showcases its thought process, providing greater transparency to users. This change marks an important step for OpenAI in enhancing user experience, making the use of AI tools more human-centered. The updated o3-mini, especially in 'high reasoning' mode.

At the end of 2024, Baidu announced its annual AI prompt for 2024: 'Answer'. This choice reflects the year-long reliance and expectations people have for AI, as well as the role AI plays in societal emotions and personal confusion. Throughout the year, many individuals began seeking answers from AI, with frequently used terms such as 'answer', 'why', 'what is', 'help me', and 'how', revealing 2024's social sentiments and showcasing the confusion and pursuit within people's hearts.

Baidu Wenku recently launched a new feature called 'AI All-Net Search', aimed at enhancing the efficiency and accuracy of information retrieval through powerful AI technology, providing users with a comprehensive and diverse knowledge graph. In this era of information explosion, users need to quickly capture key points from vast amounts of data, while traditional search engines often drown people in complex results. 'AI All-Net Search' redefines information retrieval by integrating various types of information from the internet, making every search an efficient, precise, and enjoyable journey of knowledge exploration.

A recent study led by the Technion-Israel Institute of Technology reveals that large language models (LLMs) may be more knowledgeable than they appear. Researchers found that LLMs' internal representations encode information about the correctness of their outputs, allowing them to identify the correct answer even when they ultimately generate an incorrect one. The research team focused on analyzing LLM errors in long text generation, which is more aligned with their real-world application scenarios. They constructed an error detection dataset to compare the model's outputs.

Recently, a study published in Nature revealed a thought-provoking phenomenon in the development of artificial intelligence: as large language models (LLMs) continue to evolve, they increasingly tend to provide answers with 'full confidence', even when those answers may be incorrect. This phenomenon has sparked broad discussions on AI reliability and the risks of use. The research team, led by José Hernández-Orallo from the Valencia Institute of Artificial Intelligence in Spain, conducted an in-depth analysis of AI models.

Google is testing a new feature called 'Ask Photos', allowing users to explore their photo library in an innovative way. By asking questions such as 'Where did we camp last time at Yosemite?' or 'What did we eat at the hotel in Stanley?', users can receive answers and retrieve related images. Powered by Google's Gemini AI model, it analyzes photo content to provide answers and images, supports natural language searching, and sorting by date or relevance, with support for more languages rolling out in the coming weeks. This feature has been launched on Android and iOS platforms.

MathGPTPro is an AI math learning tool developed by a Chinese genius, transforming how students worldwide learn. It outperforms GPT-4 by 20%, providing 24/7 math problem solutions, personalized learning plans, and in-depth problem-solving discussions. Features include screen reading, multiple explanation methods, LaTeX code support, and affordable subscription options. The innovation of its Chinese founders highlights their impact in global tech ....

Researchers from the University of Reading in the UK secretly submitted AI-generated exam answers, successfully tricking professors and achieving better scores than real students. The project used ChatGPT-4 to generate unedited answers, which were submitted under the guise of a fictitious student for an online assessment in an undergraduate course.

In this AI-driven era, it seems that every company is racing to use AI technology to directly provide answers to users, simplifying the decision-making process, and even bypassing thinking altogether. However, the iA writing tool takes a different approach; it not only aims to assist users with writing but also seeks to stimulate their thinking.

LexisNexis first revealed its Lexis+ AI initiative last May, utilizing large language models to answer legal questions. The company has lifted restrictions on Lexis+ AI, making it widely available to clients in the U.S. Lexis+ AI provides reliable results with 'hallucination-free' legal citation links. It enables conversational searching, document drafting, case summarization, and document analysis. LexisNexis claims that Lexis+ AI can enhance the speed of legal professionals.

Trends in the Application of Large Models: The article mentions how large model AI is rapidly infiltrating various industries, but at the same time exposes the logical reasoning flaws of large models. Limitations of Knowledge in Large Models: A founding member of OpenAI candidly stated that the knowledge of large models is much more incomplete than you might imagine. Tests in both fictional and real scenarios indicate that large models suffer from a 'Reversal Curse' and are unable to perform reverse reasoning. Doubts About the Reliability of Large Models: Logical reasoning forms the basis for large models to solve problems correctly. The Reversal Curse indicates deficiencies in the reasoning capabilities of large models, which may impact various important application areas.

A new study has found that ChatGPT's accuracy in answering software engineering questions is below fifty percent. The results show that 52% of ChatGPT's answers to software engineering questions are incorrect. While ChatGPT performs better on general questions, 77% of its answers are still overly verbose.