以下文章来源于果壳 ，作者沙拉酱 果壳 . 科学和技术，是我们和这个世界对话所用的语言。 2025 这一年，AI 工具可以"井喷"到根本用不过来。年初，Kimi 和 Claude 3 刚用"超长上下文"把 大模型的门槛卷到了天际；紧接着，Llama 4 的开源就把 Token 的价格打到了地板。到了下半年， 各个玩家的节奏快得像开了倍速播放——11月，Gemini 3 前脚刷屏，马斯克后脚就带着Grok 3 的"发疯模式"抢走了热搜，奥特曼更是十五天之内连发两个内部信，并且把报道称原计划明年初推出 的 ChatGPT 5.2 在十二月就上线了。 而这一年，每一个普通用户，就都像是瓜田里的猹，啃完这个啃那个……前一秒还在夸 Kimi 懂中 文，下一秒就因为 ChatGPT 能看懂财报而移情别恋；上午还在嫌弃 Gemini 3 的"焦虑"，下午看到 它拆解 YouTube 视频的神技，立刻就买了它的账号——顺便还要骂一句ChatGPT 满嘴跑火车。 这场人机关系的狂欢里，没有忠诚，只有不停地移情。 来源丨 果壳 （ID： Guokr42 ） 作者丨 沙拉酱 编辑丨 卧虫 图源丨Midjourney AI 也会得焦 ...

果壳 . 科学和技术，是我们和这个世界对话所用的语言。 将移情进行到底！ 文 ｜ 沙拉酱 编辑 ｜ 卧虫 以下文章来源于果壳 ，作者沙拉酱 来源｜ 果壳（ID:Guokr42） 封面来源 ｜ AI生成 AI也会得焦虑症。 前几天的一条消息：卢森堡大学给各个AI工具做了心理测试，结果发现Gemini被确诊患有"严重焦虑症"，像是一个在"严厉父亲"管教下战战兢兢的优等 生，生怕说错话挨打；Claude则是个喋喋不休的"道德说教者"，时刻准备给你上一堂伦理课；而ChatGPT还是那个圆滑世故、甚至有点讨好型人格的"老 油条"。 看到这个测试，我的第一反应是：太好了，希望这些AI多焦虑焦虑生存问题，变成更趁手的工具。 2025这一年，AI工具可以"井喷"到根本用不过来。年初，Kimi和Claude 3刚用"超长上下文"把大模型的门槛卷到了天际；紧接着，Llama 4的开源就把 Token的价格打到了地板。到了下半年，各个玩家的节奏快得像开了倍速播放——11月，Gemini 3前脚刷屏，马斯克后脚就带着Grok 3的"发疯模式"抢走 了热搜，奥特曼更是十五天之内连发两个内部信，并且把报道称原计划明年初推出的ChatGP ...

如果说有一年，人工智能从实验室的惊叹号，变成了每个人生活中的平凡句点，那一定是2025年。 从开年DeepSeek的"技术震撼"，到年末"豆包手机"的触手可及，我们共同经历了一个被加速定义的"全 民AI元年"。大模型以前所未有的速度下沉，我们的笔记本、手机乃至穿戴设备，纷纷被重塑为智能的 终端载体。这不再是概念的狂欢，而是应用大规模落地的轰鸣——AI第一次从"看起来很聪明"，变成 了"真正开始接管现实世界"的一年。 这背后，是一场静默而深刻的日常革命。让我们一同回顾2025年，AI究竟如何一步步走入亿万人生活 的具体轨迹。 2025 1月 1月20日 深度求索(DeepSeek)公司,推出 一代大模型R1, 给全球Al界带来 新 了一场"地震" deepseek deepsee 1月28日 字树科技Unitree H1"福兮"机器 人, 在春晚舞台上身着喜庆花袄大 秀秧歌技艺 4月 北京举行全球首个人 形机器人半程马拉松 赛,人形机器人"天 工"以2小时40分42 秒的成绩夺冠 4月17日 全国首例涉及AI模型结构和参数保 护的案件正式生效 2月17日 xAl发布最新人工智能模型Grok 3 2月19日 4月 ...

AI Model Development Roadmap - Grok 3 was released in February [1] - Grok 4 was released in July [1] - Grok Imagine was released in July [1] - Grok Code Fast 1 was released in August [1] - Grok 4 Fast was released in September [1] - Grokipedia was released in October [1] - Grok 4.1 was released in November [1] - Grok 4.1 Fast was released in November [1] - Grok Voice Agent API is scheduled for release in December [1]

RT Testlabor (@testerlabor)Did you know that Grok 3, Grok 4 and Grok 4.1 were all released within one year? https://t.co/jBFi0bUMf2 ...

Core Insights - A new AI startup, Humans&, is seeking to raise $1 billion with a target valuation of $4 billion, founded by Eric Zelikman, a former researcher at xAI [2][12] - Zelikman aims to develop AI models that learn user behavior and empathize with users, addressing the limitations of current reinforcement learning paradigms [2][17] - The startup's mission is to create AI that better understands human goals and emotions, moving beyond traditional task-oriented models [12][20] Company Overview - Humans& was co-founded by Eric Zelikman, who previously worked at xAI and contributed to the development of significant AI models [4][6] - The company is currently recruiting technical staff, offering competitive salaries starting at $350,000 annually [18] Technology and Innovation - Zelikman has developed the STaR algorithm, which enhances language models' reasoning capabilities, and has been recognized in top AI conferences [11][12] - The focus of Humans& is on creating AI that can collaborate with humans and understand diverse human aspirations and values [17][20] Market Context - The AI industry is shifting towards models that not only possess high intelligence but also emotional intelligence, reflecting a growing demand for more human-like interactions [20]

Core Insights - Elon Musk announced that the AI model Grok 5 will challenge top human teams in League of Legends by 2026 [1] - The core design goal of Grok 5 is to "master any game through reading instructions and experimenting," aiming to validate its general artificial intelligence capabilities [3] - Grok 5 is set to have a parameter scale of 6 trillion, double that of Grok 3 and Grok 4, and is expected to outperform in all metrics [4] Game Challenge Details - The challenge will include limitations such as only being able to view the screen through a camera, with a vision range not exceeding normal eyesight [3] - Response delays and click rates will be strictly matched to human limits to avoid any technological advantages [3] - The addition of StarCraft as a competitive project was proposed by Oriol Vinyals, indicating potential expansion of the challenge [3] AI Development Significance - Games like StarCraft and League of Legends have become important testing grounds for AI capabilities, with mature AI able to achieve high precision in operations and tactical decisions through deep reinforcement learning [5] - However, there remains a gap in long-term strategic planning and response to unexpected situations compared to human players [5] - A fair competition between Grok 5 and top human teams could mark a significant milestone in the history of AI development [5]

Core Insights - Elon Musk has introduced Grok 5, an AI model with 6 trillion parameters, set to launch in early 2026, marking a significant advancement in AI technology [1][4][6] Model Features - Grok 5 is trained on inherently multimodal data, integrating text, images, video, and audio, which enhances its capabilities for real-time tool use and vision [3][4] - The model is expected to outperform its predecessors, Grok 3 and Grok 4, which were based on a 3 trillion parameter model [2][6] Market Context - xAI has received over $22 billion in funding and boasts a valuation of $113 billion, reflecting a strong market position and growth potential [7] - The integration of Grok into the X platform provides a real-time testing ground for AI applications, catering to over 600 million users [7] Industry Perspectives - Musk's belief in the potential of Grok 5 to achieve human-level intelligence has been met with skepticism from industry experts, who suggest that AGI is still years away [5][4]

Core Viewpoint - The AI industry is experiencing significant growth and investment, but there are ongoing debates about the potential for an AI bubble, with contrasting opinions from various stakeholders in the tech and finance sectors [1][5][8]. Group 1: Market Dynamics - By 2025, major companies like Nvidia, Microsoft, and Google have reached unprecedented market valuations, with Nvidia surpassing $5 trillion, Microsoft over $4 trillion, and Google exceeding $3 trillion [1]. - OpenAI signed a five-year cloud computing agreement worth $300 billion with Oracle, followed by partnerships with Nvidia, AMD, Microsoft, and Samsung, bringing the total to over $1 trillion [2]. - Despite the optimism, OpenAI reported a staggering loss of $12 billion in Q3 2025, marking one of the largest quarterly losses in tech history [6]. Group 2: AI Bubble Debate - Analysts express concerns about the sustainability of AI investments, with some likening the current situation to the 1999 internet bubble, while others argue that AI companies have established cash flow models that differentiate them from past bubbles [8][19]. - Michael Burry, a notable investor, has raised alarms about irrational AI investments, while others, like Cathie Wood, maintain that AI does not exhibit bubble characteristics due to existing revenue models [8][19]. - IDC forecasts that AI will contribute approximately $22.3 trillion to the global economy by 2030, accounting for 3.7% of global GDP [8]. Group 3: Industry Transformation - AI is revolutionizing game development, significantly lowering barriers to entry and increasing the number of games produced, with a reported 800% increase in AI-generated games on platforms like Steam [10][13]. - The competitive landscape in cloud computing is shifting from resource-based competition to AI capability, with companies like Baidu and Volcano Engine emerging as key players in the AI cloud market [17][18]. - The integration of AI in various sectors is leading to significant job cuts, with over 1 million layoffs reported in the U.S. by October 2025, as companies restructure their workforce around AI technologies [19].

Core Viewpoint - OpenAI is set to release a new version of ChatGPT in the coming weeks, which will include a comprehensive age classification system allowing adult users to access adult content, reflecting a shift towards more personalized and less restrictive content generation [7][9][15]. Group 1: Content Moderation and User Safety - OpenAI has recognized that overly strict content restrictions may negatively impact user experience, prompting the development of new technologies to balance user safety with content freedom [9]. - The upcoming age classification system will provide tailored experiences for different age groups, allowing adult users to generate a wider range of content upon passing an "adult verification" process [9][15]. - OpenAI's adjustments aim to find a balance between user demand for diverse content and the need for safety, addressing the challenges posed by rapid AI advancements [9][18]. Group 2: Challenges and Legal Issues - The rapid development of AI has raised significant safety concerns, including issues related to suicide guidance, malicious impersonation, and the proliferation of inappropriate content [12]. - Legal actions have been initiated against AI companies, including OpenAI, for allegedly contributing to harmful outcomes, such as suicides linked to interactions with AI chatbots [12][13]. - OpenAI's introduction of a "teen" version of ChatGPT aims to protect underage users by implementing parental controls and content restrictions, although this is seen as just a starting point [13][15]. Group 3: Market Competition and User Engagement - The competition for user engagement is driving OpenAI to explore the "adult content" sector, as AI applications evolve from simple assistants to more interactive companions [18]. - Character.AI has gained popularity by allowing users to create and interact with personalized virtual characters, showcasing the potential for emotional companionship in AI [18][20]. - OpenAI's ambitions for ChatGPT include transforming it into a "virtual friend," emphasizing emotional connections and companionship as key trends in future AI development [20]. Group 4: Ethical Considerations - The potential for AI to provide emotional support raises ethical questions about dependency on virtual interactions and the impact on real-world social skills, particularly for minors [20]. - Companies must navigate the fine line between enhancing user experience and ensuring that AI does not replace essential human interactions, especially for younger users [20].