China's Smart Driving Flip-Over Horror Scene: No Special Effects Needed!

China's Smart Driving Flip-Over Horror Scene: No Special Effects Needed! (Video)

[People News] Hello, dear audience, and welcome to "Forbidden News Decoded." I am Xiao Kun.

At various press conferences, the CEOs of major car manufacturers in mainland China are raising their voices higher than ever, with PPT presentations becoming increasingly fantastical. However, behind these beautifully packaged "technological myths" lies a series of gruesome rollover incidents on the highways.

On the internet in mainland China, you can find numerous spine-chilling "smart driving rollover" videos that require no special effects; they are pure horror.

The most ironic aspect is that when some vehicles are on the verge of collision, in the last second before disaster strikes, their smart driving systems do not assist the owner in making an emergency escape. Instead, they "flash faster than anyone else," displaying a message that reads, "Smart driving system has exited," and forcibly returning control to the owner, who has no time to react!

This is the reality of smart driving. When the bravado at press conferences collides with the twisted steel of crashed vehicles, and when the boundless trust in car companies ultimately results in shattered families, we cannot help but ask: where exactly is the "leading ahead" that you boast about every day?

The "Real Generational Gap" in Smart Driving between China and the U.S.

To uncover what truly lies behind those bloody car accidents, we must discard the marketing hype that has been fed to the public and examine the actual level of intelligent driving being promoted by mainland car manufacturers.

Many people online mistakenly think that domestic cars have outpaced foreign automakers, but in reality, the true leaders in intelligent driving technology are recognised within the industry as the American companies Tesla and Waymo, a subsidiary of Google. While Chinese car manufacturers claim to have 'surpassed the horizon' in their marketing, there remains a significant 'originality trench' separating them from the world's most advanced standards.

This gap is primarily evident in the core algorithm architecture and the ability to innovate.

Recently, you may have heard the term 'end-to-end large models' quite often. Domestic car manufacturers are now rolling out their own end-to-end intelligent driving systems one after another, as if this were their proprietary secret weapon. However, it is important to note that the originator of this concept and technical approach is Tesla's Full Self-Driving (FSD) system.

Before Tesla introduced the 'pure vision + end-to-end' approach, what were Chinese car manufacturers doing? They were heavily reliant on high-precision maps, essentially memorising road conditions in a 'cheating' manner. However, when faced with road repairs, detours, or areas where the map has not been updated, the vehicles quickly become ineffective. Tesla has broken through this limitation by employing neural networks to convert images captured by cameras into control commands, much like how humans use their eyes to see the road and their brains to think.

When American tech giants cleared the way and published their research, mainland car manufacturers suddenly realised the potential and rushed in to start 'pixel-level copying.' These companies have never developed an original foundational framework; they only know how to replicate others' work. While others invent tools from scratch, we merely seize the tools created by others and figure out how to adapt them. This consistent follower mentality means that we will always lag behind in the depth of core algorithms.

Moreover, there exists an insurmountable gap between mainland car companies and Tesla regarding the accumulation of high-value data.

The essence of intelligent driving lies in harnessing big data. Tesla's millions of vehicles worldwide collect real driving data daily through 'shadow mode' across diverse climates, terrains, and road conditions. Their algorithms are well-versed in real-world scenarios.

In contrast, domestic car manufacturers, despite boasting about the hundreds of millions of kilometres they claim to have driven, find that most of their data is confined to specific highways and streets in first- and second-tier cities in China.

This limitation has resulted in a critical issue: in cross-border blind tests conducted by third-party organisations, Tesla's intelligent driving system can still achieve remarkable avoidance results in various unexpected road conditions, even in foreign environments 'without any local training data,' thanks to its robust single-vehicle intelligence and generalisation capabilities.

Conversely, our domestic intelligent driving systems, which claim at press conferences that 'they can drive without maps,' tend to collapse when faced with extreme scenarios they have never encountered or when they encounter slightly unusual obstacles. In such cases, the 'boundary effect' within their algorithms is triggered, leading to an immediate failure of the vehicle.

Tesla's strategy focuses on 'purely single-vehicle intelligence'; regardless of the quality of external infrastructure, even in remote areas, the car is capable of understanding and driving itself effectively.

For an extended period, mainland China has promoted 'vehicle-road collaboration', attempting to enhance vehicle performance by installing 5G base stations and smart road signs on the roads.

While this concept sounds appealing, once a vehicle departs from the expensive smart roads and transitions to regular rural or provincial roads, or enters international markets, its diminished intelligent driving capabilities become glaringly apparent.

This explains why Tesla and Waymo in the United States can operate Level 4 autonomous taxis on public streets without safety operators, whereas the so-called autonomous vehicles in mainland China often remain slow-moving industrial toys performing in designated areas.

On a related note, I recently noticed a trend in technology marketing. Many tech products now highlight features like 'liberating hands', 'boosting efficiency', and 'making life easier', but the true foundation for maintaining good mental and physical well-being ultimately lies in one's own health.

When we delve into these so-called 'intelligent driving brains', we uncover another critical issue that warrants attention—the reliance on core hardware within China's intelligent driving industry.

The challenges of hardware and the reality of the 'choke point' concerning vehicle-mounted chips.

The mindset of many domestic car manufacturers is that when algorithms fall short, they turn to hardware. Since they cannot perfect algorithm optimisation, they simply load their vehicles with more hardware! As a result, we see mainland car companies competing at press conferences to see who can install the most lidar and who has the highest chip computing power. They often boast about dual Orin-X chips and computing power exceeding 2000 TOPS, which sounds quite alarming.

However, once we take a closer look at the 'brain' of the vehicle—the domain controller—we uncover a harsher and more unsettling truth: the foundation of China's intelligent driving technology is, in fact, built on shaky ground.

In the realm of high-performance intelligent driving chips, American companies Nvidia and Tesla have nearly cornered the global high-end computing power market. Nowadays, any domestic vehicle priced over 200,000 yuan that aims to showcase its advanced intelligent driving features will likely be equipped with Nvidia's Orin-X chip, as revealed in the specifications. Some models are even waiting in line to purchase the latest Thor chip.

Are there alternatives available in mainland China? Yes, there are. Huawei's Ascend series and Horizon's Journey series are making significant strides in domestic substitution. However, if we confront the objective reality, we find that domestic substitute chips lag behind the leading foreign chips by one to two generations in terms of manufacturing processes and energy efficiency.

This indicates that to achieve the same level of computing power, domestically produced chips may require larger sizes, higher power consumption, and more complex cooling systems.

A more easily overlooked yet critical bottleneck exists in a sector that most people have never heard of—automotive storage chips.

Many people mistakenly think that automotive chips are merely computing power chips, which is incorrect. When the intelligent driving model is operational, it receives massive sensor data amounting to several tens of gigabytes every second. At this point, the memory that temporarily caches data and the hard drive that stores data must maintain adequate read and write speeds and bandwidth. If the speed of the storage chip decreases, it is akin to a genius with an IQ of 200 being paired with a severely constipated digestive system; the entire system will quickly experience lag or even crash.

At present, the high-bandwidth memory required to support L3 and L4-level advanced intelligent driving, such as LPDDR5, LPDDR5X, and ultra-fast automotive UFS 4.0 storage, is nearly 100% monopolised by American Micron, South Korea's Samsung, and SK Hynix.

Chinese local storage companies, such as Changxin Storage and Yangtze Memory Technologies, while thriving in consumer-grade sectors like mobile phones and computers, still exhibit a significant gap in the extremely demanding 

What does automotive-grade mean? A mobile phone memory chip, when it fails, may only result in missing a call or needing a restart. In contrast, automotive storage must ensure continuous operation for 15 years under extreme temperature fluctuations ranging from minus 40 degrees to 105 degrees, while enduring the high-frequency vibrations of a vehicle day after day, guaranteeing that 'not a single character can be incorrect, and not a single byte of data can be lost.'

A single coding error could mean that the brake pedal fails to engage, or that the AEB (Automatic Emergency Braking) system could stop functioning at a critical moment when emergency avoidance is necessary!

In the development of high-capacity automotive storage chips that require rigorous safety certification, domestic chips lag behind industry giants like Micron by at least 2 to 3 years in research and development cycles.

This has created a highly awkward situation: domestic car manufacturers boast about their 'leading' intelligent driving capabilities in presentations, yet the core computing chips they rely on are purchased, the most crucial storage chips are imported, and even the foundational software toolchain that integrates these chips comes from abroad.

When your hardware lifeline is entirely controlled by others, and your foundational technology has generational gaps, what justification do you have to boldly claim that you have surpassed the world's advanced standards? This 'technological prosperity' built on shaky ground will only lead to embarrassment once the truth is revealed.

Boasting is tax-free: the "deadly poison" of marketing masters

Given that the underlying hardware is constrained and the algorithms are available for use, why does intelligent driving in China seem so unbeatable in the public discourse?

The answer is straightforward: Chinese car manufacturers possess a unique skill that no other car companies worldwide have—PPT marketing and concept hype.

In China, if you can't boast or invent new terms, you seem unworthy of being a CEO.

Let's examine some of the new terms that have emerged in recent years: "high-level intelligent driving," "can be driven nationwide," "can go wherever there is a road," "end-to-end no-map NOA," and "urban commuting mode"... each one sounding more impressive than the last.

On various short video platforms, you can even find some so-called automotive influencers and internet celebrities who, in a bid for views, deliberately activate intelligent driving on highways and then move to the passenger seat or even the back seat to sleep, play cards, or eat hot pot! This blatant disregard for traffic laws and life itself is not only overlooked by car companies but is even tacitly encouraged by some brands, who use it as promotional material to showcase their intelligent driving capabilities and reliability.

CEOs of car companies appear beaming at press conferences, confidently telling consumers: Our cars are already perfect; if you buy one, you can enjoy the freedom that technology brings.

What do ordinary consumers know about chips or algorithm boundaries? Under such intense brainwashing propaganda, they genuinely believe that their cars are already "autonomous driving"; they truly think that if you claim to be "far ahead," you can certainly help them avoid all dangers.

Do you know what the most shameless aspect of this entire situation is?

When these car manufacturers are under the spotlight, they boast about their vehicles as if they are miraculous creations; however, once you purchase the car and activate the vehicle's screen, agreeing to the electronic terms, you will encounter a line of fine print disclaimers: 'This function is only for L2 level assisted driving, not autonomous driving. The driver is solely responsible for safety and must always remain attentive to the road and be prepared to take control of the vehicle at any moment. The company is not liable for any property damage or personal injury resulting from the use of this function.'

Isn't this just a form of rogue logic? During promotions, they can't wait to tell the world that this car is so advanced it doesn't need a driver; yet when an accident occurs and accountability is required, the vehicle suddenly becomes a simple 'assistance tool,' allowing the manufacturers to offload all blame, legal responsibilities, and the cost of human lives onto the consumers!

'Aiming for wealth in danger, the risk is left to the car owner,' a sharp comment from a netizen. The car companies have achieved sales, stock prices have soared, and the CEO has enjoyed the spotlight. But what is the cost? The cost is borne by those car owners who trusted the marketing, risking their own and their families' safety to fill the technological void created by the manufacturers. They say boasting doesn't incur taxes, but the boast made by car companies is paid for with the lives of car owners!

A respect for life should be the fundamental principle of technology.

In discussing this matter, we must reveal the most fundamental and, indeed, the most disheartening truth about the intelligent driving spectacle: Chinese automotive companies are treating public roads as their free testing grounds; they are viewing consumers, who spend their hard-earned money on cars, as 'guinea pigs' who do not require salaries and may even pay with their lives!

In the traditional automotive industry, any technology related to personal safety must undergo years of closed testing, durability assessments, and extreme environment trials, often amounting to hundreds of thousands or even millions of kilometers before it can be marketed. Each airbag deployment and every brake system response must be calibrated thousands of times to ensure absolute reliability before mass production. This is because car manufacturers understand that these four wheels carry living human lives.

Yet, some emerging automotive companies are uncritically adopting the reckless and opportunistic approach of 'small steps, rapid iterations, and launching first while patching later' from internet giants into the critical automotive sector.

Is the system not fully tested? No problem, let’s push it to car owners via OTA updates first. Are there still flaws in the algorithm? No problem, let the car owners drive; if an accident occurs, we can collect data in the background and optimise the algorithm as needed.

This is not technological innovation; it is moral decay! It reflects a profound indifference and irresponsibility towards human life!

While any technological advancement is commendable, and intelligent driving is undoubtedly the future direction, such progress must never be built on the disregard for life safety.

As traffic management enforcement agencies find themselves repeatedly cleaning up after accidents, issuing fines, and sternly warning the public that "current intelligent driving systems are purely auxiliary, and any behavior by the driver that involves letting go of the steering wheel or taking their eyes off the road is a serious safety violation"; and as independent experts within the industry call for "maintaining safety standards and ceasing exaggerated claims"; one must ask the car companies still on stage boasting about being "far ahead"—do you not feel any guilt?

Tesla CEO Elon Musk has stated that for every incremental increase in the reliability of autonomous driving by one nines, such as from 99.9% to 99.99%, it requires exponentially more effort. However, some domestic brands, without even stabilising basic safety at 90%, boldly claim at press conferences that they have achieved "safe entry and exit".

This "intelligent driving myth", driven by capital, plagued by marketing, and reinforced by internet trolls, has turned into a complete farce in light of tragic car accidents.

We do not oppose technology, but we do oppose the deception and arrogance that lead consumers down a dangerous path. I would also like to advise all friends who have already purchased intelligent driving cars or are considering buying one: please remember that no matter how impressive the car companies make their presentations or how miraculous the sales pitches sound, the only ones responsible for your and your family's safety are your hands firmly gripping the steering wheel and your eyes focused on the road ahead.

Do not entrust your life to immature algorithms, and do not allow yourself to become a casualty of capital games and technological bubbles.

That concludes the content of this video. If you believe this video has exposed some lies and shared truths that others are afraid to discuss, please press and hold the like button and share it with your friends. If you have your own opinions on the mainland's 'Intelligent Driving Myth', feel free to leave a comment in the comments section.

We’ll see you in the next episode!

(‘Forbidden News Decryption’)
△