山姆·奥特曼在斯坦福为 AI 尺度定律辩护

今日内容一览：[00:21] 山姆·奥特曼在斯坦福为 AI 尺度定律辩护[01:04] iRobot 的 Roomba 如何开启家用机器人革命[01:43] AI 创业公司提供免费保洁以换取训练数据[02:27] 加州大学伯克利分校研究发现 AI 导致大学成绩通胀---[00:00] Welcome to the Daily Tech Briefing. I'm your host. Today, we discuss Sam Altman's defense of AI scaling laws at Stanford and look back at the history of iRobot's Roomba. We also examine a company offering free apartment cleaning to train robots, and a new study from UC Berkeley on how AI is inflating college grades. Let's get started.欢迎收听每日科技简报。我是主持人。今天我们将探讨山姆·奥特曼在斯坦福为 AI 尺度定律的辩护，并回顾 iRobot Roomba 的发展历史。我们还将关注一家通过提供免费保洁服务来训练机器人的公司，以及加州大学伯克利分校关于 AI 如何导致大学成绩通胀的新研究。让我们开始吧。---[00:21] Sam Altman Defends AI Scaling Laws at Stanford. OpenAI CEO Sam Altman strongly defends the scaling of large language models. He believes that betting against this scaling is misguided (误导的，迷失方向的). During a speech at Stanford, Altman criticized AI skeptics (怀疑者). He said a whole generation of researchers held the field back because they did not trust scaling. For example, some critics call these models a dead end. But Altman argues the data shows otherwise. He noted that an OpenAI model recently solved a hard math problem that had stumped human experts. This proves that artificial intelligence can find new knowledge. However, Altman admitted that AI is still bad at tasks that require long-term planning and high human judgment.山姆·奥特曼在斯坦福为 AI 尺度定律辩护。OpenAI 首席执行官山姆·奥特曼强烈支持扩大大型语言模型的规模。他认为，不相信这种尺度定律是完全被误导的。在斯坦福大学的一次演讲中，奥特曼批评了人工智能领域的怀疑论者。他说，整整一代研究人员阻碍了该领域的发展，因为他们不相信尺度定律的力量。例如，一些批评者称这些模型是死胡同。但奥特曼认为数据证明了相反的结论。他指出，OpenAI 的一个模型最近解决了一个长期困扰人类专家的数学难题。这证明了人工智能可以发现新的知识。然而，奥特曼也承认，在需要长期规划和高度人类判断的任务上，人工智能的表现仍然很差。（来源：The Decoder）---[01:04] How iRobot's Roomba Started the Home Robot Revolution. The famous Roomba robot vacuum started a revolution in home cleaning. In the early days, the Roomba was a very simple machine. It had no smart maps; it just bumped around the house until its battery died. Even so, users loved their Roombas and often gave them human names. iRobot, the company behind Roomba, worked for nearly ten years to develop the vacuum. The company almost collapsed (倒闭，崩溃) several times before the product became a success. Although a huge market grew around robot vacuums, iRobot eventually struggled to keep up with new competitors (竞争对手).iRobot 的 Roomba 如何开启家用机器人革命。著名的 Roomba 智能扫地机器人开启了家庭清洁的革命。在早期，Roomba 是一款非常简单的机器。它没有智能地图；它只是在屋里到处乱撞，直到电池耗尽。即便如此，用户仍然非常喜欢他们的 Roomba，并且经常给它们起人类的名字。Roomba 背后的 iRobot 公司花了近十年的时间来研发这款吸尘器。在产品取得成功之前，该公司曾几次濒临破产崩溃。虽然围绕扫地机器人成长出了一个巨大的市场，但 iRobot 最终还是在与新竞争对手的竞争中挣扎求生。（来源：The Verge Tech & Culture）---[01:43] AI Startup Offers Free Cleaning in Exchange for Training Data. A technology firm called Micro AGI is offering free apartment cleaning in New York City. In exchange, the company records every corner of the home to collect training data. The cleaners wear cameras on their caps to record their hand movements. This data helps train future humanoid robots to perform tasks with high dexterity. The firm plans to sell this anonymized (匿名化的) data to other AI companies. However, privacy experts warn consumers about this practice. They call it a dangerous form of data bribing. They also warn that these recordings can reveal sensitive personal information. Still, the company's founder argues that this free service is an honest transaction (交易).AI 创业公司提供免费保洁以换取训练数据。一家名为 Micro AGI 的科技公司正在纽约市提供免费的公寓清洁服务。作为交换，该公司会记录房屋的每一个角落以收集训练数据。清洁工在帽子上佩戴摄像头，以记录他们的手部动作。这些数据有助于训练未来的类人机器人，使其能够以高度的灵巧性执行任务。该公司计划将这些去隐私的匿名数据出售给其他人工智能公司。然而，隐私专家对这种做法向消费者提出了警告。他们称这是一种危险的“数据贿赂”形式。他们还警告说，这些录像可能会泄露敏感的个人隐私信息。不过，该公司的创始人认为，这种免费服务是一笔诚实的交易。（来源：BBC Technology）---[02:27] UC Berkeley Study Finds AI Inflates College Grades. A new study from UC Berkeley suggests that artificial intelligence is inflating (使膨胀，通胀) college grades. Researchers analyzed more than five hundred thousand grades at a large public university. They found that grades rose sharply in courses with many writing and coding assignments after ChatGPT launched. However, this increase mostly happened on homework, which students do without supervision (监督，管理，指导). In courses that rely on supervised exams or oral presentations, grades did not change. This suggests that students are outsourcing (外包) their homework to AI rather than learning better. The study warns that grade inflation could lead to worse hiring decisions by employers. It also suggests that schools should change how they design exams.加州大学伯克利分校研究发现 AI 导致大学成绩通胀。加州大学伯克利分校的一项新研究表明，人工智能正在导致大学成绩通胀。研究人员分析了一所大型公立大学的 50 多万个成绩。他们发现，在 ChatGPT 推出后，在有大量写作和编程作业的课程中，成绩大幅攀升。然而，这种增长主要发生在家庭作业上，而这些作业是学生在没有监督的情况下完成的。在依赖监督考试或口头汇报的课程中，成绩并没有发生变化。这表明学生正在将他们的家庭作业外包给人工智能，而不是更好地学习。研究警告说，成绩通胀可能会导致雇主做出更糟糕的招聘决策。研究还建议学校应该改变他们的考试设计方式。（来源：The Decoder）---[03:18] 📚 今日单词：supervision- noun 监督，管理，指导- /ˌsuːpərˈvɪʒn/- 词源：Derived from the Latin 'supervidere', where 'super' means 'over' and 'videre' means 'to see'. It literally means 'overseeing' or watching from above.---[03:21] 词组/例句- without supervision 在没有监督的情况下- However, this increase mostly happened on homework, which students do without supervision. 然而，这一增长主要发生在学生在没有监督的情况下完成的家庭作业上。---[03:37] That is all for today's Daily Tech Briefing. Thank you for listening, and we will see you next time.以上就是今天每日科技简报的全部内容。感谢您的收听，我们下期再见。前往小宇宙评论区与主播互动