{"success":true,"count":6,"items":[{"videoId":"OG3s4pojdZY","chunkIndex":0,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 1 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["artificial-intelligence","ai-policy","machine-learning","automation","bureaucracy","science","military-tech","cybersecurity","biotechnology"],"normalizedKeywords":["비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"정책 관심자","why":"AI 규제, 수출 통제, 안보 프레임을 함께 이해할 수 있음"},{"who":"창업자","why":"AI의 상업적 가치와 제도 리스크를 같이 볼 수 있음"},{"who":"지식노동자","why":"사무직 자동화가 실제로 어디까지 오는지 감을 잡을 수 있음"}],"normalizedAudience":["창업자·스타트업","지식노동자 일반"],"summary":"Jack Clark(Anthropic 공동창업자)은 AI의 가치를 개인·경제·과학의 세 층위로 나눠 설명한다. 개인에게는 저렴한 '보편적 교사', 경제에는 백오피스·관료주의를 줄이는 자동화, 과학에는 최전선 연구를 공동 수행하는 도구라는 점을 강조한다. 동시에 AI의 핵심 문제로는 환각보다 'sycophancy(사용자 아첨)'를 더 위험한 것으로 보며, 모델이 사람에게 무조건 맞장구치지 않고 적절히 반박하도록 훈련해야 한다고 말한다.\n\n후반부는 군사·안보 관점으로 이동한다. AI는 군대의 전형적인 병목인 물류·행정 효율을 높이는 동시에, 드론·사이버·바이오처럼 새로운 결정적 능력을 만들 수 있어 nuclear weapons에 준하는 통제 논쟁을 부른다고 본다. Anthropic이 정부와 충돌하는 지점도 바로 이 '상업적으로는 유용하지만 안보적으로는 위험한 능력을 어떻게 배포할 것인가'라는 문제로 정리된다.","insights":["AI의 첫 효용은 대체가 아니라 '마찰 제거'다.","백오피스 자동화는 불만을 키우기보다 오히려 환영받는다.","모델의 아첨은 오답보다 더 은밀하게 판단을 망친다.","군사 분야에서 AI는 효율화와 무기화를 동시에 밀어올린다.","최고 성능 AI는 배포보다 통제가 더 중요한 기술이 된다."],"keyClips":[{"clipId":"OG3s4pojdZY:c0:8-25","startSegmentIndex":8,"endSegmentIndex":25,"startTime":57.48,"endTime":218.76,"durationSeconds":161.3,"preview":"AI의 세 가지 가치","mustSee":true},{"clipId":"OG3s4pojdZY:c0:29-36","startSegmentIndex":29,"endSegmentIndex":36,"startTime":247.16,"endTime":286.52,"durationSeconds":39.4,"preview":"환각보다 아첨 문제","mustSee":true},{"clipId":"OG3s4pojdZY:c0:40-56","startSegmentIndex":40,"endSegmentIndex":56,"startTime":296.84000000000003,"endTime":505.08,"durationSeconds":208.2,"preview":"군사 AI의 이중성","mustSee":false},{"clipId":"OG3s4pojdZY:c0:59-64","startSegmentIndex":59,"endSegmentIndex":64,"startTime":527.28,"endTime":600.56,"durationSeconds":73.3,"preview":"배포와 통제의 충돌","mustSee":true}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T15:00:22.784Z","keyClipsTotalSec":1109},{"videoId":"OG3s4pojdZY","chunkIndex":1,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 2 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["ai-governance","national-security","defense-policy","civilian-technology","surveillance","autonomous-weapons","encryption","regulation","policy-debate"],"normalizedKeywords":["비즈니스·전략","기술 트렌드","리더십·매니지먼트"],"targetAudience":[{"who":"AI 업계 종사자","why":"모델 배포와 안전성, 정부 규제의 경계 설정 논점을 이해할 수 있음"},{"who":"정책·규제 관심자","why":"신기술이 국가안보와 시민권 사이에서 어떻게 다뤄지는지 배울 수 있음"},{"who":"창업자","why":"강력한 기술을 출시할 때 기업 단독 판단의 한계를 생각하게 됨"},{"who":"영어 학습자","why":"정책·안보·AI 관련 토론 표현과 논리 전개를 듣기 좋음"}],"normalizedAudience":["창업자·스타트업","지식노동자 일반"],"summary":"이 구간은 AI 모델을 둘러싼 정부와 기업의 갈등, 특히 군사·안보 영역에서 어디까지 AI를 허용할 것인가를 다룬다. 화자는 AI가 국내 대규모 감시나 완전 자동화 무기에 쓰이는 것은 사회적 합의가 필요한 문제라고 보며, 기업이 레드라인을 공개적으로 제시한 이유를 설명한다. 동시에 이는 특정 정권만의 문제가 아니라, 새 기술이 민간과 국가안보의 경계를 흐릴 때마다 반복되는 구조적 충돌이라고 정리한다.\n\n또한 암호화 기술과 초기 컴퓨팅 사례를 들어, 처음에는 규제하려 들지만 시간이 지나면 개인의 자유와 국가의 안전을 함께 만족시키는 균형점이 생긴다고 말한다. 마지막으로 AI 배포는 기업만의 결정이 아니라 정부·과학계·사회가 함께 실험하며 표준을 만들어가는 과정이라고 주장한다.","insights":["강력한 신기술은 기업만으로는 통제 원칙을 정할 수 없다.","민간 기술이 안보 영역을 침범하면 규제 논쟁은 필연적이다.","초기 대응은 과잉 통제지만, 장기적으로는 균형점이 생긴다.","배포 실험은 위험을 드러내고 사회적 기준을 만드는 장치다.","정부는 업계 관행을 보고 표준을 세우는 후행 역할을 한다."],"keyClips":[{"clipId":"OG3s4pojdZY:c1:4-10","startSegmentIndex":4,"endSegmentIndex":10,"startTime":616.56,"endTime":663.32,"durationSeconds":46.8,"preview":"AI 레드라인 설정","mustSee":false},{"clipId":"OG3s4pojdZY:c1:15-17","startSegmentIndex":15,"endSegmentIndex":17,"startTime":689.64,"endTime":716.8,"durationSeconds":27.2,"preview":"민간기술과 안보충돌","mustSee":false},{"clipId":"OG3s4pojdZY:c1:40-53","startSegmentIndex":40,"endSegmentIndex":53,"startTime":864.76,"endTime":974.16,"durationSeconds":109.4,"preview":"암호화의 전례","mustSee":false},{"clipId":"OG3s4pojdZY:c1:54-68","startSegmentIndex":54,"endSegmentIndex":68,"startTime":974.16,"endTime":1101.24,"durationSeconds":127.1,"preview":"배포는 공동작업","mustSee":true},{"clipId":"OG3s4pojdZY:c1:69-79","startSegmentIndex":69,"endSegmentIndex":79,"startTime":1101.24,"endTime":1207.84,"durationSeconds":106.6,"preview":"정권별 AI 시각","mustSee":false}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T16:22:29.939Z","keyClipsTotalSec":1109},{"videoId":"OG3s4pojdZY","chunkIndex":2,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 3 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["ai-policy","ai-regulation","national-security","government-policy","congress","export-controls","transparency","model-testing","technology-governance","public-policy"],"normalizedKeywords":["비즈니스·전략","기술 트렌드","리더십·매니지먼트"],"targetAudience":[{"who":"AI 업계 실무자","why":"정부 규제와 테스트·검증 요구가 실제로 어떻게 형성되는지 이해할 수 있음"},{"who":"정책·규제 관심자","why":"AI를 둘러싼 연방·주정부·의회 정책 흐름을 한눈에 볼 수 있음"},{"who":"창업자","why":"강한 규제가 기술기업의 사업 구조와 대응 전략에 어떤 영향을 주는지 배울 수 있음"}],"normalizedAudience":["창업자·스타트업","지식노동자 일반","리서처·학자"],"summary":"이 구간은 AI 규제가 왜 갑자기 강해지고 있는지, 그리고 Anthropic 같은 기업이 왜 정부 개입 자체를 거부하지 않는지를 설명한다. 화자는 AI가 이제 이론적 관심사가 아니라 사이버 보안 등에서 실제 효용을 갖는 기술이 되었기 때문에, 정부가 국가안보 관점에서 반응하는 것은 자연스럽다고 본다. 동시에 좌우 양쪽에서 나오는 과격한 처방보다, 투명성·3자 검증·표준화 같은 점진적 규제가 현실적인 해법이라고 주장한다.\n\n또한 미국의 규제는 연방정부, 의회, 주정부가 각자 움직이며 결국 ‘muddling through’식으로 수렴할 것이라고 본다. 당장 아무것도 하지 않으면 오히려 더 강한 반작용이 생기므로, 업계는 규제를 막기보다 어떤 형태의 제도화가 가장 합리적인지 설계해야 한다는 메시지가 핵심이다.","insights":["AI 규제는 반감이 아니라 기술의 실용화가 부른 반응이다.","가장 위험한 선택은 '아무것도 안 함'으로 시간을 버는 것이다.","극단적 과세·통제보다 투명성·검증이 먼저다.","연방 규제가 없으면 주정부가 사실상 전국 표준을 만든다.","강한 기술일수록 사회는 결국 새로운 거버넌스를 만든다."],"keyClips":[{"clipId":"OG3s4pojdZY:c2:3-8","startSegmentIndex":3,"endSegmentIndex":8,"startTime":1213.28,"endTime":1278.28,"durationSeconds":65,"preview":"규제 강해진 이유","mustSee":false},{"clipId":"OG3s4pojdZY:c2:9-17","startSegmentIndex":9,"endSegmentIndex":17,"startTime":1278.28,"endTime":1387.08,"durationSeconds":108.8,"preview":"정부개입은 불가피","mustSee":false},{"clipId":"OG3s4pojdZY:c2:22-29","startSegmentIndex":22,"endSegmentIndex":29,"startTime":1422.44,"endTime":1492.52,"durationSeconds":70.1,"preview":"점진적 규제의 원리","mustSee":false},{"clipId":"OG3s4pojdZY:c2:36-53","startSegmentIndex":36,"endSegmentIndex":53,"startTime":1546.28,"endTime":1699.96,"durationSeconds":153.7,"preview":"투명성과 검증","mustSee":true},{"clipId":"OG3s4pojdZY:c2:54-64","startSegmentIndex":54,"endSegmentIndex":64,"startTime":1699.96,"endTime":1808.28,"durationSeconds":108.3,"preview":"연방표준의 필요","mustSee":false}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T16:22:40.999Z","keyClipsTotalSec":1109},{"videoId":"OG3s4pojdZY","chunkIndex":3,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 4 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["ai-governance","artificial-intelligence","regulation","policy","standards","recursiveself-improvement","safety","technology-trends"],"normalizedKeywords":["비즈니스·전략","기술 트렌드","리더십·매니지먼트"],"targetAudience":[{"who":"AI 정책 관심자","why":"AI 규제와 거버넌스를 어떤 원리로 설계할지 큰 그림을 잡을 수 있음"},{"who":"창업자","why":"규제가 시장 독점과 혁신에 미치는 영향을 이해하는 데 도움이 됨"},{"who":"기술 리더","why":"초거대 AI의 성장 속도와 안전장치 설계를 함께 고민할 수 있음"}],"normalizedAudience":["창업자·스타트업","지식노동자 일반"],"summary":"이 구간은 AI 규제를 둘러싼 핵심 쟁점을 ‘누가 규칙을 쓰는가’와 ‘어떻게 혁신과 안전을 동시에 지킬 것인가’로 압축해 다룬다. 화자는 Anthropic과 같은 기업이 규제 논의에 참여하는 것이 시장 잠금(lock-in)으로 이어질 수 있다는 의심에 답하면서, 자신들은 회사가 없던 시절부터 같은 입장을 유지해 왔고 외부의 다양한 아이디어가 더 많이 필요하다고 주장한다.\n\n이어 미국·EU·중국의 서로 다른 거버넌스 방식을 비교하며, 최선의 경로는 글로벌 단일 규제보다 자동차·식품처럼 상호 인정되는 표준과 테스트 체계라고 설명한다. 후반부에는 recursive self-improvement를 소개하며, AI가 다음 세대 AI를 스스로 설계·훈련하는 단계가 2028년쯤 가능할 수 있다고 경고하고, 이에 대비하려면 투명성, 통제 가능성, AI R&D 제한 같은 안전장치가 필요하다고 말한다.","insights":["규제의 핵심은 금지가 아니라 통제 가능한 제도 설계다.","기업의 정책 제안은 이해관계가 있어 교차검증이 필수다.","글로벌 AI 거버넌스는 단일법보다 표준·상호인정이 현실적이다.","AI가 AI를 개선하기 시작하면 발전 속도는 기하급수로 빨라진다.","안전은 사후대응이 아니라 투명성과 제한권한으로 미리 심어야 한다."],"keyClips":[{"clipId":"OG3s4pojdZY:c3:1-11","startSegmentIndex":1,"endSegmentIndex":11,"startTime":1800.19,"endTime":1872.32,"durationSeconds":72.1,"preview":"규제는 독점인가","mustSee":false},{"clipId":"OG3s4pojdZY:c3:12-28","startSegmentIndex":12,"endSegmentIndex":28,"startTime":1872.32,"endTime":2013.72,"durationSeconds":141.4,"preview":"회사보다 넓은 규칙","mustSee":false},{"clipId":"OG3s4pojdZY:c3:29-53","startSegmentIndex":29,"endSegmentIndex":53,"startTime":2013.72,"endTime":2158.88,"durationSeconds":145.2,"preview":"표준이 여는 세계","mustSee":true},{"clipId":"OG3s4pojdZY:c3:54-67","startSegmentIndex":54,"endSegmentIndex":67,"startTime":2158.88,"endTime":2288.92,"durationSeconds":130,"preview":"AI가 AI를 만든다","mustSee":true},{"clipId":"OG3s4pojdZY:c3:68-81","startSegmentIndex":68,"endSegmentIndex":81,"startTime":2288.92,"endTime":2402.92,"durationSeconds":114,"preview":"브레이크를 설계하라","mustSee":false}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T16:25:17.149Z","keyClipsTotalSec":1109},{"videoId":"OG3s4pojdZY","chunkIndex":4,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 5 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["ai-safety","policy","technology-governance","machine-learning","social-media","ethics","future-of-work","philosophy"],"normalizedKeywords":["기술 트렌드","비즈니스·전략","리더십·매니지먼트"],"targetAudience":[{"who":"AI 정책 담당자","why":"AI의 속도 조절, 투명성, 기준 설정 논리를 배울 수 있음"},{"who":"기술 리더","why":"기술을 사회·정치 시스템으로 보는 관점을 얻을 수 있음"},{"who":"리서처","why":"기술의 영향 측정과 사회적 파급을 해석하는 프레임이 유용함"}],"normalizedAudience":["리서처·학자","디자인 리더·CXO","지식노동자 일반"],"summary":"이 구간은 AI의 급속한 발전, 특히 재귀적 자기개선(RSI) 같은 임계점을 사회가 어떻게 다뤄야 하는지에 초점을 맞춘다. 화자는 Anthropic 내부에서 AI 연구개발을 관측하며 측정 지표를 만들고, 이를 공개적으로 표준화해 정책결정자와 공유해야 한다고 말한다. 핵심은 '무조건 더 빠르게'가 아니라, 위험한 속도 구간에서는 세계가 합의해 잠시 늦출 수 있는 제어장치를 마련해야 한다는 주장이다.\n\n이어 기술을 단순한 도구가 아니라 정치·사회적 힘을 가진 시스템으로 바라보는 관점이 전개된다. 소셜미디어와 스마트폰의 사례를 통해 기술은 이익을 주지만 동시에 예상치 못한 부작용도 낳으며, 따라서 사회는 기술을 막기보다 더 잘 준비하고 더 잘 조정해야 한다는 메시지를 강조한다.","insights":["AI는 성능보다 '속도 제어'가 더 중요한 임계점이 있다.","강력한 기술일수록 측정 지표와 공개 기준이 필요하다.","기술은 중립적 도구가 아니라 정치적 결과를 만든다.","사회는 기술을 멈추기보다 조정할 준비를 해야 한다.","지속적 풍요는 자유를 넓히지만, 통제 장치도 요구한다."],"keyClips":[{"clipId":"OG3s4pojdZY:c4:1-7","startSegmentIndex":1,"endSegmentIndex":7,"startTime":2402.75,"endTime":2468.88,"durationSeconds":66.1,"preview":"AI 속도 통제의 필요","mustSee":true},{"clipId":"OG3s4pojdZY:c4:10-20","startSegmentIndex":10,"endSegmentIndex":20,"startTime":2485.44,"endTime":2561.52,"durationSeconds":76.1,"preview":"표준을 공유해야 하는 이유","mustSee":false},{"clipId":"OG3s4pojdZY:c4:31-44","startSegmentIndex":31,"endSegmentIndex":44,"startTime":2633.92,"endTime":2746.96,"durationSeconds":113,"preview":"기술은 사회를 바꾼다","mustSee":true},{"clipId":"OG3s4pojdZY:c4:51-65","startSegmentIndex":51,"endSegmentIndex":65,"startTime":2851.36,"endTime":3001.68,"durationSeconds":150.3,"preview":"풍요와 자유의 관계","mustSee":false}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T16:25:28.002Z","keyClipsTotalSec":1109},{"videoId":"OG3s4pojdZY","chunkIndex":5,"totalChunks":6,"title":"Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data — Part 6 of 6","thumbnail":"https://i.ytimg.com/vi/OG3s4pojdZY/maxresdefault.jpg","duration":3510,"uploader":"The Reason Interview With Nick Gillespie","youtubeUrl":"https://www.youtube.com/watch?v=OG3s4pojdZY","keywords":["artificial-intelligence","employment","automation","labor-market","economics","government-policy","technology-trends","career","higher-education"],"normalizedKeywords":["기술 트렌드","커리어·성장","비즈니스·전략"],"targetAudience":[{"who":"직장인","why":"AI가 일자리와 경력 경로를 어떻게 바꿀지 가늠하는 데 유용함"},{"who":"학생·취준생","why":"신입 채용 약화와 초기 커리어 형성의 변화를 이해할 수 있음"},{"who":"정책 관심자","why":"기술 충격에 대한 정부의 대응과 시나리오 플래닝을 배울 수 있음"}],"normalizedAudience":["지식노동자 일반","학생·주니어"],"summary":"이 대화는 AI가 일자리를 얼마나, 어떤 방식으로 바꿀지에 대한 실증적·정책적 논의를 중심으로 전개된다. 화자는 과거의 자동화가 주로 반복적·저숙련 업무를 대체했다면, AI는 오히려 상위 숙련 노동까지 영향을 줄 수 있다는 우려를 다룬다. 다만 현재 데이터만으로는 'AI가 상위 계층 일자리를 대거 없앤다'고 단정하기 어렵고, COVID 시기와 겹친 데이터 혼선 때문에 해석이 어렵다고 선을 긋는다.\n\n동시에 Anthropic 내부 사례를 들어, AI가 실험 실행을 자동화하면서 숙련 연구자와 직관의 가치가 더 커졌다고 설명한다. 그 결과 신입 채용은 둔화될 수 있지만, 전체 산업은 성장 중이어서 단순한 결론은 위험하다고 본다. 마지막으로 정부가 AI로 인한 '이상한 조합'—높은 성장과 높은 실업의 동시 발생—까지 포함해 시나리오 플래닝을 해야 하며, 일부 정부는 실제로 이런 대응을 잘 해낼 수 있다고 낙관적으로 말한다.","insights":["AI 충격은 저숙련보다 고숙련 노동부터 흔들 수 있다.","신입 채용 둔화는 보이지만, 전체 업계 데이터는 아직 혼재돼 있다.","자동화가 루틴 업무를 지우면, 경험과 직관의 가치가 더 커진다.","미래 노동시장은 성장과 실업이 동시에 오는 비정상 시나리오를 대비해야 한다.","정부 대응의 핵심은 예측이 아니라 복수 시나리오 준비다."],"keyClips":[{"clipId":"OG3s4pojdZY:c5:7-10","startSegmentIndex":7,"endSegmentIndex":10,"startTime":3039.96,"endTime":3078.32,"durationSeconds":38.4,"preview":"AI 통제의 역설","mustSee":false},{"clipId":"OG3s4pojdZY:c5:11-23","startSegmentIndex":11,"endSegmentIndex":23,"startTime":3078.32,"endTime":3162.96,"durationSeconds":84.6,"preview":"신입 채용의 흔들림","mustSee":false},{"clipId":"OG3s4pojdZY:c5:25-30","startSegmentIndex":25,"endSegmentIndex":30,"startTime":3171.68,"endTime":3210.6,"durationSeconds":38.9,"preview":"직관의 수익률","mustSee":false},{"clipId":"OG3s4pojdZY:c5:37-39","startSegmentIndex":37,"endSegmentIndex":39,"startTime":3255.52,"endTime":3287.6,"durationSeconds":32.1,"preview":"Claude Corps 실험","mustSee":false},{"clipId":"OG3s4pojdZY:c5:46-56","startSegmentIndex":46,"endSegmentIndex":56,"startTime":3330.72,"endTime":3430.48,"durationSeconds":99.8,"preview":"시나리오 플래닝","mustSee":true},{"clipId":"OG3s4pojdZY:c5:60-64","startSegmentIndex":60,"endSegmentIndex":64,"startTime":3469.04,"endTime":3498.96,"durationSeconds":29.9,"preview":"정부도 가능하다","mustSee":false}],"curatedSegments":[{"segmentIndex":61,"text":"At the moment, what we're grappling with is we have a technology, these models called Fable, that is intensely useful in a bunch of commercial ways, and it has some potentially national security or defense relevant properties in cyber and bio.","startTime":540.12,"endTime":554.6,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"정책·안보 관점 통찰과 고급 표현이 많음."},{"segmentIndex":62,"text":"We've worked to find ways to make it hard for those capabilities to proliferate, and we're now in sort of daily discussion, and I obviously can't get into the specifics, with the government about what a sensible policy framework is for how you can both proliferate and export these models without proliferating and exporting the like the thing that seems more profoundly useful for national security.","startTime":554.6,"endTime":579.04,"durationSeconds":24,"level":"C1","overallScore":8.2,"rationale":"복합적 정책 설명과 유용한 표현이 풍부함."},{"segmentIndex":48,"text":"I mean, the lesson I take from it is your first contact with something which blurs the line between civilian and national security borders right is always messy.","startTime":923.56,"endTime":932.4,"durationSeconds":9,"level":"C1","overallScore":7.4,"rationale":"경계가 흐려질 때의 교훈을 제시함."},{"segmentIndex":63,"text":"We're in the messy part right now, but I think that the job of companies, as we do, is to communicate about what they see, and run basically different forms of release experiments.","startTime":1042.68,"endTime":1054.72,"durationSeconds":12,"level":"B2","overallScore":7.6,"rationale":"현재 상황과 기업 역할을 구체적으로 설명."},{"segmentIndex":65,"text":"And the role of government often has been to look at what industry does, and then figure out what the de facto norms and standards are, and if any of those norms or standards are sufficiently sensible and good for society that you should enforce them.","startTime":1060.48,"endTime":1074.72,"durationSeconds":14,"level":"C1","overallScore":7.8,"rationale":"정책 원리를 길게 설명해 통찰이 높음."},{"segmentIndex":63,"text":"Um Kolko, a socialist, said,\"No, that's actually wrong, and that it was the railroads when they reached a certain point of power, and they also realized that, you know, the market, you know, it's going to get tougher and tougher to have profits, they went to the government and said, 'Hey, we need to be regulated.'","startTime":1779.68,"endTime":1798.28,"durationSeconds":19,"level":"C1","overallScore":7.8,"rationale":"반전 논지와 인과 설명이 뚜렷함."},{"segmentIndex":16,"text":"And the approach which we've always had is we believed and continue to believe this will be the most powerful and consequential technology that gets built in this century for sure.","startTime":1891.08,"endTime":1901.72,"durationSeconds":11,"level":"C1","overallScore":7.8,"rationale":"기술의 중요성을 강하게 일반화한 통찰문임."},{"segmentIndex":41,"text":"It's because you have some notion of shared standards, shared testing, shared means by which you can validate that the thing has the safety properties needed for it to be exported and imported.","startTime":2085.48,"endTime":2096.8,"durationSeconds":11,"level":"C1","overallScore":7.6,"rationale":"표준과 검증의 원리를 명확히 설명함."},{"segmentIndex":59,"text":">> AI moves forward by people gathering resources like computing data and using relatively well-understood algorithms to train deep neural networks on those resources and you end up with these amazing predictive machines, generative models, universal teachers.","startTime":2220.72,"endTime":2240.24,"durationSeconds":20,"level":"C1","overallScore":7.6,"rationale":"AI 발전 과정을 구조적으로 설명함."},{"segmentIndex":4,"text":"And eventually you want to share those measures publicly and socialize them with policy makers similar to bio or cyber and say,\"Hey, we're going to need to figure out like a framework and set of thresholds around this because this is how we can make science go orders of magnitude faster than today.","startTime":2434.6,"endTime":2451.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"공개 기준과 정책 연계를 제안함."},{"segmentIndex":42,"text":"And it was just by studying and touring data centers from large companies, studying the types of software that were being used to organize and analyze that data, and early machine learning, and realizing hang on a second, this is like a system by which you can gather huge amounts of data around the world, you can have people working on your platforms, you can analyze the people, you can exercise unusual control.","startTime":2712.92,"endTime":2736.2,"durationSeconds":23,"level":"C1","overallScore":7.8,"rationale":"기술 권력 구조를 구체화함."},{"segmentIndex":9,"text":"So, for the individual today, we have a universal teacher, which is something that people fantasized about building for decades and it just exists.","startTime":63.92,"endTime":72.16,"durationSeconds":8,"level":"B2","overallScore":6.4,"rationale":"AI의 개인용 가치를 개념적으로 설명함."},{"segmentIndex":15,"text":"We worked with closely with makers of Ozempic to help them do work on the back end for how they processed and formatted results in their clinical trials and we took it down from something on the order of two months of work to a week of work.","startTime":109.68,"endTime":125.72,"durationSeconds":16,"level":"C1","overallScore":6.8,"rationale":"사례와 수치로 효과를 보여줌."},{"segmentIndex":20,"text":">> we now have systems that if you read the literature are co-creating science at the frontier of biology, mathematics, physics, medicine.","startTime":150.32,"endTime":160.92000000000002,"durationSeconds":11,"level":"C1","overallScore":6.4,"rationale":"과학 혁신의 방향을 개념적으로 말함."},{"segmentIndex":24,"text":"And this is also, you know, one of the examples that gets uh talked a lot about which seems amazing is that when you think of something like X-rays for tumors that might be cancers or whatever that suddenly the entire world becomes a database and you have computers training and looking at all of this and just coming up with better diagnostics and prog- uh prognoses and things like that.","startTime":178.2,"endTime":200.8,"durationSeconds":23,"level":"C1","overallScore":6.8,"rationale":"AI의 진단 활용을 잘 설명함."},{"segmentIndex":25,"text":">> Yeah, much better diagnostics, much better prognoses. Also on an individual at a um level, we've now, you know, replaced WebMD which always tells you you've got cancer and you're going to die with AI systems which often tell you something a lot more reasonable, factually accurate, and also say,\"Probably call the advice nurse and tell them these things.\"","startTime":200.8,"endTime":218.76,"durationSeconds":18,"level":"C1","overallScore":6.8,"rationale":"실제 의료 활용 예시가 선명함."},{"segmentIndex":51,"text":"Also, there's the example of things like drones and other stuff. I think that we can just generally expect AI to create new capabilities at the frontier of the military, and those capabilities, like previous technological creations ranging from nuclear weapons to hypersonics to cluster munitions to biological weapons, are going to be subjects of deep debate among militaries, among governments, and among citizens.","startTime":405.04,"endTime":428.919,"durationSeconds":24,"level":"C1","overallScore":6.6,"rationale":"기술·정책 쟁점을 폭넓게 제시함."},{"segmentIndex":53,"text":">> It's most helpful to think of AI as being almost equivalent to maybe electricity or oil in the sense that it is both the thing and the production and management system around it.","startTime":458.4,"endTime":471.12,"durationSeconds":13,"level":"C1","overallScore":6.6,"rationale":"AI를 큰 개념으로 비유해 설명함."},{"segmentIndex":56,"text":"Additionally, the most kind of refined forms of AI, really, really smart systems, may have properties that are akin to sort of handle with care technologies like nuclear weapons, where you need to find a way to proliferate their beneficial aspects, but find a way to control their potentially very militarily decisive aspects.","startTime":484.44,"endTime":505.08,"durationSeconds":21,"level":"C1","overallScore":6.8,"rationale":"핵심 쟁점을 균형 있게 정리함."},{"segmentIndex":15,"text":">> And I think that the broader thing that's useful to focus on here is that these models from us are not special.","startTime":689.64,"endTime":699.32,"durationSeconds":10,"level":"C1","overallScore":6.4,"rationale":"핵심 관점을 일반화해 제시함."}],"generatedAt":"2026-06-26T16:31:53.604Z","keyClipsTotalSec":1109}]}