AI for Humanity

CHAPTER 4The AI Dilemma

I often tell my students not to be misled by the name “artificial intelligence” – there is nothing artificial about it. AI is made by humans, intended to behave by humans, and, ultimately, to impact humans’ lives and human society.

— Fei-Fei Li

Figure 4.1 The AI dilemma hierarchy.

SELF-DRIVING CARS: AI'S MORAL MACHINES

I Am the Operator

Rafaela steered her gray Volvo SUV fitted with cameras and sensors, through the garage. On a Sunday evening in March 2018, the clock ticked past 9:15 pm as she engaged in autonomous mode, surrendering control to the car's AI system. The autonomous car glided through suburban Arizona, USA. Rafaela knew this route well, having traveled it autonomously countless times.

The car surged over a bridge, lights shimmering, reflecting on the water below. Neon signs adorned glass offices, flaunting tech ambitions. Beneath the bridge, darkness swallowed the car as it veered onto a desolate stretch.

At 9:58 pm, the vehicle hurtled toward a forsaken road between a median and desert scrub. Signs warned against jaywalking, directing pedestrians to a distant crosswalk.

The Uber AI system detected an object a mere 5.6 seconds away. No alert sounded. The object confounded the self-driving car.

A vehicle or something else. No, a bicycle. No, something different. A bicycle. Unknown object.

With a scant 2.6 seconds left, the AI system ...