本文来自微信公众号： 腾讯科技 ，作者：晓静，编辑：徐青阳，原文标题：《技术指数级发展，可 怕的是全世界竟无察觉｜Anthropic CEO最新访谈》，题图来自：视觉中国 "我90%确信，2035年前人类将迎来'数据中心里的天才国度'——甚至可能就在一两年内。" Anthropic CEO达里奥·阿莫代伊 （Dario Amodei） 说出这句话时，语气平静得像在预言明天的天 气。 但真正让他抓狂的不是技术进展太快，而是全世界竟然毫无察觉。在接受美国知名博客主持人德瓦克 什·帕特尔 （Dwarkesh Patel） 近150分钟的深度专访中，阿莫代伊反复强调一个观点： 我们离AGI 的终点比任何人想象的都要近，而公众还在讨论那些老掉牙的政治话题。 帕特尔：现在"规模扩展"的假设到底是什么？预训练的扩展定律大家都懂，但强化学习扩展好像没有 公开的规律可循。 阿莫代伊：我现在的假设跟2017年写《大计算块假说》 （The Big Blob of Compute Hypothesis） 时是一样的，与图灵奖得主、强化学习之父里奇·萨顿 （Rich Sutton） 的《苦涩的教训》 （The Bitter Lesso ...

Core Viewpoint - The CEO of Anthropic, Dario Amodei, predicts that by 2026-2027, AI will evolve into a "Country of Geniuses in a Datacenter," with intelligence comparable to thousands of Nobel laureates working together [2][8][9] - Anthropic is experiencing a staggering annual revenue growth of 10 times, expecting to reach $10 billion by 2025, driven by advancements in AI capabilities [2][11] Group 1: AI Growth and Predictions - Amodei asserts that AI is nearing the end of its exponential growth phase, with significant qualitative changes expected in the next 2-3 years [5][6] - The transition from "smart high school student" to "professional-level" AI models has been rapid, with improvements in programming and mathematical capabilities [6][8] - Amodei expresses high confidence in achieving the vision of a genius AI nation within the next decade, citing a 90% certainty for a 10-year timeline and a 50/50 chance for the next 1-2 years [9][42] Group 2: Revenue Growth and Financial Strategy - Anthropic's revenue trajectory is described as "bizarre 10x per year growth," with projections of $1 million in 2023, $10 million in 2024, and $9-10 billion in 2025 [11][12] - Amodei explains the cautious approach to capital investment in computing power, emphasizing the need for revenue growth to align with capacity expansion to avoid bankruptcy risks [13][14] Group 3: AI in Software Engineering - Amodei outlines three stages of AI evolution in software engineering, with the first stage already achieved where models write 90% of code lines [16][50] - The second stage will see models handling 90% of end-to-end tasks, while the third stage will involve models taking over complex engineering tasks [18][53] - The expectation is that AI will significantly enhance productivity in software engineering without leading to mass unemployment among engineers [20][54] Group 4: Challenges and Future Developments - Amodei acknowledges potential geopolitical risks and societal upheavals as variables that could impact the timeline for achieving advanced AI capabilities [9][13] - The company is actively researching continuous learning capabilities for AI, which may be realized in the next couple of years [108][109] - There is an ongoing discussion about the efficiency of AI in learning and adapting compared to human learning processes, with a focus on the need for models to achieve a level of contextual understanding [100][101]

Core Insights - Anthropic recently completed a $30 billion Series G funding round, achieving a valuation of $380 billion, marking the second-largest single funding round in venture capital history, with an annual revenue of $14 billion [2] - The CEO of Anthropic, Dario Amodei, predicts that the AI industry will likely reach a trillion-dollar revenue level by 2030, driven by technological and diffusion indices [3][17] - Amodei's aggressive forecast suggests that within 1 to 3 years, AI systems will reach or exceed the capabilities of Nobel Prize winners in various fields [5] Company Strategy and Growth - Anthropic's revenue is projected to grow approximately tenfold each year, from nearly zero to $1 billion in 2023, $10 billion in 2024, and around $90-100 billion in 2025, with significant increases already noted in January 2025 [14][48] - The company has adopted an aggressive yet calculated investment strategy in computing resources, emphasizing the importance of early procurement to avoid potential bankruptcy due to demand forecasting errors [15] - The internal perception at Anthropic indicates that AI tools have significantly enhanced productivity, contributing to an overall acceleration of 15-20% in operations [12] Industry Dynamics and Predictions - The AI industry's competitive landscape is expected to resemble that of cloud computing, characterized by a few dominant players and high entry barriers, ensuring that profits will not be driven to zero [16] - Amodei believes that while AI diffusion into the economy is rapid, it will not happen instantaneously due to factors like corporate procurement processes and compliance reviews [13] - The anticipated "genius nation in data centers" is expected to emerge within 1 to 3 years, fundamentally transforming various professional fields [8][41] Technological Insights - The scaling laws for pre-training and reinforcement learning (RL) remain effective, supporting the hypothesis that large computational blocks are essential for AI development [9] - Continuous learning is not deemed necessary for models, as pre-training and RL generalization, combined with longer context windows, are likely sufficient for performance [10] - The spectrum of coding capabilities ranges from AI writing 90% of code to potentially replacing software engineering entirely, though full replacement is still some distance away [11] Safety and Ethical Considerations - Amodei advocates for transparency in AI safety standards, suggesting that regulations should evolve as risks are validated, rather than imposing blanket bans [21][22] - The potential for AI to dissolve authoritarian structures is viewed optimistically, akin to the early expectations surrounding social media [23] - The importance of building data centers in developing countries is emphasized to ensure they do not fall behind in the AI-driven economy [24] Cultural and Operational Insights - Maintaining company culture is a priority for Anthropic, with regular all-hands meetings and open communication to foster cohesion among employees [27] - Decision-making speed is highlighted as critical, with the potential for significant historical decisions to be made in brief moments [28]

Core Insights - Anthropic's CEO Dario Amodei predicts that by 2026, AI will evolve into a "Country of Geniuses in a Datacenter," where AI systems will exhibit intelligence comparable to thousands of top minds working together [3][107] - The company is experiencing an extraordinary revenue growth trajectory, with expectations of reaching $10 billion in 2024 and $90-100 billion in 2025, marking a bizarre 10x annual growth rate [4][42] - Amodei emphasizes the importance of responsible investment in computational power, linking it to revenue growth and the accuracy of predictions to avoid catastrophic risks [6][8] Group 1: AI Growth and Evolution - Amodei asserts that AI is nearing the end of its exponential growth phase, transitioning from "smart high school students" to "PhD-level" capabilities, with significant advancements in programming and mathematics [2][10] - The rapid advancements in AI capabilities are not merely about increasing parameters but represent a fundamental upgrade in intelligence, moving from data fitting to autonomous generalization [2][4] Group 2: Revenue Projections - Anthropic's revenue is projected to grow from $0 to $100 million in 2023, from $100 million to $1 billion in 2024, and to $90-100 billion in 2025, indicating a remarkable growth curve [4][42] - The company has already added several billion dollars in revenue in the first month of 2023, reinforcing the expectation of continued rapid growth [4][44] Group 3: Financial Strategy - Amodei explains that the expansion of computational power must align with revenue growth and predictive accuracy to mitigate the risk of bankruptcy [6][8] - The current strategy is described as "responsibly aggressive," allowing for sufficient computational investment to capture significant upside while maintaining survival through high margins and cash flow [8] Group 4: AI in Software Engineering - Amodei outlines three stages of AI evolution in software engineering, predicting that within 1-3 years, AI will handle all responsibilities of senior software engineers, leading to a massive productivity boost [9][33] - The first stage has already been achieved, with models writing 90% of code lines, and the next stages will involve handling end-to-end tasks and understanding complex codebases [11][33] Group 5: Future Predictions and Challenges - Amodei expresses high confidence (90%) in achieving the vision of a "Country of Geniuses" within ten years, with a 50/50 chance of significant advancements occurring in the next 1-2 years [3][21] - Potential geopolitical risks, such as disruptions in the chip supply chain, are noted as the only significant uncertainties that could impact this timeline [3][21]