OpenAI启动深入研究：高级研究的AI代理

本周，Openai推出了一个名为Deep Research的新AI代理。就像它发布的第一个AI代理一样，Deep Research是一种AI代理，能够根据初始用户指令自动执行复杂的任务。

OpenAI Deep Research: An AI agent for advanced research

Openai深入研究：高级研究的AI代理

OpenAI unveiled its latest AI agent, dubbed Deep Research, designed to autonomously execute complex end-to-end research tasks based on user instructions. Unlike Operator, which was geared towards short and simple assignments, Deep Research is optimized for in-depth, extensive workflows, akin to a researcher conducting advanced investigations in various fields. The AI agent can take anywhere from 5 to 30 minutes to return results, depending on the task's complexity, مشابه to the extended duration required for a human to complete involved research.

Openai推出了其最新的AI代理，称为深入研究，旨在根据用户说明自主执行复杂的端到端研究任务。与经营者不同，该操作员针对简单和简单的作业而言，深入研究对深入，广泛的工作流程进行了优化，类似于研究人员在各个领域进行高级研究。 AI代理可以花费5到30分钟才能返回结果，具体取决于任务的复杂性，即人类完成涉及研究所需的延长持续时间。

“Deep research is built for people who do intensive knowledge work in areas like finance, science, policy, and engineering and need thorough, precise, and reliable research. It can be equally useful for discerning shoppers looking for hyper-personalized recommendations on purchases that typically require careful research, like cars, appliances, and furniture,” the announcement reads.

“深入研究是为在金融，科学，政策和工程等领域进行大量知识工作的人们建立的，需要彻底，精确和可靠的研究。这对于寻找通常需要仔细研究（例如汽车，电器和家具）的购买的购物者而言，这同样有用。”公告写道。

In OpenAI’s demonstration, Deep Research was showcased conducting comprehensive market research for a product and identifying the best skis for a user based on their specifications.

在Openai的演示中，展示了深入的研究，为产品进行全面的市场研究，并根据用户的规格确定最佳滑雪板。

Deep Research is currently available exclusively to GPT Pro users, who pay $200 per month for OpenAI's premium tier service. However, both Operator and Deep Research will be rolled out to all tiers of OpenAI subscribers in the future, according to the company’s statement.

目前，GPT Pro用户专门提供深入的研究，他们每月为OpenAI的高级层服务支付200美元。但是，根据该公司的声明，将来将向OpenAI订户的所有层次进行操作员和深入研究。

I see the release of Deep Research as a direct response to the recent buzz surrounding DeepSeek's entry into the AI scene. When DeepSeek made headlines last week, it sparked discussions about whether U.S.-based AI companies, like OpenAI, could justify their sky-high valuations. OpenAI’s move seems like a counterattack against the negative sentiment that Deep Seek’s entry cast over U.S. tech companies, attempting to shift the narrative back in their favor and reminding the world why they remain a dominant force in AI.

我认为Deep Research的发布是对DeepSeek进入AI场景的最新嗡嗡声的直接反应。当DeepSeek上周成为头条新闻时，它引发了关于美国AI公司（例如Openai）是否可以证明其高价估值的讨论。 Openai的举动似乎是反对Deep Seek对美国科技公司的负面情绪的反击，试图将叙事转移回他们的支持，并提醒世界为什么他们仍然是AI中的主导力量。

DeepSeek’s rapid rise and fall: From AI disruptor to global security threat

DeepSeek的迅速崛起和下降：从AI颠覆者到全球安全威胁

Last week, DeepSeek took the AI world by storm, making headlines for allegedly matching OpenAI's performance while spending just $6 million to develop its generative AI model, compared to the multi-billions spent by OpenAI. The news sent ripples through the tech industry and raised concerns about the valuations and competitiveness of U.S.-based AI firms.

上周，DeepSeek席卷了AI世界，与OpenAI花费的数十亿美元相比，据称与Openai的表现相匹配，而据称与Openai的表现相匹配，这使其成为头条新闻。这一消息使科技行业发动了涟漪，并引起了人们对美国AI公司的估值和竞争力的担忧。

But this week? DeepSeek has nearly vanished from the conversation. The AI has been banned from government devices in South Korea, Australia, and Taiwan over national security concerns. Microsoft (NASDAQ: MSFT) is looking into whether DeepSeek used OpenAI research and data without authorization, and corporations and government agencies are worried that user data entered into DeepSeek is being funneled directly to China, posing a security risk.

但是这个星期呢？ DeepSeek几乎从谈话中消失了。在国家安全问题上，AI已被韩国，澳大利亚和台湾的政府设备禁止使用。 Microsoft（NASDAQ：MSFT）正在研究DeepSeek是否在未经授权的情况下使用了OpenAI研究和数据，公司和政府机构担心将输入DeepSeek的用户数据直接向中国汇入，带来了安全风险。

All of these factors, combined with skepticism over whether DeepSeek was truly able to create its model for the amount of money and computing infrastructure that they claimed, set the stage for the Chinese competitor to be publicly disparaged and almost wiped from our collective memory within a span of just one week.

所有这些因素，加上对DeepSeek是否真正能够为他们声称的金钱和计算基础设施创建其模型的怀疑，为中国竞争对手公开贬低并几乎从我们的集体记忆中削弱了舞台跨度仅一周。

We reproduced DeepSeek R1-Zero in the CountDown game, and it just works Through RL, the 3B base LM develops self-verification and search abilities all on its own You can experience the Ahah moment yourself for < $30 Code: https://t.co/B2IsN1PrXVHere's what we learned ? pic.twitter.com/43BVYMmS8X

我们在Countdown游戏中重现了DeepSeek R1-Zero，它仅通过RL起作用，3B基础LM可以自己体验自我验证和搜索能力。 .co/b2isn1prxvhere我们学到了?

On the other hand, an AI research team at UC Berkeley managed to replicate DeepSeek’s core capabilities for just $30. While this lends some credibility to DeepSeek’s claim of achieving a low-cost AI breakthrough, it also aligns with OpenAI CEO Sam Altman's statement that being a copycat is easier than being a pioneer, especially in a field like AI where the foundational work has already been done by companies like OpenAI, paving the way for later entrants.

另一方面，加州大学伯克利分校的AI研究团队设法以30美元的价格复制了DeepSeek的核心能力。尽管这对DeepSeek的主张获得了低成本AI的突破，但它也与Openai首席执行官Sam Altman的说法保持一致，即成为模仿者比成为先驱者更容易由Openai等公司完成，为后来的进入者铺平了道路。

Regardless, I don’t think we will be hearing much more about DeepSeek in the United States. The U.S. has a vested interest in ensuring that U.S.-based AI companies remain the dominant forces and are likely playing a role in suppressing foreign inventions like DeepSeek. I expect that to continue, whether it be through direct intervention, trade restrictions, or negative PR. It’s clear that foreign AI firms—especially those from China—face significant roadblocks to global expansion, set in place by the United States.

无论如何，我认为我们不会听到更多关于美国DeepSeek的消息。美国对确保总部位于美国的AI公司仍然是主要力量的既得利益，并且很可能在压制诸如DeepSeek之类的外国发明中发挥作用。我希望无论是通过直接干预，贸易限制还是负PR，都将继续下去。很明显，外国人工智能公司（尤其是来自中国的公司）对全球扩张的巨大障碍，这是美国实现的。

Google lifts ban on AI weapons

Google举起禁令AI武器

Google (NASDAQ: GOOGL) has quietly revised its AI ethics policy, removing a key clause that previously prohibited the company from pursuing AI applications for weapons and surveillance. This change represents a significant shift in Google’s stance on AI development.

Google（NASDAQ：GOOGL）已悄悄修改了其AI道德政策，删除了一项重要条款，该条款以前禁止该公司追求AI申请武器和监视。这种变化代表了Google对AI开发的立场的重大转变。

Following the update, Google released a blog post to justify its decision, “Since we first published our AI Principles in *}2018, the technology has evolved rapidly… AI has moved from a niche research topic in the lab to

更新后，Google发布了一篇博客文章以证明其决定是合理的：“自从我们首次发布了我们在 *} 2018中的AI原理以来，该技术已经迅速发展了……AI已从实验室的利基研究主题转移到