深网腾讯新闻 . 以下文章来源于深网腾讯新闻 ，作者胡世鑫 腾讯新闻出品栏目，关注科技和TMT领域公司、事件和人物中的故事，探究背后的深层逻辑。 来源丨 深网腾讯新闻 （ID： qqshenwang ） 作者丨 胡世鑫 编辑丨 叶锦言 图源丨Midjourney 山东考生李响，近日收到了上海某二本院校人工智能 (AI) 专业录取通知书，兴奋劲还没过，便把信息晒到某社交媒体平台。让他没想到的是，评论区的"过 来人"几乎清一色"唱衰"："双非"学这个专业真是"天坑"…… 《 2025 高考季数据报告》显示：今年，工学专业热度断层领先，人工智能 (AI) 专业首次进入热门专业 Top10 。考生对于该专业就业前景展现出较大期 待， AI 发展促使更多考生愿意学习、从事 AI 行业。 AI 教育赛道的快速扩张，与产业界对 AI 人才的狂热需求形成了鲜明呼应。近日， Meta CEO 扎克伯格又从 OpenAI 挖走数十名骨干，签约奖金高达 1 亿 美金，引发关注。 收到录取通知的喜悦还没散去，李响就被社交媒体泼了盆冷水，他在评论区看到的尽是"毕业即失业"的劝退留言。而首批 AI 专业毕业生正经历着冰火两重 天：有人手 ...

Group 1 - The core message emphasizes that China's supply chain is a remarkable achievement and that open-source AI from China acts as a catalyst for global progress, allowing various countries and industries to participate in the AI revolution [2][3] - Huang Renxun highlighted that over 1.5 million developers in China are utilizing NVIDIA's platform to turn innovations into reality, with world-class models like DeepSeek, Tencent Mix Yuan, MiniMax, and Baidu Wenxin Yiyan driving rapid global AI development [3] - AI is described as a foundational infrastructure akin to electricity and the internet, reshaping supply chains and transforming production and logistics methods [3] Group 2 - AI is set to become central to every industry, enterprise, product, and service, sparking a new industrial revolution and presenting new growth opportunities for China's exceptional supply chain ecosystem [4] - NVIDIA aims to collaborate with long-term partners and new acquaintances to create a prosperous future in the AI era [4]

Core Viewpoint - Manus, once a popular AI Agent product, has cleared its official social media accounts, leading to speculation about its potential exit from the market. This follows recent layoffs and a strategic shift to relocate core technical staff to Singapore, indicating a possible connection between these events [2][3]. Group 1: Company Actions and Strategic Decisions - Manus has laid off staff in its domestic operations and moved key technical personnel to Singapore, citing operational efficiency as the reason for these adjustments [2]. - The company received $75 million in Series B funding led by Benchmark, with a valuation of $500 million, but faced scrutiny from the U.S. Treasury due to investment restrictions related to AI technology in China [2]. - The decision to clear social media accounts aligns with the company's relocation to Singapore, which was reportedly a requirement from Benchmark [3]. Group 2: Market Position and Competition - Manus initially gained attention through a registration system that made its product hard to access, leading to high demand for invitations. However, as other AI products matured, Manus's visibility diminished [4][5]. - The company missed the opportunity to capitalize on its initial popularity and failed to establish a strong user base compared to competitors like deepseek, which successfully engaged a wide audience [5]. Group 3: Capital and Investment Trends - Manus's approach has focused heavily on capital, with its founder having a background in mobile internet and previous ventures supported by significant investment [6]. - The global venture capital landscape is experiencing a slowdown in the number of deals, while larger transactions are becoming more common, indicating a trend towards concentration in the market [7][8]. Group 4: Importance of the Chinese Market - The Chinese AI market is seen as a critical growth area, with a mature user base and significant investment from major tech companies. Manus's move away from China raises questions about its future success [9][10]. - Statistics indicate that by June 2025, China's AI core industry is expected to exceed 1.2 trillion yuan, with an application penetration rate of 38%, highlighting the market's potential [10]. Group 5: Lessons for the Industry - Manus's decline serves as a cautionary tale for other AI players, emphasizing the need to focus on product development and market engagement rather than solely relying on capital [11]. - Companies in the AI sector should prioritize deepening their presence in the Chinese market to leverage its growth potential and avoid the pitfalls experienced by Manus [11].

Core Insights - The AI sector is showing positive momentum, with the CSI Artificial Intelligence Theme Index rising by 0.3% and the Shanghai Stock Exchange Sci-Tech Innovation Board AI Index increasing by 0.2% [1] - Six major AI language models, including DeepSeek and Baidu's Wenxin Yiyan, scored no less than 50 out of 60 on a national college entrance examination essay, demonstrating their strong capabilities in language understanding and creation [1] - CITIC Securities indicates that the tech sector has recently rebounded from its lows and remains in a high cost-performance range, with improving risk appetite leading to significant gains in overseas markets, particularly in the tech sector, which will reflect on A-shares [1] Investment Opportunities - The AI industry chain is highlighted as a key area for investment, focusing on upstream computing power autonomy and downstream application innovation [1] - The Artificial Intelligence ETF (159819) and the Sci-Tech Innovation AI ETF (588730) cover the entire AI industry chain, providing convenient tools for investors to capitalize on industry development opportunities [1] - The latest scale of the Artificial Intelligence ETF (159819) exceeds 16 billion yuan, making it the largest among similar products [1] Index Composition - The Sci-Tech Innovation AI Index tracks 30 large-cap stocks involved in AI foundational resources, technology, and application support, with over 85% of its composition in the electronics and computer sectors [4] - The index is designed to focus on computing power and application segments, reflecting the growth potential in these areas [4]

作者 | 第一财经 刘晓洁、郑栩彤 一年一度高考来临，考生之后，按惯例是各家AI的答题时间。这一年AI模型圈加速迭代进化，让AI写作文，水平会有提高吗？让名师来打分，各大模型的成绩排名有多大变化？ "相较去年，AI进步很快，不再机械单一，变得有思想了。"在看完几家模型的作文后，四川南充市嘉陵一中语文教师李东林对第一财经表示。 2024年，第一财经写了一期九大模型"决战"高考，当时李东林老师在看完AI作文后，认为它们"缺一点情感和灵气"，到今年，他觉得AI已经补上了这方面，但仍达不到细腻。 "人写作，可以刻意抒情和感性，即使是思辨类题目，也需要去契合某些人的感受。"李东林认为，这是AI仍然缺少的部分。 湖南省中学语文特级教师左建国有同样的看法，他觉得Al作文整体有提升，但提升幅度不大。因为有关时代、社会与生活方面的新素材并不多，明显储存不够，尤其是在抒发个人情感方面， 仍然是短板，缺少生命的温度。 左建国老师曾担任高考作文阅卷组的副组长，在高考阅卷方面有十几年的经验。"Al作文已经把考场作文变成一种可计算的拼图，形式上固然能逼近完美，但个性化的思考，以及拨动人心的 语言，几乎没有看到，这是AI与真人思维的真正差 ...

Group 1 - The core viewpoint of the articles is that AI writing models have shown significant improvement over the past year, becoming more sophisticated and capable of expressing ideas, although they still lack emotional depth and personal touch compared to human writing [1][2][3] - The ranking of AI models has changed, with Google's Gemini and DeepSeek emerging as top performers, while previous leaders like Yuanbao and Tongyi Qianwen have dropped in rankings [2][3] - The evaluation of AI-written essays indicates that while models can perform well on structured topics, they struggle with more abstract or nuanced prompts, highlighting the limitations of AI in understanding complex human emotions [5][6] Group 2 - The performance of various AI models in the essay evaluation shows a range of scores, with Gemini and DeepSeek achieving average scores of 61.5, while models like ChatGPT and Zhiyu performed poorly, indicating a disparity in capabilities [3][12][14] - The feedback from educators emphasizes that while AI can generate coherent essays, they often lack the depth and emotional resonance that human writers can provide, which is crucial for achieving high scores in subjective evaluations [2][5][20] - The assessment process involved a specific high school essay prompt, which was designed to test the models' ability to engage with contemporary cultural and philosophical themes, further illustrating the challenges AI faces in nuanced writing tasks [6][20]

Core Viewpoint - The article discusses a simulated exam for high school essays using six prominent domestic AI writing models, highlighting their capabilities and limitations in generating essays that reflect the depth of thought and language expression expected in high-stakes assessments [2][4][24]. Group 1: AI Models and Evaluation - Six AI models participated in the simulated exam: Tencent Yuanbao, Deepseek, Baidu Wenxin Yiyan, Quark, ByteDance Doubao, and iFlytek Xinghuo [2][4]. - The evaluation was conducted by three literary critics and involved anonymous peer reviews among the AI models, creating a multidimensional assessment matrix with a maximum score of 60 [4][23]. Group 2: Essay Themes and Content - The essays generated by the AI models explored themes of national spirit, personal struggle, and cultural resilience, drawing on historical and literary references [6][18]. - Each essay presented a unique perspective, with some focusing on the challenges of expression in difficult times, while others emphasized the connection between individual fate and national identity [10][18][24]. Group 3: Performance and Feedback - The AI models demonstrated strong language fluency and emotional depth, particularly in their use of cultural references and narrative techniques [24][25]. - However, there were noted deficiencies in the substantive content and depth of thought, with critiques pointing to a tendency for superficiality and a lack of coherent argumentation [24][25].

Group 1 - The article discusses a simulated essay examination conducted by six domestic AI models in response to the Chinese college entrance examination essay prompt, showcasing the writing capabilities of these AI systems [2][3][28] - The essay prompt encourages deep thinking and expression, reflecting the cultural and historical context of the Chinese nation, and serves as a test of language expression and creativity [3][5] - The evaluation of the essays was conducted by a panel of literary critics and through anonymous peer reviews among the AI models, ensuring a comprehensive assessment of their writing abilities [4][27] Group 2 - The essays generated by the AI models exhibit a range of themes, including cultural heritage, national spirit, and personal struggle, reflecting the historical significance of the materials used [6][10][18] - The critiques from literary experts highlight the strengths and weaknesses of each AI-generated essay, noting aspects such as thematic focus, logical coherence, and emotional depth [8][12][19][28] - Overall, the AI models demonstrated proficiency in language fluency and emotional resonance, but there remains room for improvement in substantive content and critical thinking [28]

Core Insights - AI agents are redefining the search landscape, shifting from "machines adapting to humans" to "humans adapting to machines" [2] - The emergence of AI large models has disrupted the traditional search market, previously dominated by giants like Google and Baidu [2][3] - The competition in the search industry is evolving towards "mind share" rather than just "traffic share" [3][4] Industry Dynamics - The competition is characterized by intensified technological stratification, with leading firms leveraging computational power and data advantages to create barriers [3] - Ecosystem competition is heating up, as traditional search engines integrate their content ecosystems to maintain dominance [3] - The value of entry points is being redefined, with AI search evolving from a tool to a traffic entry point [3] User Behavior Changes - Google search volume in Safari has declined for the first time in 20 years, while user numbers for startups like Perplexity are surging [4] - Apple's plans to integrate third-party AI search in Safari indicate a potential erosion of traditional search engine dominance [4] Business Model Evolution - The traditional advertising model in the search industry is being challenged, with subscription and enterprise services emerging as new directions [5] - Search results are evolving from "webpage lists" to "knowledge products," enhancing the added value of AI search [5] Challenges Ahead - There are significant challenges, including technical bottlenecks, ethical risks, and the need for ecosystem collaboration [6] - AI hallucination issues persist, particularly in handling ambiguous semantics [6] - Data privacy and algorithmic bias are critical ethical concerns as AI agents become more proactive in user environments [6]

Group 1 - The report by Zhong An Technology and Zhong An Financial Technology Research Institute explores the application of large language models (LLMs) in the financial and insurance industries, concluding that LLMs present new opportunities but face challenges in implementation that require multi-party collaboration [1] - The development of large model technology is diversifying globally, with vertical models emerging to provide tailored industry solutions. China has made progress in computing autonomy and data optimization, leading to a trend of functional differentiation and specialization in its ecosystem [1][24] - New technologies are driving down the costs of training, operation, and inference for large models, prompting a restructuring of processes in the financial industry. Financial enterprises need to balance acquisition, inference, and operational costs while selecting appropriate deployment models and roles [1][12] Group 2 - Domestic models like DeepSeek and Tongyi Qianwen have achieved breakthroughs in cost control and inference performance, providing better technical options for insurance institutions while ensuring data security and compliance [1][15] - Insurance institutions are accelerating the integration of large models, focusing on internal efficiency improvements across the entire insurance business chain and back-office management. Caution is advised during pilot applications to address data security and AI hallucination issues [1][16] - The value of data elements is becoming more prominent, with the financial and insurance industries building high-quality datasets through horizontal, vertical, and government-enterprise collaboration mechanisms to promote intelligent transformation [1][19] Group 3 - The application of large language models in the financial and insurance sectors is transitioning from pilot exploration to systematic integration, with initial deployments focusing on low-risk, low-intervention auxiliary business scenarios such as intelligent customer service and smart claims [6][7] - The introduction of large language models is not only enhancing process efficiency but also driving a deep transformation in information processing paradigms and decision-making logic within the industry [8][9] - The rise of large language models is reshaping the operational philosophies, business logic, and value creation models of financial institutions, leading to trends such as precision financial services and cross-industry ecological collaboration [9][10] Group 4 - The evolution of large model technology is characterized by a shift from purely algorithmic breakthroughs to the construction of systemic capabilities that integrate model deployment, business processes, and system interfaces [29][30] - The deployment capabilities of large models are transitioning from "usable" to "adaptable," with future competition likely focusing on building flexible deployment mechanisms across architectures and scenarios [31] - The emergence of vertical large models is addressing the specific needs of industries like finance and healthcare, enhancing precision and efficiency in tasks such as risk assessment and compliance checks [40][41]