AI and Society: Self-driving Cars and Other Challenges

Hello students, welcome to Kwickprep. A car with no driver stops at a signal, turns, and parks itself. But if it crashes, who is to blame? Today we learn how a self-driving car decides, and who is responsible when AI gets it wrong. Then we look at jobs, privacy and safety, and run a class debate.

Let us begin with the meaning. A self-driving car is a vehicle that senses what is around it and drives with little or no human help. It is also called an autonomous vehicle, where autonomous means able to act on its own. It uses AI to see the road, decide what to do and control the car. Engineers use levels from zero to five to describe how much of the driving the car does.

These levels come from an international body of automotive engineers. At level zero, the human does all the driving. At levels one and two, the human drives while AI helps. Many new cars in India have such driver assistance, like lane keeping. At level three, AI drives in limited conditions but a human must be ready to take over. At level four, the car drives itself fully, but only in a mapped area, like robotaxi zones in some cities abroad. Level five means driving anywhere without a human, and no such car exists yet.

A self-driving car senses the world with several sensors. Cameras read signs, signals and lane lines, but they struggle in fog or darkness. Radar uses radio waves to measure the speed and distance of other vehicles. LiDAR uses laser pulses to build a three dimensional map, but it is costly and affected by heavy rain. GPS and maps tell the car where it is, though signals weaken in tunnels. Ultrasonic sensors detect very close objects while parking.

Now let us see how the car decides, in a loop. First, it senses the world by reading all its sensors. Then it perceives, which means it identifies objects, like a cyclist, a cow or a traffic signal. Next, it predicts what each object will do in the next few seconds. Then it plans a safe path and speed. Finally, it acts by steering, braking or speeding up, and the whole loop repeats many times every second.

Pause and predict. A ball suddenly rolls onto the road in front of the car. The camera sees only the ball, and no child. What should a good self-driving car predict? A child may run after the ball, so the car should slow down early, just as a careful human driver would.

No system is perfect, so let us see how AI can go wrong. A sensor may miss a pedestrian crossing a dark road. The AI model may meet a case it never saw during training, like a cart pulled by a camel. A software update may contain a bug. Or a hacker may break into the car's computer system.

When an AI car causes harm, who is responsible? The manufacturer may be blamed for a faulty design. The programmers may be blamed for a bug in the code. The owner or driver may be blamed for ignoring warnings or not taking over. The government may be blamed if the rules were unclear. And the team that collected the training data may be blamed if the data missed certain situations.

This is the problem of accountability, which means someone must answer for a decision. An AI system cannot be sent to court or pay a fine. So humans and companies must stay accountable for what their AI does. Clear laws are needed to decide who pays for the damage. It is harder because many AI models are a black box, which means even their makers cannot fully explain each decision.

Self-driving cars also raise wider concerns for society. For jobs, millions of truck, taxi and auto drivers may lose work, though new jobs and reskilling can help. For privacy, the cameras record people and places all the time, so data protection laws like India's Digital Personal Data Protection Act matter. For safety, AI may reduce human errors, but it can make new kinds of mistakes, so strict testing is needed. For security, a hacked car is dangerous, so strong cybersecurity is essential.

These challenges go beyond cars. Bias happens when AI gives unfair results for some groups, for example in loan or job screening. Deepfakes are fake videos or voices made by AI that look real. Misinformation can spread faster when AI writes convincing fake news. And people may depend on AI so much that they stop thinking for themselves.

Indian roads bring their own challenges. Traffic is mixed, with autos, cycles, handcarts, pedestrians and cattle on the same road. Lane rules are often not followed, which makes prediction harder. As of now, India does not permit fully driverless cars on public roads, and a human driver is required. But driver assistance features are allowed and are becoming common in new cars.

Now let us run a structured class debate. First, set the motion, like, self-driving cars should be allowed on Indian roads. Next, form two teams, one For the motion and one Against, with a moderator to keep time. Then each team gives an opening speech of two minutes with its main points. After that, each team gives a rebuttal, which means replying to the other side's points. Finally, both teams close, and the audience votes, before the class reflects together.

A good debate follows simple rules. Use facts and real examples, not only feelings. Attack the argument, never the person who made it. Speak only in your turn, within your time limit. Listen carefully and note the other team's points so you can reply. The judges score on evidence and clarity, not on who speaks loudest.

Let us revise what we learned today. A self-driving car senses, perceives, predicts, plans and acts, in a fast loop. Automation has levels from zero to five, and level five does not exist yet. When AI gets it wrong, makers, programmers, owners and governments may share responsibility. Society worries about jobs, privacy, safety and security. And a structured debate uses a motion, two teams, openings, rebuttals and a vote.