L2? L2.99? China and the U.S. are both failing to regulate self-driving cars

Marketing language often refers to the future, not the present — and drivers in both countries are getting confused, and sometimes hurt.

A NIO ES6 electric vehicle on display.

While auto industry disruptors are known for taking the world one step closer to self-driving cars, the actual cars being sold are still far from fully autonomous.

Photo: VCG via Getty Images

The tech worlds of the United States and China are usually operating on different timelines: Either the Americans lead in one field or the Chinese lead in another. But this past week, the two countries have found themselves in sync on one issue: the failure of proclaimed self-driving cars to … actually drive themselves safely.

Auto industry disruptors who have made a name for their pioneering autopilot systems are being questioned in both countries about whether these systems are causing severe travel accidents. In the United States, the National Highway Traffic Safety Administration just opened an investigation into Tesla's Autopilot feature following 11 traffic accidents that happened in a specific night driving scenario. In China, NIO, one of the leading Chinese autonomous vehicle makers, got grilled in the media after a fatal car crash that happened when NIO's autopilot system was at work.

In both countries, the controversies and media attention go back to the discrepancy between dream and reality: While these new auto brands are known for taking the world one step closer to self-driving cars, the actual cars being sold are still far from fully autonomous. Beijing and Washington are increasingly asking whether these car makers are promising too much and misleading drivers into over-dependence on their vehicles.

For now, neither Chinese or American auto regulators have found the answer. It's the drivers of Tesla, NIO and many more smart car brands who are shouldering the cost of this still-futuristic dream.

Pick a number between (L)2 and (L)3

On Aug. 12, 31-year-old entrepreneur and investor Lin Wenqin died in a highway traffic accident in southeast China. Photos of the scene show that the NIO ES8 that Lin was driving, a high-end SUV model comparable to Tesla's Model X, crashed into a heavy-duty vehicle parked on the highway. According to an obituary released by Lin's friends and family two days later, Lin's vehicle was on NIO's "Navigate on Pilot self driving mode" when the accident happened.

The possibility that a self-driving system is to blame for the accident immediately caught social media's attention. To date, the police haven't updated the public on their investigation, and the company NIO said in an announcement that it won't release any information before the police inquiry is finished. But when answering a question from the Chinese publication Beijing News, NIO emphasized that "NOP is still an assisted driving feature. It should not be seen as a self-driving feature." NIO didn't respond to Protocol's request for comment.

The question is whether the victim, and the drivers of NIO and other autonomous vehicles, know the difference.

The United States and China have similar standards to define how "autonomous" a vehicle is. Cars are grouped into six categories, from L0, where the driver controls everything, to L5, where the car performs all tasks. Today, the Japanese brand Honda boasts the only mass-produced model that has reached L3, the Honda Legend, a critical milestone that frees the driver from watching the road.

But the clarity on paper doesn't translate well into reality, especially when it's not always in the car manufacturer's interest to admit how limited their self-driving technology is.

The many Chinese car manufacturers that have yet to reach L3 have focused on how to make their assisted driving technology sound more appealing than their competitors'. As a result, they have marketed their cars using terms like "L2 Plus" (BYD), "L2.5" (Jetour), "L2.75" (XPeng) and even "L2.99" (Chery).

The numbering system is confusing, but it gets even worse when companies are coming up with acronyms. Between NIO's NOP (Navigate on Pilot) and NAD (NIO Autonomous Driving), Tesla's NAP (Navigate on Autopilot), and XPeng's NGP (Navigation Guided Pilot), it's become hard to tell which technology is more advanced.

In the United States, Tesla has gotten into plenty of trouble for calling its Autopilot system "Full Self-Driving." In China, other car brands have deployed the same tactics to, purposefully or not, blur the boundaries between assisted driving and self-driving. The most egregious example, according to Xing Lei, an auto industry analyst and former chief editor at the Beijing-based China Auto Review, is when the Shanghai-based brand Weltmeister publicly advertised in April that it had "the first mass-produced self-driving car in the country," equating to L4. But later it was revealed that technology was only available when the car was parking.

Blurred lines

Perhaps the bigger problem is the gulf between assisted driving — what an L1 or L2 vehicle actually offers — and "self-driving" or "autonomous driving."

After the recent highway accident, Li Xiang, founder and CEO of Chinese car brand Li Auto, called on the industry to refrain from using the word "自动" (which means "self" or "autonomous") in its marketing before reaching the L4 stage.

On paper, NIO seems to have done a decent job explaining its technology. According to Chinese publication LatePost, in NIO's 100,000-character-long driver's manual for its ES8 model, it only says "self-driving" once, and that's to warn that autopilot does not equate to self-driving. On its website, the word also never appeared when describing its currently available models.

"I've talked to three or four salespersons of Tesla China and two salespersons of NIO," a Shenzhen-based Tesla owner who wishes to stay anonymous told Protocol. "When I asked them if they have self-driving cars, they all seemed well-trained and told me it is assisted driving, not self-driving."

But some buyers aren't content with NIO's efforts to teach the difference. In a discussion forum on NIO's app, one buyer nicknamed "the unfillable stomach" wrote: "I can tell you loud and clear that it's impossible" for everyone to understand NIO's autopilot system correctly. "At least when I bought the car, I was told by everybody that Navigate on Pilot gives you self-driving."

Drivers clearly differ in their views. Soon after the accident, 500 NIO owners signed an open letter in support of the company, pledging the company has done a good job teaching them the differences between assisted driving and autonomous driving. But then came the backlash, with other NIO drivers starting a hashtag "Reject the Open Letter" on NIO's app.

Customers often want to think that their cars are more powerful than they actually are. "After autonomous vehicle companies exaggerate their capabilities, some consumers, especially those who are young and chasing after new technologies, may try to show off or play with the system," Xing told Protocol, "and that's very dangerous."

What, me regulate?

Evidently, neither Chinese nor American regulators have a handle on this problem. On the same day of the fatal accident, the Chinese Ministry of Industry and Informational Technology published a new regulation that asked autonomous car makers to ensure there's a safety mechanism to detect whether the driver's hands are still on the wheel. But the U.S. and China are both on a blank page when it comes to determining whether auto companies are exaggerating their technologies, and to what extent that should be forbidden.

Meanwhile, car makers continue to depict a great driverless future, even if that's not yet offered by a particular product they currently sell. While NIO has been careful about not calling its current pilot system "self-driving," it freely sprinkles the word in when describing its latest technology, NAD, a more sophisticated system that will be released on NIO cars in 2022. Once released, NAD will bring about "a safer and more relaxing autonomous driving experience from point A to point B," the website says.

As always, the devil — or in this case, the reality — hides in the fine print. At the bottom of the NAD page, a line reads, in light grey color and smaller text: "NAD in the description is designed for driver assistance purposes only and cannot fully replace driver's control or handle all possible traffic, weather and road conditions."


What Chinese tech CEOs are telling Beijing

The Chinese style of lobbying, in the form of “two sessions” proposals, is more open but less effective and helpful in deciphering what tech companies and Beijing want.

NPC and CPPCC proposals are supposed to be expert advice on what the Chinese government should do next. But they can often double as open lobbying efforts.

Photo: Paul Yeung/Bloomberg via Getty Images

The CEOs of China’s largest tech companies have a wishlist for their government — at least, a version of their wishlist they can say publicly — for this year: Watch out for metaverse risks! The more electric vehicles, the merrier! Make AI greener!

Every year around March, Beijing holds the most important national political meetings: the National People's Congress (NPC) and the Chinese People's Political Consultative Conference (CPPCC), usually referred to as the “two sessions.” Thousands of members — government officials, entrepreneurs, workers and academics — gather in Beijing for legislative discussions and to submit their “suggestions” (for NPC) and “proposals” (for CPPCC) to the national government.

Keep ReadingShow less
Zeyi Yang

Zeyi Yang is a former reporter with Protocol | China. Previously, he worked as a reporting fellow for the digital magazine Rest of World, covering the intersection of technology and culture in China and neighboring countries. He has also contributed to the South China Morning Post, Nikkei Asia, Columbia Journalism Review, among other publications. In his spare time, Zeyi co-founded a Mandarin podcast that tells LGBTQ stories in China. He has been playing Pokemon for 14 years and has a weird favorite pick.


How the CAC became Chinese tech’s biggest nightmare

The Cyberspace Administration of China’s core functions have expanded from content control to data security and privacy, and it now affects the entire digital economy.

CAC's rising power and expanding territory are evident in the spike in the number of releases, policies and regulations it issues.

Photo: Jordan McAlister via Getty Images

The Cyberspace Administration of China (CAC) rose to fame as China’s central internet censor and it still is. But in the past few years, the agency has expanded its regulatory scope, gradually morphing into a super regulator that affects virtually every internet company in China.

The CAC has been making headlines all over the world for the past two years as its power has grown exponentially and it has become the leading Chinese regulator in data security and privacy — playing an instrumental role in China’s ongoing tech crackdown.

Keep ReadingShow less
AJ Caughey

AJ Caughey is a data researcher for Protocol | China. Previously, he worked at Meridian International Center administering State Department cultural exchange programs and as an Applied Data and Governance Fellow for the International Innovation Corps. AJ holds master's degrees from the University of Chicago's Harris School of Public Policy as well the Committee on International Relations.


Alibaba is walking a tightrope between Russia and Ukraine

Three of Alibaba’s business partners in Russia have been sanctioned. The ecommerce giant is in an awkward position facing internal and external pressures.

Tech firms based in the U.S. and Europe have been quick to support their Ukrainian employees or suspend operations in Russia, but Chinese tech companies have been much more cautious about taking a stance.

Illustration: Protocol

The day before bombs started raining down on Kyiv, Dmytro Romashko streamed for three hours on AliExpress, Alibaba’s global ecommerce platform. Sitting in his house, the 31-year-old Kyiv resident showcased an array of products, including electronic gadgets and decorative mugs, to thousands of Russian-speaking viewers. Romashko was in the same house when, at five o’clock the next morning, he watched with horror as Russian shells descended and fled to a bunker in only his underwear.

Romashko has worked with the Chinese ecommerce platform since 2014. He started out writing product reviews before eventually becoming one of its most prominent Russian-speaking livestream influencers. He has more than 268,000 followers on AliExpress, and each of his streams was regularly watched by thousands of viewers across Eastern Europe and Central Asia. While Romashko has never been an Alibaba employee, he serves the company’s ambition to bring its Chinese-style ecommerce craze to the world.

Keep ReadingShow less
Zeyi Yang

Zeyi Yang is a former reporter with Protocol | China. Previously, he worked as a reporting fellow for the digital magazine Rest of World, covering the intersection of technology and culture in China and neighboring countries. He has also contributed to the South China Morning Post, Nikkei Asia, Columbia Journalism Review, among other publications. In his spare time, Zeyi co-founded a Mandarin podcast that tells LGBTQ stories in China. He has been playing Pokemon for 14 years and has a weird favorite pick.


How China’s social media handles fake news about Ukraine

Douyin, Weibo, WeChat and Bilibili scramble to tackle misinformation about Russia’s invasion into Ukraine.

The social media platforms are also cracking down particularly hard on misogynist jokes that demean Ukrainian women.

Photo: Sean Gladwell/Getty Images

Social media platforms worldwide are waging their own battles against fake news and disinformation related to Russia’s invasion of Ukraine. Chinese social companies are no exception. In recent days, Chinese social platforms — from Douyin to Weibo, from WeChat to Bilibili — have been scrambling to remove accounts that are spreading misinformation and making warmongering remarks and vulgar jokes, including some content that seems to have endangered the lives of Chinese nationals stranded in Ukraine.

TikTok’s Chinese original version, Douyin, revealed Sunday in a post on WeChat that its short-video platform had already pulled 3,518 video clips and 12,179 comments regarding Russia’s invasion into Ukraine. It also imposed punishments, including account suspension, on almost 500 specific accounts that spread misinformation.

Keep ReadingShow less
Shen Lu

Shen Lu is a former Protocol reporter, covering China's tech industry.


Death of a Bilibili censor reignites overwork debates in China

The unexpected death of a content moderation worker at Bilibili brings the plight of Chinese censors into focus.

Content moderation is critical to keep platforms running, but the individuals toiling away behind the screens are certainly not treated as essential workers.

Illustration: Christopher T. Fong/Protocol

Content moderation is challenging, largely hidden work all over the globe. China’s tens of thousands of content moderators are no exception: They’re an invisible, overworked workforce who keep social apps from Douyin and WeChat to Bilibili and Weibo running. But unlike other overworked groups, such as gig delivery workers, their plight was nearly unknown to the public until the recent tragic death of a worker at one of China’s leading video platforms.

On Feb. 4, a 25-year-old content moderator for Bilibili died from a brain bleed. Screenshots floating online that his colleagues at Bilibili leaked to a tech blogger showed that he was asked to work extra shifts during the Lunar New Year, when almost the entire country’s workforce was on holiday. But Bilibili denied in an internal email he had worked overtime before his death, emphasizing he only worked regular eight-hour shifts the previous week.

Keep ReadingShow less
Shen Lu

Shen Lu is a former Protocol reporter, covering China's tech industry.

Latest Stories