愤怒的本质:系统而非科学
许多人表示他们现在“讨厌AI”。我不断地在评论、对话和新闻标题中听到这种声音。我理解这种情绪,我并不是要告诉你这种感觉是错误的。但我想在这段视频中尝试一种不同的方式:我不会告诉你应该如何看待AI,而是将阐述一些区别,回顾一些背景,并向你提出一系列问题。到最后,我希望你能在评论中告诉我,你的愤怒究竟指向何处,因为我认为答案可能会比“我讨厌AI”这种简单的表述更具体、更有用。我是El,我拥有计算机科学博士学位,我分析AI的发展,以理解所有这些炒作背后真正发生的事情。
让我们从一个类比开始。没有人会说“我讨厌医学”。医学(Medicine: 一门科学分支,旨在通过预防和治疗疾病来维护和改善健康)作为一门学科,延长了人类的预期寿命,根除了天花,使分娩变得可存活,并让人们与家人拥有了原本不可能拥有的岁月。它是人类最伟大的成就之一。然而,如果你是一个美国人,曾经不幸骨折并收到4万美元的医院账单,或者不得不在配药和付房租之间做出选择,或者眼睁睁看着保险公司拒绝支付医生认为你需要的治疗费用,你可能会对医疗系统抱有非常强烈的情绪。你可能会对药品定价、保险公司的守门行为、制药公司为了利润推销药物感到愤怒。这些都是完全合理的情绪,但你并非在愤怒医学本身。你愤怒的是一套经济结构、企业激励机制和政策失误,它们将一门救死扶伤的科学变成了金融恐怖的来源。你讨厌的是系统(System: 组织或相互作用的部分组成的复杂整体),而不是科学。
Original English Source
A lot of people say they hate AI right now. I hear it constantly in comments, in conversations, in headlines. And I understand it. I'm not here to tell you that you're wrong. But I want to try something a little different with this video. Instead of telling you what to think about AI, I'm going to lay out some distinctions, walk through some context, and ask you a series of questions. By the end, I'd like you to tell me in the comments where your anger actually sits. because I think the answer might be more specific and more useful than just I hate AI. I'm El. I have a PhD in computer science and I analyze AI developments to understand what actually is happening beneath all of this hype. Let's start with an analogy. Nobody says I hate medicine. Medicine is a branch of science that has extended human life expectancy, eradicated smallpox, made child birth survivable, and given people years with their families they wouldn't otherwise have had. Medicine as a discipline is one of humanity's greatest achievements. And yet, if you're an American who has ever broken a bone, god forbid, and received a $40,000 hospital bill, or had to choose between filling a prescription and paying rent, or watched an insurance company deny a claim for a procedure your doctor said you needed. You might have some very strong feelings about the medical system. You might be furious about drug pricing, about insurance gatekeeping, about pharmaceutical companies pushing medications for profit. Those are completely valid feelings, but you're not angry at medicine. You're angry at a set of economic structures, corporate incentives, and policy failures that have turned a life-saving science into a source of financial terror. You hate the system, not the science.
AI的辉煌成就:科学的力量
我认为AI也正在发生非常相似的事情,我想在这段视频中仔细剖析它。人工智能(Artificial Intelligence: 计算机科学的一个分支,旨在使机器能够执行通常需要人类智能的任务)作为计算机科学的一个分支,自1950年代以来以各种形式存在。它所做的大部分工作与聊天机器人、生成的图像或大型语言模型(Large Language Models: 基于海量文本数据训练的深度学习模型,能理解和生成人类语言)毫无关系。
以Google DeepMind开发的AlphaFold为例,它解决了一个困扰生物学家50年的问题:根据氨基酸序列预测蛋白质的三维结构。过去,每种蛋白质的这项工作需要一年或更长时间的艰苦实验。而AlphaFold则几乎为所有已知蛋白质完成了这项任务。这对于药物发现具有巨大的意义。与AlphaFold相关的研究在临床文章中被引用的可能性是典型结构生物学工作的两倍,并且更有可能出现在专利中。它在2024年获得了诺贝尔化学奖。这就是AI,它与为你写电子邮件毫无关系。
在癌症影像领域,麻省理工学院的MRAI等AI工具可以提前长达五年预测乳腺癌风险,这使得临床医生有时间更早地进行干预。在多种癌症类型中,AI诊断系统在检测医学影像中的肿瘤方面,始终与临床医生的准确率持平甚至超越。AI正被用于为聋哑和听障社区提供实时语音转文本服务。它还被用于气候建模、材料科学、农业产量预测、地震预警系统以及数十个其他领域,在这些领域,“大型语言模型”这个词从未被提及。当有人说“我讨厌AI”时,我只想温和地问:“你讨厌蛋白质折叠研究吗?你讨厌早期癌症检测吗?你讨厌为残障人士提供的无障碍工具吗?”你可能不讨厌。那么,你具体讨厌的是什么?请记住这个问题,我稍后会回来讨论它。
Original English Source
I think something very similar is happening with AI, and I want to carefully pull it apart in this video. Artificial intelligence is a branch of computer science. It has been around in various forms since the 1950s and the vast majority of what it does has absolutely nothing to do with chatbots generated images or large language models. Consider Alphafold developed by Google DeepMind solved a problem that had stumped biologists for 50 years predicting the three-dimensional structure of proteins from their amino acid sequences. This used to take a year or more of painstaking experimental work per protein. Alphold did it for essentially every known protein. The implications for drug discovery are enormous. Research linked to AlphaFold is twice as likely to be cited in clinical articles as typical structural biology work and significantly more likely to appear in patents. It won the 2024 Nobel Prize in chemistry. This is AI. It has nothing to do with writing your emails for you. In cancer imaging, AI tools like MIT's MRAI can predict breast cancer risk up to five years in advance, giving clinicians time to intervene earlier. Across multiple cancer types, AI diagnostic systems have consistently matched or surpassed clinician accuracy in detecting tumors from medical images. AI is being used for realtime speech to text for deaf and heart of hearing communities. It's being used in climate modeling, material science, agricultural yield prediction, earthquake early warning systems, and dozens of other domains where the words large language models have never been uttered. When someone says, "I hate AI," I want to just gently ask, "Do you hate protein folding research? Do you hate early cancer detection? Do you hate accessibility tools for people with disabilities?" You probably don't. So, what is it specifically that you hate? Hold that question. I'm going to come back to it.
经济困境:AI威胁感源于社会环境
我认为要理解为什么AI现在会引发如此程度的恐惧和愤怒,你必须理解它所处的时代背景,因为AI并非凭空出现,它降临在我们当前的经济体系之中。
自1985年以来,美国家庭收入中位数(US Median Household Income: 衡量家庭经济状况的核心指标,将所有家庭按收入高低排序,取中间位置的收入值)按名义价值计算增长了约255%,但房屋价格中位数(Median House Prices: 一国或地区所有房屋价格排序后的中间价格)却飙升了415%以上。房价与实际收入之比从约3.1上升到接近历史最高水平的5.1。首次购房者的平均年龄从1980年代的29岁上升到今天的40岁。截至2026年,65%的美国家庭买不起新建的中等价位房屋。在某些州,这一数字甚至超过80%。
你不需要我告诉你这是一种什么样的感受,我们都在共同经历。人们的收入与生活成本之间的差距已经持续扩大了40年。对于大多数人来说,几乎没有留下任何余地。现在,想象你正处于这种境地——紧绷、岌岌可危,离真正的麻烦只差一个糟糕的月份。然后有人告诉你,一项强大的新技术即将重塑劳动力市场。无论解释多么细致入微,无论我在视频中说什么,都无关紧要。无论这项技术具有非凡的潜力,或者它可能治愈癌症,你所听到的是“又一项可能夺走我立足之地的事物”。这种恐惧,我的朋友们,是完全理性的。不是因为AI本身具有内在威胁,而是因为我们所处的经济环境使得任何重大的颠覆(Disruption: 指一项新技术或商业模式从根本上改变现有市场格局或运营方式)都让人感到事关存亡。
回想一下智能手机的出现,iPhone(Apple公司于2007年推出的第一代智能手机,开创了现代智能手机时代)早在2007年就发布了。它确实具有颠覆性,彻底改变了整个行业,导致纸质媒体衰落,重塑了零售业,并淘汰了那些无法适应的公司。但公众主要的反应是好奇,甚至有些兴奋。没有人上街反对智能手机。这并非因为智能手机的颠覆性不如AI,而是因为它们出现在一个大多数人仍感到足够的经济稳定,能够以兴趣而非恐惧来吸收变革的时刻。当时,世界在变,但人们脚下的土地足够坚实。而AI出现时,情况并非如此。
所以,这是我的第一个问题:当你对AI感到愤怒或恐惧时,这种情绪是针对技术本身——那些数学、研究、工程,还是针对我们身处的经济环境?在这个环境中,任何重大变革都可能成为最终压垮你的最后一根稻草。因为这些是不同的问题,它们有非常不同的解决方案。
Original English Source
I think to understand why AI generates the level of fear and anger it does right now, you have to understand the world it arrived into, because AI didn't land in a vacuum it landed in this economy. Since 1985, US median household income has risen roughly 255% in nominal terms, but median house prices have surged over 415%. The ratio of home price to actual income has gone from about 3.1 to a nearrecord 5.1. The median age of a firsttime home buyer has risen from 29 in the 1980s to 40 years old today. As of 2026, 65% of American households cannot afford a newly built medianpriced home. In some states, that figure is over 80%. You don't need me to tell you what this feels like. We're all living it together. The gap between what people earn and what life costs has been widening for 40 years. And for most people, there is very little margin left. Now, imagine you're in that position, stretched, precarious, one bad month away from real trouble. And somebody tells you that a new powerful technology is about to reshape the labor market. It doesn't matter how nuanced the explanation is or what I say in a video. It doesn't matter if the technology has extraordinary potential or that it might cure cancer. What you hear is another thing that might take away the ground that I'm standing on yet again. And that fear, my friends, is completely rational. Not because AI is inherently threatening, but because the economic environment we're all living in has made any major disruption feel existential. Think about when smartphones arrived, right? The iPhone launched back in 2007. It was genuinely disruptive. It transformed entire industries, contributed to the decline of print media, restructured retail, killed companies that couldn't adapt. But the dominant public reaction was curiosity, even excitement a little bit. Nobody was marching in the streets against smartphones. And that's not because smartphones were less disruptive than AI. It's because they arrived in a moment when most people still felt enough economic stability to absorb the change with interest rather than dread. The ground felt solid enough to stand on while the world shifted. AI arrived when it doesn't. So here's my first question for you. When you feel anger or fear about AI, is it directed at the technology itself, at the mathematics, the research, the engineering, or is it directed at the fact that we exist in an economic environment where any significant change feels like it could be the one thing that finally breaks you? Because those are different problems and they have very different solutions.
AI产品:强制推行与企业责任
现在,医学类比有一个失效的地方,我认为诚实地承认这一点很重要。对于医学,你通常是选择与系统打交道。你生病了才去看医生。没有人强迫你服用你不想要的药物。但是,AI正被强加给人们,无论他们是否需要。
微软(Microsoft: 全球领先的软件、服务和解决方案提供商)已将Copilot(Copilot: 微软推出的人工智能助手,集成到其多款产品中)嵌入到其销售的几乎所有产品中。谷歌(Google: 全球知名的互联网搜索引擎和技术公司)在搜索结果顶部插入了AI生成的摘要,这往往牺牲了人们所依赖的质量。Adobe(Adobe: 图像、视频、音频和设计软件的全球领导者)在未经许多艺术家明确同意的服务条款下,使用创作者的作品来训练其生成式AI模型。在各个行业的工作场所中,员工被“告知”而非“询问”要使用AI工具,无论这些工具是否合适、准确或受欢迎。这就像是制药公司推销不必要的处方药物。
这不是医学的错,而是特定公司做出的特定决定,他们优先考虑采用率指标和季度收益,而不是他们的产品是否真正被需要或有帮助。我认为,很大一部分合理且真实的愤怒正源于此——并非针对作为科学的AI,而是针对被强制推入我们数字生活每个角落的AI产品,这些公司决定我们没有选择权。
Original English Source
Now, there is a place where the medicine analogy breaks down. And I think it's important to be honest about that. With medicine, you generally choose to engage with the system. You go to the doctor when you're unwell. Nobody forces you to take a medication you don't want. But AI is being pushed onto people whether they ask for it or not. Microsoft has embedded C-Pilot into virtually every product it sells. Google has inserted AI generated summaries at the top of search results, often at the expense of the quality people relied on. Adobe trained generative AI models on creators' work under terms of service that many artists felt they never meaningfully consented to. In workplaces across every sector, employees are being told to use AI tools not asked told regardless of whether those tools are appropriate, accurate, or welcome. This is the pharmaceutical company pushing unnecessary prescriptions version of the analogy. It is not medicine's fault, but it is the fault of specific companies making specific decisions to prioritize adoption metrics and quarterly earnings over whether their product is actually wanted or helpful. And I think this is where a very significant and very legitimate portion of the anger sits not at AI the science but at AI the product being forced into every surface of our digital lives by corporations that have decided we don't get a choice.
面对AI:技术异议与伦理考量
此外,对于AI的构建方式,也存在严肃且具体的技术异议(Technical Objections: 对技术设计、实现或应用层面提出的具体质疑或挑战)。例如,训练数据是否获得了同意?谁获得了创意作品作为训练数据的使用许可?所有权(Ownership: 对财产或知识成果拥有合法权利的状态)问题:当AI模型生成了源自艺术家风格的作品时,谁从中获利?准确性(Accuracy: 信息或系统与真实情况或预期结果一致的程度)问题:当AI系统在医学、法律或新闻等高风险背景下产生自信、看似合理但实际是无意义的胡言乱语(Plausible Nonsense: 听起来合理但实际上是错误或误导性信息,常用于描述AI的“幻觉”现象)时会发生什么?
这些都是真实存在的问题。它们是法律、伦理和技术层面的问题。我们之前的一个视频中已经深入探讨了其中一些问题,我也会链接该视频。这些问题值得认真对待,而非置之不理。但请注意,这些异议无一是对AI作为研究领域的反对。它们是对特定公司采取的特定做法的反对。这种区别很重要,因为它改变了你投入精力的方向。
所以,这是我的第二个问题:你的沮丧是针对作为技术的人工智能(Artificial Intelligence)的存在本身,还是针对特定公司构建、部署和商业化(Monetizing: 将产品、服务或资产转化为经济价值,通常指盈利)AI产品的方式?因为其中一个在当前看来已是一种自然力量,而另一个则是一系列由可以做出不同选择的人们所做的决定。
Original English Source
There are also serious specific technical objections to how AI is being built. Questions of consent, who gave permissions for creative work to be used as training data? Questions of ownership, who profits when an AI model generates something derived from an artist style? Questions of accuracy, what happens when AI systems produce confident, plausible nonsense in highstakes context like medicine, law, or journalism? These are real problems. They are legal, ethical, and technical problems. We've covered some of them in depth in a previous video that I'm going to link as well. And they deserve serious treatment, not dismissal. But notice, none of these are objections to AI as a field of research. They are objections to specific practices by specific companies. The distinction matters because it changes where you aim your energy. So here's my second question. Is your frustration directed at the existence of artificial intelligence as a technology or at the way specific corporations are building, deploying, and monetizing AI products because one of those is a force of nature at this point. The other is a set of decisions being made by people who could make different choices.
解决方案:精确归因而非简单好坏论
我想用一些更具哲学性的思考来结束这段视频,因为我认为主导AI对话的“好与坏”框架本身就是问题的一部分。我认为AI既不好也不坏,我甚至不确定这些是否是有用的类别。在生活中,只有目标(Objectives: 期望达到或实现的结果),我们努力去实现这些目标,有些事物有助于我们实现目标,有些则阻碍我们。AI可能属于其中任何一种。它完全取决于其周围的系统(System: 相互关联的组成部分集合,作为一个整体运作以实现特定目的)。
在一个拥有强大劳动保护(Labor Protections: 旨在保护工人权利和福祉的法律法规)、有意义的社会安全网(Safety Nets: 政府或社会组织提供的支持,以防止个人或家庭陷入贫困或困境)和真正监管框架(Regulatory Frameworks: 规范特定行业或活动行为的法律、规则和机构体系)的社会中,一项强大的新技术是一个机遇。人们可以保持好奇心。他们可以重新培训、适应、探索,因为即使他们跌倒,脚下的地面也不会消失。
然而,在一个大多数人离危机只有一份工资之遥的社会中,在一个住房消耗我们一半收入的社会中,在一个人们不得不在有住所和有孩子之间做出选择的社会中,在一个社会契约在四十年里似乎被悄悄撕毁的社会中。同样的技术会让人感到是一种威胁,不是因为它本身就是威胁,而是因为我们无法承受它成为威胁。而且,没有任何当权者给出理由让我们相信他们会确保它不会成为威胁。
愤怒是真实的,恐惧是真实的。我不是来劝说任何人放弃这些感受的。我只是请求你精确地指向它们。因为“我讨厌AI”这句话没有一个有用的回应。但是,“我讨厌AI产品使用我的数据却没有监管框架”,或者“我讨厌我的雇主用半成品AI替代了功能工具”,又或者“我讨厌我生活在一个如此岌岌可危的经济中,以至于任何干扰都感觉像是死刑判决”——这些话指向了实际的解决方案。监管(Regulation: 政府或机构对特定活动或行业制定和执行规则以控制其行为的过程)、劳动保护(Labor Protections)、企业问责制(Corporate Accountability: 公司对其行为及其对社会、环境和经济影响负责的义务)、经济改革(Economic Reform: 旨在改变经济结构、政策或实践以改善经济绩效的措施)——这些是可以被要求、被组织、被投票的事项。
AI作为一门科学不会消失。它已经存在了70年,不会因为人们在互联网上发泄愤怒而停止。但是,围绕它的条件——经济上的不稳定、企业鲁莽的行为、监管真空——这些都是选择。它们是由人做出的,也可以由人撤销。
所以,你已经听到了这些区别:科学与产品、技术与经济背景、合法的技术异议与普遍的恐惧。你的愤怒究竟指向何处?我真的很想知道。请在下面的评论中告诉我。如果你想了解人们产生这些感受背后的硬数据——水资源提取、噪音污染、向从未要求这一切的社区收取数十亿美元——我推荐你观看我记录AI反弹的视频。链接在下方描述中。非常感谢大家的观看。请订阅,我们下期视频再见。
Original English Source
I want to close this video with something a little more philosophical because I think the good versus bad framing that dominates the AI conversation is actually part of the problem. I don't think AI is good or bad. I'm not even sure those are useful categories. In life, there are only objectives, things we're trying to achieve, and there are things that help us meet those objectives and things that hinder us. AI could be either. It depends entirely on the systems around it. In a society with strong labor protections, meaningful safety nets, and genuine regulatory frameworks, a powerful new technology is an opportunity. People can afford to be curious. They can retrain, adapt, explore because the floor beneath them isn't going to disappear if they stumble. In a society where most people are one missed paycheck away from a crisis, where housing consumes half our income, where people have to choose between having shelter or children, where the social contract feels like it's been quietly shredded over four decades. That same technology feels like a threat, not because it is one, but because we can't afford for it to be one. And nobody in a position of power has given us any reason to believe they'll make sure that it isn't. The anger is real. The fear is real. I'm not here to talk anyone out of those feelings. I am just asking you to be precise about where you aim them. Because I hate AI is a sentence that doesn't have a useful response. But I hate that there is no regulatory framework for how AI products use my data. Or I hate that my employer is replacing functional tools with halfbaked AI alternatives. or I hate that I live in an economy so precarious that any disruption feels like a death sentence. Those are sentences that point toward actual solutions. Regulation, labor protections, corporate accountability, economic reform, things that can be demanded, organized around, voted for. AI as a field of science isn't going anywhere. It's been around for 70 years and it's not going to stop because people are angry on the internet. But the conditions around it, the economic procarity, the corporate recklessness, the regulatory vacuum, those are choices. They were made by people and they can be unmade by people. So you've heard the distinctions, the science versus the products, the technology versus the economic context, the legitimate technical objections versus the ambient dread. Where does your anger actually sit? I generally love to know. Tell me in the comments below. If you want to see the hard data behind why people feel the way they do, the water extraction, the noise pollution, the billions being charged to communities that never ask for any of this, I'd recommend my video documenting this AI backlash. The link is in the description below. Thank you all so much for watching. Subscribe and I'll see you all in the next