Angad Singh thinks there’s a good chance that within 20 years, your job will no longer exist. And if it does, a bot will do it.

“It seems obvious,” he says. “The nature of work will be completely different.” By different, he means that anyone currently doing “non-creative” work will be made redundant in favor of software or robots. So long, assembly line workers. Good luck, white-collar paper pushers. Anything that can be automated will be. Even professions once thought to require actual human beings, like law and medicine, can be at least partly robotized.

Angad Singh

“It will be super different, but in a good way,” Singh says. “It’s the way we want work to go.”

For someone predicting the mass unemployment of millions, Singh is incredibly optimistic about the future. That’s because he believes everyone has the ability to be creative, and he thinks his company, Lemonade.io, can teach them how.

Lemonade was founded in 2014 and is backed by New Enterprise Associates, which also invested in Duolingo, Box, and Coursera. Its advisers include David Kelley, the founder of design firm IDEO and Stanford’s d.school, and Jerry Yang, the cofounder and former CEO of Yahoo! Inc. Singh envisions a future where Lemonade’s creativity technology integrates with everything in the modern workflow, from Slack to Gmail, and gives employees real-time feedback on their work to help them think outside the box.

But before Lemonade can train people to be creative, it needs to train its own technology to be intelligent enough to know what creativity looks like. That may sound a bit freaky to anyone who thinks of imaginativeness as being some sort of divine gift, but actually, we’ve known for a long time that creativity can be quantified. In the ’80s, a Harvard professor named Teresa Amabile created the Consensual Assessment Technique, which is still used today and operates on the assumption that when multiple people are asked to evaluate how creative something is, they all tend to be consistent in their answers.

“People have very similar opinions of creativity,” says Singh. In other words, what we humans consider creative is pretty predictable. It usually boils down to novelty (is the idea somehow unique?) and value (does the idea solve a problem?). Computers can learn to assess both of those things through machine learning. But that takes data. Lots and lots of data. That’s where games come in.