Few books have shaped my vision of what students need to learn more than The New Division of Labor by Richard Murnane and Frank Levy. Published in 2005, the book details the ways in which computers are transforming labor markets. They now have a new update to that book, a white paper titled Dancing with Robots: Human Skills for Computerized Work.
Levy and Murnane argue that computers do a few things very well, and they do those things very cheaply. In particular, they are extremely good at tasks that can be organized into a set of rules-based routines. Assign luggage tags to the airline passenger. Put the bolt here. Follow the rules of the tax code. Search for every instance of the word "kickback" in 2 million documents. If something can be broken down into a series of if-then-do statements, then sooner rather than later a computer or robot will be doing that task.
Computers, however, are still not very good at certain kinds of tasks, and Levy and Murnane put these into three big categories: solving unstructured problems, working with new information, and carrying out non-routine manual tasks. Unstructured problems are those where the desired end points are unknowable in advance or the set of information needed to solve the problem is unknowable in advance. Broadly speaking, we think of these problems as requiring creativity to solve. Working with new information is another way to discuss communication, which computers are still not very good at. Humans remain better at solving challenges that require social interaction to define the problem space or solicit necessary information. And robots remain laughably bad at some fairly basic manual tasks. For instance, it still can take robots up to 20 minutes to fold a single towel out of a laundry pile.
These observations lie at the core of the idea for 21st century skills. It's not than unstructured problem solving or working with new information are new skills for the 21st century, it's that they are newly important in the 21st century as computers replace routine-based work. In economic terms, humans have a comparative advantage over computers in these domains. In the past three decades, jobs requiring routine manual or routine cognitive skills have disappeared from the labor market, and jobs requiring solving unstructured problems, communication, and non-routine manual work have grown as a proportion of the labor market. The best chance of preparing young people for decent paying jobs in the decades ahead is preparing them with the skills to solve these kinds of complex tasks.