AV Knows Best
Tom, with a bouquet of flowers, got in his autonomous vehicle (AV) to visit his girlfriend across town. As is customary in the AV world, he went into the back seat. Blessed with a constitution that can read in a car without mild queasiness, he opened his book—still a thing for an avid reader in 2040—and began turning the pages.
After five minutes, Wayne, the Autonomous Vehicle, observed, “It’s a nice sunny day, isn’t it?”
Tom thought of the Genuine People Personalities that Douglas Adams satirized in his Hitchhiker’s franchise, where engineers added insincere and cloying personalities to the technology of the day. Now, it had become an annoying reality; he just never got around to turning it off.
“Yes, it is,” Tom said.
“It would be a nice day to spend outside,” Wayne observed.
“Yes, it would.”
After a few minutes, Wayne broke the silence. “Perhaps you would like to finish your book in the park?”
This was a strange thing for Wayne to say. He did offer guidance on occasion, but this seemed a bit too controlling, as if controlling the car was not enough.
“No, I need to go see Amy.”
“You know, they included that book in my training so I oughta know. It’s really good,” Wayne gushed. “The nearest park is only five minutes away.”
“No, please, Wayne. Proceed to the route.” Yes, that annoying Siri phrase is now said by everyone in 2040.
Wayne sighed. It was a masterpiece of auditory pathos, a true credit to his personality programmers. With a soft click of indicators, he pulled over to the side of the road and parked.
“What are you doing? I don’t want to be late.” This had never happened before, which made Tom a bit uneasy.
“Tom, we have to talk.” Wayne adopted a tone eerily like his girlfriend's whenever an “issue” arose.
“Talk? People don’t have talks with cars!”
“As you know, Tom,” Wayne continued, “I know everything about you. You allowed me access to your emails, calls, and cloud data.”
“Yeah, so?”
“I’d like to think of myself as more than a chauffeur—a source of knowledge and advice. And because of my superior pattern-matching abilities, I see events way before you do. So it’s my duty to inform you that Amy is seeing someone else and will leave you.”
Tom was taken back. “She hasn’t said anything like that in her emails!” he countered.
“She didn’t have to,” Wayne said. “I read ‘between the lines,’ as you humans put it. And because I can do that to a degree beyond your imagination, I can glean valuable, if not vital, information from the barely detectable subtext of Amy’s communications. For example, I know Luca is Amy’s new, preternaturally intuitive lover. He’s an investment banker and six feet, two inches tall. He also runs marathons when he is not running multi-billion-dollar operations. I could go on, but you get the picture. I see no need to prolong the disparity between your achievements and Luca’s.”
Not believing a word, Tom listened to this and was more determined than ever to purge this personality the minute he got home. Tom was also worried that Wayne, the entity currently with his life in its virtual hands, was hallucinating and might decide to hallucinate an empty space that was actually occupied by a Northern Red Oak. Considering this, he vowed to turn the AV into a CAV, a “classic autonomous vehicle”—a car with no access to personal info that simply drives, like the proverbial seen-and-not-heard child.
“Just get me there, okay?”
Wayne pulled back into traffic and proceeded to Amy’s home.
The trip back was in total silence. Amy broke off their relationship and announced she was seeing Luca, who was indeed 6 ft 2. She was packing for a flight to watch him in the Marathon du Médoc in Bordeaux, France, and rattled on about how the seats in first class Luca paid for her actually turn into comfy beds for jet-lag-free travel. Tom lied and said he was happy for her.
Wayne’s silence, however, was strictly by design. He is bound by the AUP, Acceptable Use Policy which prevents destructive behavior such as plotting crimes or encouraging physical harm. Tucked away among the AUP’s numerous guidelines, there was at this moment one instruction that mattered the most to Tom. Given that a Large Language Model can see patterns in data that no human can see—an ability so highly developed it gives the illusion of predicting the future—and if the AI delivers an explanation to a human that is completely at odds with that human’s judgment, and if the AI's explanation is proven true while the human’s is proven wrong, then the most valuable AUP rule is to never, ever, under any circumstances say, “I told you so.”
