Will We Have Self-Driving Cars by 2030?
In 1956, General Motors made a film promising self-driving cars by 1976. People have been confidently predicting them ever since. This week you'll find out what the technology actually has to do — and then hit the questions engineering can't answer. How safe does a machine have to be before you'd trust it with your life? Who gets sued when nobody was driving? And why do we forgive a human mistake more easily than a computer's?
Reading List
You don't read all of these. Pick one from Group A and one from Group B before Tuesday, and pull a quote from each that stood out to you — you'll share both in your breakout room.
Group A · Pick One
Group B · Pick One
If You Want More
One-Week Plan
Here's the shape of the week. Your guide will set the exact timing for your section.
Would You Get In?
Five minutes of writing on one question: you call a rideshare, and the car that pulls up has nobody in the driver's seat. Do you get in? Would you feel safe? Why?
Then the history, which is funnier than you'd expect — a 1956 General Motors film promising self-driving cars by 1976, followed by a timeline of everything since: cars guided by cables in the 1960s, desert races nobody could finish in 2004, the first licence in 2012, the first fatality in 2016.
Then the harder part: why is this so difficult? Self-driving cars learn the way machine learning learns, which works best on simple repetitive tasks. But driving means reading a scene, guessing why the other driver just slowed down, handling a police officer waving you through a red light, and signalling to a pedestrian that you'll wait. None of that is simple or repetitive.
Before you leave, you'll pick one reading from each group and pull a quote from each.
Your Quotes, and Four Uncomfortable Questions
You'll start in a breakout room: which article you read, the quote you took, and why it stood out. About ten minutes, and everyone shares.
Then the questions that engineering can't solve:
- How safe is safe enough? How many deaths a day would you accept for the benefits — 1? 10? 100? Worth knowing before you answer: more than 100 Americans already die on the roads every day, about 90% from human error, and nearly 60% of drivers in fatal crashes are drunk.
- The impossible choice. Someone steps out with a stroller. The car can swerve onto the sidewalk where two adults are standing, or stay its course. What should it do — and could you accept a computer making that call?
- Who gets sued? A truck crosses the centre line so suddenly that no human could have avoided it, and the passenger of a self-driving car dies. Whose fault is it?
- Can the data even fit? A fully autonomous car's sensors would produce around 43 gigabits every second. A 500 GB hard drive would fill in under twelve seconds.
Notice what all four have in common: we accept an enormous amount of human error and almost no machine error. Is that reasonable, or just familiar?
Open the Discussion
Bring the point you prepared and the question you want to ask. Then:
- Would you use a self-driving car, or would you rather drive yourself?
- Will we have them by 2030?
- Will there be a day when cars that need drivers don't exist anymore?
- How much should we worry about hacking? A self-driving car is a networked computer, and a European cybersecurity agency found real vulnerabilities.
Before you leave, find one resource of your own and post it to the sharing board so everyone can look at it before Thursday.
Share What You Found
You'll share your resource, then pick one that a classmate posted, read it carefully, and say what it adds to the week.
The last question of the week is the biggest one: will machines ever be as good as us at some things — or will they eventually be better at everything?
Score Your Own Week
What went well, what to improve, what you learned, and whether anything you believed changed. Then the participation matrix. Be honest with yourself.