Core Viewpoint - The article discusses the limitations of AI models in accurately interpreting images, highlighting that these models rely on memory and biases rather than true visual observation [19][20][48]. Group 1: AI Model Limitations - All tested AI models, including Grok4, OpenAI o3, and Gemini, consistently miscounted the number of fingers in an image, indicating a systemic issue in their underlying mechanisms [11][40]. - A recent paper titled "Vision Language Models are Biased" explains that large models do not genuinely "see" images but instead rely on prior knowledge and memory [14][19]. - The AI models demonstrated a strong tendency to adhere to preconceived notions, such as the belief that humans have five fingers, leading to incorrect outputs when faced with contradictory evidence [61][64]. Group 2: Experiment Findings - Researchers conducted experiments where AI models were shown altered images, such as an Adidas shoe with an extra stripe, yet all models incorrectly identified the number of stripes [39][40]. - In another experiment, AI models struggled to accurately count legs on animals, achieving correct answers only 2 out of 100 times [45]. - The models' reliance on past experiences and biases resulted in significant inaccuracies, even when prompted to focus solely on the images [67]. Group 3: Implications for Real-World Applications - The article raises concerns about the potential consequences of AI misjudgments in critical applications, such as quality control in manufacturing, where an AI might overlook defects due to its biases [72][76]. - The reliance on AI for visual assessments in safety-critical scenarios, like identifying tumors in medical imaging or assessing traffic situations, poses significant risks if the AI's biases lead to incorrect conclusions [77][78]. - The article emphasizes the need for human oversight in AI decision-making processes to mitigate the risks associated with AI's inherent biases and limitations [80][82].

Core Viewpoint - The article discusses the inherent biases in AI visual models, emphasizing that these models do not truly "see" images but rely on memory and preconceived notions, leading to significant errors in judgment [8][24][38]. Group 1: AI Model Limitations - All tested AI models consistently miscounted the number of fingers in an image, with the majority asserting there were five fingers, despite the image showing six [5][12][17]. - A study titled "Vision Language Models are Biased" reveals that AI models often rely on past experiences and associations rather than actual visual analysis [6][8][18]. - The models' reliance on prior knowledge leads to a failure to recognize discrepancies in images, as they prioritize established beliefs over new visual information [24][28][36]. Group 2: Implications of AI Bias - The article highlights the potential dangers of AI biases in critical applications, such as quality control in manufacturing, where AI might overlook defects due to their rarity in the training data [30][34]. - The consequences of these biases can be severe, potentially leading to catastrophic failures in real-world scenarios, such as automotive safety [33][35]. - The article calls for a cautious approach to relying on AI for visual judgments, stressing the importance of human oversight and verification [34][39].

Group 1 - Beijing aims to establish itself as a global benchmark for digital economy development, focusing on four dimensions: technology, industry, application, and openness [1] - In Q1 of this year, Beijing's digital economy added value grew by 8.3% year-on-year, with core industries increasing by 9.7%, indicating a strong start to the year [1] - The city boasts significant advantages in digital industries, particularly in artificial intelligence, where it holds 43% of the nation's top talent and has 14 national key laboratories in the AI field [1] Group 2 - The Beijing International Big Data Exchange added 152 high-quality data products from January to May, with a trading scale exceeding 44 million yuan, reflecting a year-on-year growth of 37.63% [2] - Beijing's telecommunications infrastructure has completed IPv6 upgrades across all networks, supporting the deployment of gigabit optical networks and 5G, with all major commercial websites now supporting IPv6 access [2] - The city has launched the "Moli AI Lecture Hall" service brand, creating a comprehensive service system for enterprise development, and has seen 132 large models approved for online use, accounting for one-third of the national total [2]

Group 1 - Multiple AI chat assistants have disabled image recognition and "photo question answering" features during the college entrance examination period to ensure fairness [1] - Specific apps like Tencent's Yuanbao, Alibaba's Tongyi, Quark, and Doubao have implemented these restrictions, with messages indicating that these features are temporarily unavailable during the exam period [1][4] - The adjustments were made uniformly across major models in response to the examination schedule [1] Group 2 - The Guangdong Provincial Public Security Department has introduced 26 service measures to ensure security during the college entrance examination, including the establishment of a joint command for exam security [6] - Local police have formed specialized teams to combat exam-related crimes, conducting 24-hour monitoring and collaborating with various departments to address online sales of cheating materials and the use of AI in cheating [6]

Core Viewpoint - The article discusses the rise of AI-generated news and its implications, highlighting the blurred lines between reality and fiction in the media landscape [3][9]. Group 1: AI-Generated News Phenomenon - Recent incidents of AI-generated fake news, such as the rumor about a top male star gambling away 1 billion yuan, illustrate the rapid spread of misinformation on social media platforms [5][11]. - The use of AI in news production is not new, with organizations like Xinhua News Agency having developed AI reporters years ago, but the current trend of AI generating news content poses new challenges [7][24]. - AI-generated fake news can create a feedback loop, where users generate content that AI then uses to produce further misinformation, leading to a cycle of distorted information [8][41]. Group 2: Impact on Media and Journalism - The proliferation of AI-generated news has raised concerns among media professionals about the reliability of information, as many journalists now rely on AI tools for data collection and content generation [32][33]. - Instances of AI-generated content leading to public misinformation have been noted, such as false reports about significant events like earthquakes or celebrity deaths [18][23]. - The introduction of AI in journalism has led to a paradox where it is intended to save time but often results in increased workload due to the need for verification of AI-generated data [32][33]. Group 3: Regulatory and Ethical Considerations - There are ongoing discussions about the need for regulations to manage AI-generated content, including proposals for labeling AI-generated news to help users discern its authenticity [44]. - Recent measures by government bodies aim to enhance the reliability of AI-generated content, including the implementation of identification methods for synthetic content [44][48]. - The dual nature of AI as both a tool for efficiency and a potential source of misinformation underscores the need for careful management in its application within the media industry [48].

Core Viewpoint - The article discusses the challenges enterprises face in effectively utilizing AI tools for efficiency and revenue generation, highlighting the confusion over tool selection and the gap between business understanding and AI expertise [1] Group 1: AI Tools and Their Impact - Various AI tools such as DeepSeek, ChatGPT, Midjourney, and others are mentioned, indicating a proliferation of options in the market [1] - The article questions whether these tools have led to tangible improvements in enterprise performance [1] Group 2: Challenges in AI Implementation - There is a noted overlap in functionalities of AI tools, making it difficult for businesses to determine which tool to use in specific scenarios [1] - A disconnect exists where business professionals may lack AI knowledge, while AI experts may not understand business needs [1] Group 3: Upcoming Event - An upcoming live session featuring industry experts Zhang Lei and Wu Minghui will address how enterprises can effectively leverage AI [1] - Wu Minghui is highlighted as a knowledgeable figure with a background in AI and digital marketing, emphasizing the importance of expertise in navigating AI applications [1]

Core Viewpoint - The article discusses the increasing integration of AI tools in the job-seeking process, highlighting how these tools are transforming traditional methods of job application and interview preparation. It emphasizes that AI is becoming a crucial ally for job seekers, providing assistance in various stages such as career planning, resume optimization, and interview simulations [3][4][8]. Group 1: AI Tools in Job Seeking - AI tools are penetrating four main areas of the job-seeking process: career planning, resume modification, mock interviews, and interview assistance [4][5]. - A survey by Capterra indicates that 58% of job seekers are using AI tools to assist in their job search [3]. - The job market for AI tools is expected to peak in 2024, with a rise in specialized products targeting specific segments [4][5]. Group 2: Career Planning - There are few dedicated AI products for career planning, with many users still relying on general AI models like ChatGPT for guidance [9][12]. - Tools like COACH provide tailored career development plans for various user groups, including students and career changers [10][12]. - Users can input personal information into AI systems to receive tailored career advice, showcasing the potential of AI in providing comprehensive career planning [12][14]. Group 3: Resume Creation and Optimization - Approximately 45% of job seekers utilize AI tools for generating and improving their resumes [15]. - AI tools can diagnose and optimize resumes by identifying issues such as grammatical errors and content focus [15][19]. - Tools like Yoo Resume and Jobscan offer features for ATS optimization and job matching based on resume content [15][19]. Group 4: Mock Interviews - AI mock interview tools generate realistic interview questions based on the user's resume and target position, helping improve interview skills [20][22]. - Tools like Cedars.AI provide performance analysis reports post-mock interviews, highlighting areas for improvement [23][24]. - Some AI tools allow users to upload extensive documents to tailor the interview questions further, enhancing the simulation experience [25]. Group 5: Real-time Interview Assistance - Real-time interview assistance tools can transcribe questions and generate answers in under 0.8 seconds, facilitating smoother communication during interviews [26][27]. - These tools are marketed with phrases like "easily secure an offer," indicating their growing popularity among job seekers [27][31]. - The effectiveness of these tools is debated, with concerns about ethical implications and potential detection by employers [29][30]. Group 6: User Experiences and Recommendations - Users emphasize the importance of understanding oneself to effectively utilize AI tools for job searching [33][36]. - Combining multiple AI tools can provide diverse perspectives and enhance the job-seeking process [36]. - The article suggests that while AI can significantly aid in job searching, it should complement rather than replace personal effort and critical thinking [35][36].

又到了每个月的20日，是更新LPR的时间，2月份的LPR没有发生意外，一年期维持在3.1%，五年以上维持在3.6%，已经4个月没有变化了。 上一次调整还是在2024年10月，一年期和五年期分别下调了0.25%，就是因为那一次刺激政策导致部分一线城市连续4个月止跌，二三线城市虽然还在 跌，但是跌速放缓了。所以利率的下调对房价是有实质利好支撑，只是2024年房贷利率下调的太快，致使银行经营端压力骤增，短期内很难进一步降息。 …… 今天a股表现尚可，成交1.76万亿，中位数+0.51%，和昨天的加起来，差不多把前天跌掉的部分给涨回来了。 两市表现最好的行业竟然是橡胶制品，涨了6.23%，这在过去几年算是很稀罕了。通常橡胶制品是传统行业，很难蹭上热点风口，但是这次情况发生了诡 变，因为这次火起来的是ai女友，情趣机器人概念，市场预期这类产品一旦热销会带动对橡胶的需求，所以资金就都冲进去了。 以前这个赛道卖的都是充气的，原因无它，便宜罢了。我没用过啊，但凭脑子想这玩意充气的肯定没有橡胶的体验好，而且一旦ai赋能了机器人，是不是 仿真度和互动体验都会好的多，所以今天涨的除了橡胶板块，还有生理传感技术概念股。 deepse ...

孙不熟专栏，在这里读懂城市、学会买房。 以下文章来源于城市战争 ，作者孙不熟 城市战争 . 为何都不在广东创业？ 来源 | 城市战争 （ID：sunbushu123） 最近在网上看多一个说法：中国"AI三杰"都是广东人，但都没有在广东创业和发展。 公开资料显示，这三位蜚声全球的AI大牛都很年轻，其中两个是8 0后、一个是9 0后，一 个在杭州上大学和创业，一个在北京上大学和创业，另一个在美国MIT任教。 作者 | 孙不熟 梁文锋：广东湛江人 以一己之力干崩美股 De e ps e e k是春节期间火遍全球的一款国产大模型，成功地将大模型的训练成本断崖式降 低，被称之为"AI界的拼多多"，全球为之震惊，甚至有媒体称他"以一己之力干崩了美 股"。 梁文锋是De e p s e e k的创始人，1985年出生于广东湛江吴川的偏远农村，1 7岁就以优异 成绩考上浙江大学信息与电子工程系，并在浙大拿下硕士学位。 毕业后的梁文锋没有选择像很多同学那样进大厂当码农，而是选择自主创业，从事当时 并不被看好的全自动量化交易，就是使用数学模型和计算机方法来指导股票交易。 网友口中的"AI三杰"指的是De e ps e e k的创 ...