Bots on the ground: China leads humanoid race

概览

This episode of The Intelligence covers three main stories: China’s fast-growing humanoid robot industry, Meta’s large legal settlement over teen social media harms, and the enduring appeal of Shakespeare and Company, the English-language bookshop in Paris.

The China segment argues that humanoid robots are advancing quickly as an industry, especially in China, but useful household or workplace droids remain far away. The bottleneck is not only hardware, but the enormous amount of real-world movement data needed to train robots to act safely and intelligently.

The Meta segment frames the $17 billion settlement as a landmark for social media regulation, though not yet a “big tobacco” moment in financial scale. The settlement may become a template for actions against TikTok, YouTube, Snap and other platforms.

The final cultural segment explains why Shakespeare and Company remains globally famous: its literary mythology, Paris location, deliberate mystique, book-buying rituals, merchandise, events and appeal as an offline community space.

分段落总结

[01:12] Episode lineup

[事实] Rosie Blore introduces the episode of The Intelligence from The Economist. [事实] The episode previews stories on teen social media use after a Meta ruling and a famous English-language bookshop in Paris.

[01:52] China’s humanoid robot spectacle

[事实] The World Humanoid Robot Games took place in Beijing, with robots competing in events such as powerlifting, boxing, football and running. [事实] A humanoid called Tiangong Ultra is described as running 100 metres in 9.39 seconds before crashing into a wall. [事实] Other robots crashed, fell, and one caught fire. [事实] Don Weinland says the event shows China is at the forefront of the humanoid robotics industry. [事实] Chinese companies are expected to sell around 50,000 humanoid robots this year, about triple last year’s figure.

[03:18] Why useful humanoid robots are hard

[事实] Don says training robots is more complicated than training chatbots because robots must learn physical interaction with the real world. [事实] He uses picking up a glass of water as an example: a robot must understand grip pressure, liquid movement and safety. [事实] The episode says humanoid robots may need around 100 million hours of real-world data, described as more than 11,000 years of data. [推测] The key limitation is not only mechanical ability, but the lack of enough high-quality physical-world training data.

[04:30] Real machine data and egocentric data

[事实] Don explains that “real-time machine data” comes from moving an actual humanoid robot and recording its motions. [事实] He says “egocentric data” can be collected from people wearing headsets, gloves, suits or boots with sensors while they move around. [事实] Both types of data can be used to improve robotic movement. [事实] China is becoming a front runner in real machine data through robot training centres.

[05:20] Inside a Wuhan robot training centre

[事实] Don and a colleague eventually gained access to the Hubei Humanoid Innovation Center in Wuhan. [事实] The centre contains staged environments including a coffee station, grocery store, pharmacy and factory line. [事实] At each station, human-robot pairs train robots to perform tasks such as pouring coffee, stocking shelves and picking up boxes. [推测] The centre is designed to turn ordinary service and logistics tasks into repeatable robot-training scenarios.

[06:41] China’s scale advantage and cost problem

[事实] Don says China is good at scaling things up and has already built 53 robot training centres, with 34 more planned. [事实] The Wuhan centre can fit about 100 humanoids, while a centre being built in Shanghai can fit 1,000. [事实] One city near Shanghai is building six such centres. [事实] Don says humanoids can cost up to $15,000 each, and training can cost 500 to 700 yuan per hour. [推测] China’s infrastructure buildout gives it an advantage, but the economics remain difficult if the industry truly needs tens of millions of training hours.

[08:03] JD.com’s mass data-collection plan

[事实] Don says JD.com is recruiting 100,000 staff and 500,000 people outside the company to wear sensor gear and record their movements. [事实] If successful over two years, JD.com may collect 10 million hours of real-world data. [事实] Don presents this as evidence that Chinese companies are exploring alternative ways to gather robot-training data. [推测] Large Chinese companies may be able to use their workforce and logistics networks as data-gathering infrastructure.

[08:52] Who is buying today’s limited robots

[事实] Don says many humanoid robots are bought by training centres. [事实] He describes training centres as a key source of demand for robots that currently cannot do very much. [事实] He says the industry is being driven by the promise of better robots in the future. [推测] Current sales partly reflect investment in future capability rather than present-day usefulness.

[09:38] The long-term robot vision

[事实] Don says household robots that clean, serve food and care for pets may be possible at some point, but he thinks that future is far off. [事实] He compares today’s humanoid robot moment to smartphones in the mid-2000s, before people fully understood their future uses. [推测] The industry’s eventual applications may be hard to predict from today’s awkward early machines.

[12:55] Parents and states take on Meta

[事实] The episode describes parents outside a California courthouse speaking about harms they attribute to Meta, which runs Facebook and Instagram. [事实] A consortium of 29 states sued Meta, claiming the platforms harm children and that the company knew about the damaging effects. [事实] The trial lasted five days and ended before Mark Zuckerberg took the stand. [事实] Meta reached a settlement described as worth up to $17 billion.

[14:33] Terms of the Meta settlement

[事实] Alice Fullwood says the settlement has two main parts: financial damages and restrictions on teen use of Facebook and Instagram. [事实] Meta agreed to pay at least $12 billion and as much as $17 billion over 10 years. [事实] The settlement requires new processes to assess users’ ages. [事实] Users believed to be under 18 will face restrictions including a two-hour daily cap across Facebook and Instagram, a midnight-to-6am blackout period, and limits on notifications. [事实] Teen users will have likes hidden and will be offered an option to switch from algorithmic feeds to chronological feeds.

[15:56] How effective the restrictions may be

[事实] Alice says the $17 billion figure is large but not comparable in scale to the big tobacco settlement of the 1990s. [事实] She says Meta’s maximum fine is about 10% of its previous year’s revenue and about one-eighth of what it plans to spend on capital expenditure such as data centres this year. [事实] Alice argues the more interesting question is whether platform changes will make Meta’s apps less addictive or compulsive for teens. [推测] The behavioural restrictions may matter more than the fine if they alter how teenagers actually experience the platforms.

[17:14] A template for other social platforms

[事实] Alice says the Meta settlement should be seen as a template for deals with other social media companies. [事实] The states can collect only $12 billion unless similar settlements are secured against other companies, which would raise the fine to $17 billion. [事实] The settlement creates incentives for states to pursue platforms such as TikTok, YouTube and Snap. [事实] Pennsylvania’s attorney general announced a lawsuit against Snapchat shortly after the settlement, and Snap’s share price fell by about 8%.

[18:43] Global and continuing legal pressure

[事实] Alice says the Meta settlement is the biggest one Meta has brokered so far, but only the tip of the iceberg. [事实] She says Australia banned young people from social media last year, and similar laws had been considered in France but were not proceeding at that moment. [事实] In the US, some states are still pursuing individual cases or other settlements with Meta. [事实] Thousands of personal injury lawsuits and more than 1,000 school district lawsuits have also been filed against the company. [事实] Early trials for those cases are expected to go ahead in February next year.

[19:48] Why the tobacco analogy only goes so far

[事实] Alice says the big tobacco analogy is appealing because both involve a potentially addictive product alleged to harm users. [事实] She says the lawsuits focus on what social media companies knew about harmful effects, similar to tobacco litigation. [事实] She notes that tobacco companies still performed very well financially after their settlement. [事实] Alice says Meta remains highly profitable and investors do not appear to view the settlement as a major threat. [推测] The settlement may increase regulation without fundamentally undermining Meta’s business model.

[21:33] Shakespeare and Company’s mythology

[事实] The episode turns to Shakespeare and Company, a famous English-language bookshop on the left bank of the Seine in Paris. [事实] Alexandra Suich Bass says tourists sometimes repeat false stories, such as Shakespeare having lived there or Sylvia Plath having founded it. [事实] Sylvia Whitman, the current owner, says guides often tell imaginative versions of the shop’s history. [事实] Her father George Whitman founded the shop 75 years ago under the name Le Mistral.

[22:42] The shop’s literary inheritance

[事实] George Whitman renamed the shop Shakespeare and Company after Sylvia Beach’s earlier bookshop, founded in 1919. [事实] Sylvia Beach published Ulysses when others considered it obscene and was friends with writers including Ernest Hemingway. [事实] The episode says legend has it that Beach gave George Whitman permission to adopt the name after her shop closed. [事实] George Whitman named his daughter Sylvia. [事实] Sylvia Whitman says her father encouraged some myths, including a playful claim about Walt Whitman being his grandfather.

[23:34] Why visitors keep coming

[事实] Shakespeare and Company receives 2,000 to 3,000 visitors per day, about one million per year. [事实] The shop has appeared in many films, and people still queue to enter. [事实] Photos and videos are prohibited inside, which Alexandra says enhances the shop’s mystique. [事实] Buyers can receive a stamp inside their book showing where it came from. [事实] The shop sells classics, luxurious hardbacks and merchandise, and hosts events with famous writers.

[24:34] A hard formula to copy

[事实] Alexandra says Shakespeare and Company’s formula is hard to replicate because of its unique history and central Paris location. [事实] She says the shop points to a broader trend in which independent bookshops are thriving in some countries. [事实] These shops serve as community gathering spaces for people who want offline experiences with others. [事实] Some locals complain that Shakespeare and Company attracts too many tourists. [推测] The shop’s success depends on both authentic history and carefully maintained scarcity around the visitor experience.

[25:31] Tomorrow’s episode preview

[事实] Rosie previews an upcoming Weekend Intelligence episode about the Ig Nobel Prizes. [事实] The preview says some Ig Nobel-winning findings have later proved useful, and some winners have gone on to win actual Nobel Prizes. [事实] This year’s Ig Nobels will be held outside America for the first time in 35 years. [事实] The episode says subscribers will be able to listen.

播客点评/总结

The strongest part of the episode is its contrast between spectacle and substance. The robot games provide memorable images, but the discussion quickly moves to the more important constraint: the massive, expensive process of collecting useful real-world data for humanoid robots.

The Meta segment is valuable because it avoids treating the settlement purely as a headline number. It explains both the practical restrictions on teen users and the legal incentives that could spread similar rules across other platforms.

The Shakespeare and Company story is lighter but effective. It shows how a cultural institution can turn history, myth, scarcity and ritual into lasting commercial appeal.

[推测] The episode is best suited for listeners who want concise briefings across business, technology, law and culture rather than a deep single-topic investigation. Its main limitation is that each story is necessarily compressed, especially the technical details of robotics and the legal mechanics of the Meta settlement.