内容来源：https://www.wired.com/story/google-launches-gemini-3-ai-bubble-search/

内容总结：

【科技讯息】谷歌正式推出迄今最智能AI模型Gemini 3，强调其不仅是先进的对话工具，更是推动现有产品升级的核心引擎。在人工智能领域投资过热引发泡沫担忧之际，谷歌高管表示公司已通过将AI深度整合至搜索、地图等成熟产品中构建了"防波堤"。

谷歌DeepMind首席执行官德米斯·哈萨比斯在接受专访时坦言，当前AI初创企业估值存在虚高现象，但谷歌凭借其完整技术栈与多元化产品矩阵，即便面临行业调整也能保持稳定。值得注意的是，谷歌正通过三项关键数据证明AI战略成效：自然语言搜索查询量实现两位数同比增长，视觉搜索使用率激增70%，而新推出的"AI概览"功能更推动核心搜索业务查询量提升10%。

技术层面，Gemini 3在LMArena等权威评测中超越OpenAI的GPT-5，展现出更强的多模态理解与复杂问题分解能力。该模型已开始赋能创新工具开发，如能将文稿自动转化为播客的NotebookLM，并即将通过订阅服务向全球6.5亿月活用户开放。

尽管行业对通用人工智能(AGI)的探索仍需要5-10年技术积累，但谷歌正通过13万开发者生态与20亿用户产生的训练数据持续迭代模型。与此同时，公司据悉即将与苹果达成协议，为iPhone的Siri助手提供AI支持，这或将成为移动端AI竞争的重要转折点。

（综合自谷歌发布会及高管专访信息）

中文翻译：

谷歌正式推出迄今最智能的人工智能模型Gemini 3，该模型具备顶尖的推理、多模态交互与编程能力。随着人工智能泡沫论甚嚣尘上，这家科技巨头特别强调，此次发布不仅是推出智能模型和聊天机器人，更是从即日起全面提升谷歌现有产品矩阵的重要举措，包括其利润丰厚的搜索引擎业务。

"我们就像谷歌的引擎舱，正在将人工智能注入每个角落，"谷歌母公司Alphabet旗下专注人工智能的子公司DeepMind首席执行官德米斯·哈萨比斯在发布会前接受《连线》专访时表示。他坦言当前人工智能市场确实存在虚高现象，众多未经市场验证的初创企业获得数十亿美元估值。谷歌与其他人工智能公司亦投入数百亿美元建设新数据中心用于训练模型，这引发了市场对行业崩盘的担忧。

但哈萨比斯认为即使人工智能泡沫破裂，谷歌仍能安然无恙。该公司已运用人工智能增强谷歌地图、Gmail和搜索引擎等产品。"若市场下行，我们将更依赖现有产品矩阵；若市场向好，我们拥有最广泛的产品布局和最前沿的科研实力，"哈萨比斯说道。

谷歌正利用人工智能开发诸如NotebookLM（可将文字内容自动生成播客）和AI Studio（支持人工智能应用原型设计）等热门新工具。该公司甚至探索将技术嵌入游戏和机器人领域，哈萨比斯表示无论整体市场如何变化，这些布局都将在未来数年带来巨大回报。

即日起，用户可通过Gemini应用和"AI概览"功能使用Gemini 3，后者是谷歌搜索中整合常规搜索结果进行信息合成的特色功能。演示显示，当用户查询物理学三体问题等信息时，Gemini 3能即时生成定制化的交互式可视化解析。

谷歌搜索产品副总裁罗比·斯坦在发布会前的简报会上透露，采用自然语言表述的搜索查询量同比增长达到"两位数"，这类查询主要针对AI概览功能。同时，依托Gemini图片分析能力的视觉搜索量激增70%。

尽管在人工智能领域投入巨资并取得关键突破（包括发明支撑大多数大语言模型的Transformer架构），谷歌仍在2022年被ChatGPT的横空出世所震动。这款聊天机器人不仅将OpenAI推至人工智能研究领域的中心舞台，更通过提供全新且更便捷的网络搜索方式，对谷歌核心业务构成挑战。

随着谷歌逐渐缩小与OpenAI的差距，关于人工智能将迅速取代传统搜索的担忧正在消退。据彭博社报道，谷歌即将与苹果达成协议，为后者的虚拟助手Siri提供Gemini技术支持。图像生成编辑工具Nano Banana据传已获得用户热烈反响。最关键的是，生成式人工智能尚未侵蚀谷歌利润丰厚的搜索业务——Alphabet在七月季度财报中披露，AI概览功能推动搜索查询量增长10%。

而OpenAI八月发布的最新前沿模型GPT-5则令人略显失望。有专家评价其表现平庸，用户也抱怨其人格化设定趋于刻板。

谷歌宣称Gemini 3在LMArena等主流模型评测平台的多个关键指标上超越GPT-5及其他模型。该模型在将复杂问题分解处理的模拟推理，以及长周期规划任务中表现更优，这将提升使用工具和网络的人工智能代理的功能水平。

"这是我们最智能的模型，"DeepMind首席技术官科拉伊·卡武克乔格鲁在预发布简报中表示，"在多模态理解领域，这是全球最卓越的模型。"他补充说，谷歌庞大的用户基数正在助力模型优化：Gemini应用月活用户达6.5亿，1300万开发者使用谷歌模型，每月20亿人使用AI概览功能。用户与聊天机器人或人工智能应用的交互数据可作为训练素材，例如当模型需要增强特定领域专业知识时。卡武克乔格鲁还指出，谷歌自主研发芯片和运营数据中心的能力构成独特优势："我们采取差异化的全栈技术路径。"

谷歌表示Gemini 3将在未来数周向Google AI Plus（月费19.99美元）和Google AI Pro（月费249.89美元）订阅用户开放。同期推出的还有基于Gemini 3开发的新型AI编程工具Antigravity。

无论泡沫是否存在，哈萨比斯坚信Gemini 3将成为构建更强大人工智能的未来平台。"我认为距离真正的通用人工智能仍需五到十年，"他坦言，"这可能需要在此前不断优化的模型基础上，再实现一至两项重大突破。"

英文来源：

Google has introduced Gemini 3, its smartest artificial intelligence model to date, with cutting-edge reasoning, multimedia, and coding skills. As talk of an AI bubble grows, the company is keen to stress that its latest release is more than just a clever model and chatbot—it’s a way of improving Google’s existing products, including its lucrative search business, starting today.
“We are the engine room of Google, and we're plugging in AI everywhere now,” Demis Hassabis, CEO of Google DeepMind, an AI-focused subsidiary of Google’s parent company, Alphabet, told WIRED in an interview ahead of the announcement.
Hassabis admits that the AI market appears inflated, with a number of unproven startups receiving multibillion-dollar valuations. Google and other AI firms are also investing billions in building out new data centers to train and run AI models, sparking fears of a potential crash.
But even if the AI bubble bursts, Hassabis thinks Google is insulated. The company is already using AI to enhance products like Google Maps, Gmail, and Search. “In the downside scenario, we will lean more on that,” Hassabis says. “In the upside scenario, I think we've got the broadest portfolio and the most pioneering research.”
Google is also using AI to build popular new tools like NotebookLM, which can auto-generate podcasts from written materials, and AI Studio which can prototype applications with AI. It’s even exploring embedding the technology into areas like gaming and robotics, which Hassabis says could pay huge dividends in years to come, regardless of what happens in the wider market.
Google is making Gemini 3 available today through the Gemini app and in AI Overviews, a Google Search feature that synthesizes information alongside regular search results. In demos, the company showed that some Google queries, like a request for information about the three-body problem in physics, will prompt Gemini 3 to automatically generate a custom interactive visualization on the fly.
Robby Stein, vice president of product for Google Search, said at a briefing ahead of the launch that the company has seen “double-digit” increases in queries phrased in natural language, which are most likely targeted at AI Overviews, year over year. The company has also seen a 70 percent spike in visual search, which relies on Gemini’s ability to analyze photos.
Despite investing heavily in AI and making key breakthroughs, including inventing the transformer model that powers most large language models, Google was shaken by the sudden rise of ChatGPT in 2022. The chatbot not only vaulted OpenAI to center stage when it came to AI research; it also challenged Google’s core business by offering a new and potentially easier way to search the web.
Fears that AI could quickly supplant regular search appear to be fading as Google gains on OpenAI. The company is nearing a deal with Apple to use Gemini for the iPhone maker’s virtual assistant Siri, according to Bloomberg. Nano Banana, a capable AI tool for generating and editing images, has reportedly been a hit with users. Most importantly, generative AI does not yet seem to be eating Google’s lucrative search business. Alphabet said in its quarterly earnings this July that AI Overviews had driven a 10 percent increase in search queries.
Meanwhile, OpenAI’s latest frontier model, GPT-5, was a bit of a disappointment when it arrived in August. Some pundits called it underwhelming, and users complained about the shift to a more formal persona.
Google says Gemini 3 outperforms GPT-5 and other models on several key leaderboards, including LMArena, a popular site that lets users score models. The company says the model is better at performing simulated reasoning that involves breaking problems into parts and at planning over longer periods, which can improve the functionality of agents that use tools and the web.
“This is our most intelligent model,” Koray Kavukcuoglu, CTO of Google DeepMind, said during the prelaunch briefing. “It is the best model in the world for multimodal understanding.”
Kavukcuoglu added that Google’s huge user base is helping the company improve its models. The Gemini app has 650 million monthly users, there are 13 million developers working with Google’s models, and 2 billion people use AI Overviews each month. As users interact with a chatbot or an AI app, their responses can be used as training data—showing, for instance, when a model needs to improve expertise in a particular area. Kavukcuoglu adds that Google’s ability to design silicon and operate data centers also give it an edge. “We have a very differentiated full-stack approach,” he said.
Google says that Gemini 3 will be rolled out to Google AI Plus and Google AI Pro subscribers, who pay $19.99 and $249.899 per month, respectively, in the coming weeks. The company is also launching a new AI programming tool called Antigravity that is powered by Gemini 3.
Bubble or not, Hassabis says that Gemini 3 will be a platform for building more capable AI in the future. “I still think we are five to 10 years away from what I would call proper full AGI,” he says. “And that may require one or two breakthroughs on top of the models that are just getting better and better.”