达里奥·阿莫迪不到30秒就提到了AI制造生物武器的风险。这位Anthropic首席执行官是该公司周二发布的IPO路演视频中第一个出现的人。
It takes Dario Amodei just over 30 seconds to mention the risk of AI-created bioweapons in his IPO pitch. The Anthropic CEO is the first face you see in the company's IPO roadshow video it released on Tuesday.
他始终围绕主题，称AI是“人类历史上最有前景的技术”，能够“捕捉认知的本质”。他强调，安全始终是这家AI实验室的首要考量。
He stays on message, calling AI "the most promising technology in human history" that is "capturing the essence of cognition." He emphasizes that safety is top of mind for the AI lab.
这段视频将Anthropic的价值观和IPO推介浓缩成一份42分钟的“鸡尾酒”，希望能说服华尔街投资其公开发行。每一个字都至关重要——因此我们进行了统计。
The video is Anthropic's values and IPO pitch bottled up together in a 42-minute cocktail it hopes will convince Wall Street to invest in its public offering. Every word counts — so we counted.
《商业内幕》分别要求ChatGPT和Claude分析视频的逐字稿，以找出最常见的词汇和主题。
Business Insider asked ChatGPT and Claude to separately analyze a transcript of the video to identify the most prevalent words and themes.
视频本身由Anthropic研究中常见的鲜艳图形，以及包括百时美施贵宝和埃森哲在内的顶级客户的一系列证言组成。
The video itself is a mixture of brightly colored graphics that have become a staple of Anthropic's research and a series of testimonials from top customers, including Bristol Myers Squibb and Accenture.
一旦排除对“Claude”和“模型”的预期提及，算力便是主导。OpenAI首席执行官山姆·阿尔特曼喜欢说算力即命运。
Compute is king, once you exclude the expected mentions of "Claude" and "models." OpenAI CEO Sam Altman likes to say compute is destiny.
算力可以说是Anthropic上市的原因，因为为数据中心及其中的芯片融资成本高昂。
Compute is arguably the reason Anthropic is going public, as financing data centers and the chips that fill them is expensive.
根据其修订版S-1文件，Anthropic计划在算力和基础设施上花费895亿美元，其中近一半预计将在2030年底前投入。Anthropic的安全使命是这家AI初创公司的基石。
According to its amended S1, Anthropic is set to spend $895 billion on compute and infrastructure, nearly half of which is expected to be spent by the end of 2030. Anthropic's safety mission is the cornerstone of the AI startup.
27次提及还不包括阿莫迪提到的AI可能助长生物武器创造的风险等内容。如果按主题分析，安全会更加突出（尽管仍不及算力）。Anthropic迅速解释说，它不认为安全是障碍。“我们发现安全并没有阻碍我们，”路演视频的旁白解释道。
The 27 mentions also don't factor in things like Amodei's mention of the risk of AI enabling the creation of bioweapons. If you analyze by theme, safety becomes more prevalent (though it still doesn't trump compute). Anthropic is quick to explain that it doesn't view safety as an obstacle. "We're finding safety doesn't hold us back," the roadshow video narrator explains.
“事实上，正是因为我们可以信任Claude的工作，我们才能将其发现应用于现实世界，造福所有人。”
"In fact, it's because we can trust Claude's work that we can take its discoveries and apply them in the real world in ways that will benefit everyone."