Do you still use Yahoo? Do you still remember MySpace? Compaq? Kodak? The cases of startups with superior ideas dethroning well-established incumbents are legion. This is the beauty of “creative destruction” – the term coined by Innovation prophet Joseph Schumpeter almost a century ago. Incumbents have to keep innovating, lest they be overtaken by a new, more creative competitor. Arguably, at least in sectors shaped by technical change, entrepreneurial innovation has kept markets competitive far better than antitrust legislation ever could. For decades, creative destruction ensured competitive markets and a constant stream of new innovation. But what if that is no longer the case?
The trouble is that the source of innovation is shifting – from human ingenuity to data-driven machine-learning. Google’s self-driving cars are getting better through the analysis of billions of data points collected as Google’s self-driving cars roam the street. IBM Watson detects skin cancer as precisely as the average dermatologist because it has been training itself with hundreds of thousands of skin images. Siri and Alexa are getting better at understanding what we say because they never stop learning. Of course, it takes plenty of talented, creative people to build these products. But their improvement is driven less by a human “aha-moment” than by data and improvements in how machines learn from it.
The Risks and Rewards of AISponsored by SASAssessing the opportunities and the potential pitfalls.
Sometimes companies have to go out and collect a specific kind of data – think of Google’s cars roaming the streets of Silicon Valley. And sometimes companies pay for access to data so that their systems can learn. But more often than not, the data that fuels innovation is being generated by users interacting with an existing digital service. When we accept Siri’s suggestion, it’s feedback to Siri that she got it right. And when we surf away from Amazon’s product recommendation, it’s another feedback signal that we weren’t so happy. It’s the same when a driver in a Tesla takes over from assisted driving, or when we accept (or don’t accept) Google auto-completing our search query. This feedback data is incredibly valuable because it is the raw material feed into machine learning tools; it’s the very resource that fuels data-driven innovation. And the more you have, the better you get. Take self-driving cars as an example. During 2016, self-driving cars by major international car manufacturers improved by roughly a third. That’s a significant jump. But Google collected far more data per car to feed a more advanced machine learning system, and its cars improved by 400%– an amazing jump in innovation, and more than ten times as much as cars utilizing less data.
But if innovation is founded on data rather than human ideas, the firms that benefit are the ones that have access to the most data. Therefore, in many instances, innovation will no longer be a countervailing force to market concentration and scale. Instead, innovation will be a force that furthers them.
This would be a fundamental change to competition, and could cause market after market to become concentrated – as has already been happening in the U.S. If this happens, conventional antitrust measures won’t be much help, because they restrain uncompetitive behavior – but large companies using their data to learn and innovate isn’t illegal. In fact, they’re acting perfectly efficiently, using the benefits of their scale to squeeze novel insights out of troves of data.
The specter of companies with access to data becoming data-driven innovation leaders, leaving smaller competitors and startups behind in the dust, should concern policymakers intent on ensuring that markets stay dynamic and competitive. Their challenge is less to realize the problem than to devise a solution that keeps markets competitive without stifling data-driven innovation on the whole.
Most business leaders, on the other hand, face a very different challenge in this world of data-driven innovation. To compete against digital champions, they will have to overcome not just scale and network effects but especially these new data-driven feedback effects. For many innovative companies, the next few years will be a time of reckoning: as the power of data-driven innovation increases, these more conventional innovators will have to find access to data to continue to innovate. That necessitates at least two huge adjustments. First, they need to reposition themselves in the data value chain to gain and secure data access. That’s difficult if, for instance, all the data is captured upstream in the data value chain. Just ask suppliers in car manufacturing, or book publishers. Second, as innovation moves from human insight to data-driven machine learning, firms need to reorganize their internal innovation culture, emphasizing machine learning opportunities and putting in place data exploitation processes. This is hard because it often runs counter to an engineering culture that has long championed human ingenuity.
The adjustment will be so severe that numerous innovative firms will falter in the coming years, overtaken by more data-savvy competitors. And those that succeed will look very different than they do today. But firms wanting to stay innovative have no other choice. You may be doing well with your innovative company today, but as the source of innovation shifts, you will need to as well.