The economic situation is volatile, uncertain, complex, and ambiguous, so trying to predict anything about the future of work seems foolish. Barring any major global catastrophes, however, here are a few safe bets.

Anything that can be done by a machine, will be.

Humans have been using tools to carry out low-skilled tasks since before the agricultural revolution, but a perfect storm of technological, economic, and political conditions is set to accelerate this process dramatically. We’ve probably had the technology to replace restaurant and store cashiers with kiosks and apps for several years, for example, but up until now it hasn’t made economic sense to do so. The irresistible political pressure to raise the minimum wage, however, is incentivizing companies to invest in machine labor for these jobs. At $7/hour, it still made sense to hire humans to take orders. At $15/hour, it becomes more attractive for companies to invest in automation. The upfront costs are substantial, but the long-term benefit is a “cashier” who rarely makes mistakesn and never calls off, gets sick, or demands health benefits.

But it’s not just the low-skilled jobs that will be replaced. With advances in artificial intelligence and robotics, many skilled blue-collar and white-collar tasks will be automated as well. It’s important not to get stuck thinking about this as if it will be like The Jetsons, though. It’s unlikely that any given business executive will be replaced by the Execubot 3000. More likely, we’ll see 20% of that executive’s tasks being automated. This doesn’t mean the executive will do 80% of the job and take a few hours off every day. It means that a given organization will employ 20% fewer executives.

This same process of partial-job replacement is already in full-swing in some sectors. In some chain restaurants, for example, you now have the option to pay your bill on a tabletop kiosk. There is still a human server to take your order, bring your food, and refill your drinks, but with the bill-paying process handled by machines, the server is able to serve more tables, resulting in fewer servers for the restaurant overall. A similar process is taking place in retail stores. Instead of employing cashiers for each check-out, many grocery and big box stores now have a single employee troubleshooting a bank of self-checkouts. And, of course, ATMs have been partially replacing bank teller jobs for several decades already.

There will be fierce competition in the labor force for the tasks that can’t be done by a machine.

Let’s look at that executive job again. Of the 80% of tasks that can’t be automated, a significant portion of thoseĀ can be outsourced. There is nothing about living in close proximity to a corporation’s headquarters that makes one uniquely qualified to carry out certain tasks. If a self-employed contractor or consultant in India or the Ukraine is just as capable of doing some of these tasks–and they’re willing to do it for a fraction of the cost of a U.S. executive–then it’s hard to imagine a legal or political restriction that will prevent exactly that from happening.

Human interaction will be a luxury good.

Many people simply hate interacting with machines and will pay a premium for the opportunity to talk to a real person. Much will depend on what proportion of the population is willing to pay this premium and how much they’re willing to pay. Some people might be willing to pay an extra 10% for their groceries if it means they can interact with a human cashier, for example, but they may draw the line a 25% premium. If a critical mass of such people isn’t present in a given community, there may be no local options for that service.