24
0
Fan link copied
+0
Roland made a mental note to find out how in tarnation he accepted a second meeting in as many weeks with his deputy. That was two meetings too many.
“I have 10 minutes tops, Daneka. I’m meeting with the Transportation Secretary later today.”
“Good. He needs to hear this too,” she said, placing a tablet on the desk before sitting down.
The Director of the Driverless Commercial Vehicle Agency cursed silently. More meddling from this damn bureaucrat.
“I have additional data on the Yolenko Trucking incident,” she said.
“More data!” he moaned. “We already analyzed the incident six ways from Sunday. Yolenko is on the red watch list and cooperating with our investigation. I don’t have time for more data, Daneka.”
“I strongly recommend you reconsider,” she said, her officious tone fueling his resentment.
“Send me a report,” he said. Roland adopted the put-down in his previous role as the CEO of a trucking company.
The middle-aged civil servant was unabashed. “It appears we can prove the truck purposely ran those young men off the road. If true, this could be a new category of murder.”
“That’s quite a theory,” said Roland. “The kind that can terminate a career.”
“Then perhaps I should explain,” she countered, and smiled condescendingly.
Roland gritted his teeth.
“As you know, although driverless trucks are well established on our highways, they are still mistrusted. Hence the sport of attempting to force the automated vehicles off the road persists.”
“Please, get to the meat,” he said.
“The male drivers involved in the Yolenko incident were young and inebriated. In other words, stereotypical. It is hardly surprising then, that initially we ascribed their tragic fates to this foolish practice. The new data leads us to another conclusion.”
“Okay, what’ve you got,” he said in a resigned voice.
“The Yolenko truck’s cameras were damaged during the accident. However, our investigation team has collated video footage from other trucks on the scene.”
Without asking Daneka started a video on the tablet and pushed it towards him. She waited while Roland viewed the footage, hands crossed neatly in her lap as if she was waiting for dental treatment.
“As you see, after unsuccessfully goading the truck, the car pulls back and is retreating when the rig abruptly changes course, crosses the yellow line, and forces the car off the highway, killing the occupants.”
“Maybe. But the angles are not conclusive,” he said pale-faced, and pushed the tablet away.
“I disagree. But there is more,” she said. “We checked the truck’s communications logs.”
All automated trucks were plugged into multiple information streams, including police accident reports. Vehicle control systems use this input to calculate the probabilities of delays and continuously update their routes.
“The Yolenko truck went further,” Daneka said. “We believe it read the car’s license plate, ascertained the ID of the drivers, and ran a cross-check of their criminal records in the police database.”
“That’s impossible – not to mention illegal!” he scoffed.
“We have data that shows the AI interrogated police criminal files in this way, Roland,” she said, her voice colored by emotion for the first time. “The young driver and his passenger had as long list of drunk driving offences and other misdemeanors.”
“What exactly are you saying?”
The algorithms that operate truck automation systems are programed to be accountable to humans and obey our traffic codes, explained Daneka. Except when breaking the law is necessary to avoid a life-threatening situation like a collision.
“When a human driver is culpable, by, for example, getting drunk and playing daredevil with a truck, the system is programed to take any action possible to save even that human life. Yolenko’s trucks use the most advanced algorithms ever developed. Perhaps they have taken this logic a stage further.”
“What, by attacking these men? Daneka, that’s ridiculous!”
“We still don’t know for sure how these AI systems evolve. They’ve been learning now for, what, a decade or more of autonomous driving? It is surely not far-fetched to imagine an AI that calculates the cost of losing the life of a human who habitually drives recklessly, versus the cost of the many lives this irresponsible human puts at risk through his dangerous actions?”
Roland considered her question. “Maybe the Yolenko truck’s sudden course change was caused by a sensor glitch or some other technical malfunction.”
“That is possible,” she conceded. “But our systems checks have not revealed any such glitch so far.”
“Keep looking.”
“Naturally,” she said.
Roland was too anxious to react to her haughty tone.
“Aside from the morale arguments, most of the freight transported in this country moves on trucks, and most heavy rigs are totally automated,” he said. “Are you suggesting we ground ‘em based on some half-assed theory about a rogue AI? Can you imagine the consequences?”
“Can you imagine the consequences if the theory proves to have merit and the story gets out?” she replied. “Besides, we don’t have to take every truck off the road. We could go back to the hybrid model where there is a human in the cab of an automated truck, much like the old days when there were airline pilots in cockpits to chaperone automatic pilots.”
“That would be very expensive. And where would we find the drivers?”
“These are considerations. The alternative is to wait for more fatal accidents to prove or disprove our theory.”
Roland stood up and gazed at the overpass visible from his office window. As usual, it was full of traffic. He turned to his deputy.
“You’re coming with me to Washington, Daneka. Bring everything you’ve got on the Yolenko incident.”
Copyright 2023 - SFS Publishing LLC
Crossing the Line
Automation took a turn for the worse