By Tesla is set to unveil its long-anticipated robotaxi prototype, the “Cybercab,” marking a pivotal moment for the electric vehicle giant’s autonomous driving ambitions.
As Elon Musk pushes Tesla deeper into self-driving technology, the company is betting on a risky, AI-driven approach known as “black box” AI.
While competitors like Waymo and Cruise rely on a combination of sensors and mapping for safety, Tesla’s approach simplifies the process but comes with significant challenges, especially in navigating rare driving scenarios known as “edge cases.”
Unlike its rivals, Tesla’s self-driving system leans heavily on “computer vision,” a method using cameras to simulate human sight, paired with end-to-end machine learning, a form of artificial intelligence that processes raw data to make driving decisions.
This approach powers Tesla’s existing “Full Self-Driving” (FSD) feature, which still requires human supervision despite its name.
The company aims to build fully autonomous robotaxis using this same technology.
What is black box AI?
“Black box” AI refers to a form of artificial intelligence where the decision-making process is not easily interpretable or transparent.
Essentially, it operates like a “black box,” meaning inputs go in, and decisions come out, but understanding the logic or reasoning behind those decisions is difficult.
The key challenge with black box AI is that if something goes wrong, such as a misjudgment or accident, it is hard to pinpoint why the error occurred, making it difficult to prevent similar issues in the future.
This lack of transparency raises safety concerns, especially in critical situations known as “edge cases,” where unpredictable driving conditions or rare scenarios arise.
Unlike more transparent AI systems, black box AI can make it difficult to troubleshoot and enhance reliability.
What experts think about Tesla’s strategy
However, industry experts, including former Tesla engineers, have pointed out critical flaws in Tesla’s strategy.
One major concern is the unpredictability of “black box” AI, which lacks transparency in decision-making.
If something goes wrong, it is challenging to determine the cause of the error, making it harder to address safety concerns.
In contrast, companies like Waymo use additional sensors like radar and lidar to provide a clearer picture of the vehicle’s environment and ensure safer operation.
Tesla’s competitors, including Alphabet’s Waymo, Amazon’s Zoox, and General Motors’ Cruise, have already launched robotaxi fleets in select cities, but Tesla hopes to set itself apart by offering more affordable self-driving vehicles capable of navigating anywhere.
This bold move comes at a critical time for Tesla, as it faces slowing EV sales and increased competition from Chinese automakers.
While Musk has long promised the arrival of fully autonomous vehicles, he has yet to deliver on his vision.
In 2019, he predicted Tesla would have operational robotaxis by 2020, but that milestone has yet to be reached.
The announcement of this week’s robotaxi reveal follows Tesla’s decision to scrap plans for a $25,000 mass-market electric vehicle, raising questions about the company’s future priorities.
Tesla’s reliance on AI-driven computer vision presents unique challenges in managing “edge cases” — rare but critical driving scenarios that even advanced systems can struggle to handle.
Despite these hurdles, Tesla’s self-driving strategy has its supporters.
Meanwhile, Tesla’s expansive fleet of vehicles equipped with cameras provides the company with vast amounts of data, allowing it to refine its self-driving technology faster than competitors with smaller fleets.
However, critics argue that Tesla’s exclusive focus on AI and computer vision leaves it vulnerable to unpredictable errors that could result in dangerous outcomes.
As the race for autonomous driving continues, Tesla faces increasing pressure to deliver on Musk’s promises while ensuring the safety and reliability of its robotaxi technology.
While the potential rewards are enormous, the risks of Tesla’s “black box” AI approach are equally significant.
The post Tesla bets on ‘black box’ AI for robotaxis: What is it and why it matters for self-driving tech? appeared first on Invezz
