Portuguese 
English 
Spanish 
3 min of reading 
0 comments 
Camera With Artificial Intelligence Misidentifies Cell Phone Use, and Driver Is Fined for Just Scratching His Head While Driving.
An unusual case caught attention in the Netherlands. A driver was fined 380 euros for allegedly talking on his cell phone while driving. However, according to him, he was just scratching his head. The mistake was made by a camera with artificial intelligence.
Accusatory Photo and Surprise at the Fine
The incident occurred in November of last year. Tim Hansen received the violation notice a month after driving on a Dutch highway.
The fine indicated use of a cell phone while driving. Hansen, however, did not recall making any calls during that trip.
-
A doctor saw a 3-month-old baby turn blue due to lack of oxygen in a hospital without equipment and made a decision that kept the child alive with a cake box until help arrived; she maintained the situation for four hours in the interior of RN before the transfer.
-
Submerged fissure in Iceland looks like a window between two continents: glacial water filtered for up to 100 years traverses 50 km of volcanic rocks and reveals, with visibility over 100 meters, the separation between North America and Eurasia underwater.
-
Soviet villagers found a “human figure” with open eyes next to a capsule fallen in the snow and thought they had found a dead cosmonaut, until they discovered it was Ivan Ivanovich, the mannequin that tested the spacecraft before Gagarin became the first man in space.
-
African country mobilizes students, public servants, and religious communities in a giant task force, plants 350 million trees in just 12 hours, surpasses world record, and now aims for 50 billion seedlings by 2026 to restore forests that have shrunk from 35% to just 4% of the territory.
Curious, he accessed the website of the Central Judicial Collection Agency and examined the photo generated by the automatic system.
At first glance, it did look like he was holding a phone. But upon closer inspection, he realized that his hand was empty. He was just scratching the side of his head.
Despite this, the fine was validated by a human employee after checking the image. In other words, the system’s error was not identified during the review.
Expert Explains How Errors Occur
Tim Hansen works with information technology. He develops algorithms for image analysis and editing.
He used this experience to explain how the technology used by the Dutch police works.
According to him, these AI models need to predict binary responses: yes or no. And, in many cases, the system can make mistakes. In his case, the algorithm concluded there was a cell phone in his hand. But there was nothing.
“This is a false positive,” Tim explained. “The model predicted that I was holding a phone, but it was wrong. A perfect model only gets it right, but that is very rare.”
Incomplete Data Can Induce Error
Hansen also pointed out a possible technical reason for the failure. The technology was likely trained with many images of people using cell phones near their faces.
But it may have few images of people just with their hands near their heads.
This causes the system to automatically associate the hand’s position with cell phone use, even when there is no device at all. He suggested that the training database be expanded with more varied images.
Human Filter Also Fails
Even with the use of artificial intelligence, there is still a human reviewer.
This person must check the image and decide if the violation actually occurred. But, in Hansen’s case, the reviewer also confirmed the error.
For him, this shows that human supervision, although important, does not guarantee that all mistakes will be corrected. And it reinforces the need for more caution in using technologies that directly impact people’s lives.
Be the first to react!