As technologies go, AI is unique in many respects, many analysts argue.
While that’s undoubtedly true, in certain ways, AI is behaving just like many new technologies before it. Take the question of whether AI is a “disruptive technology” and its impact on the competitive dynamics of an industry.
Ben Thompson, one of the more astute analysts of technology, argued in August in a post titled “Autonomy and Innovation” that “it speaks to the incredible capability of AI that it is setting up to be both [a sustaining and disruptive innovation].”
But it turns out that AI isn’t at all “incredible” or surprising in this respect.
Rarely is a new technology inherently “disruptive” or “sustaining.” The same technology can enable both.
What matters is the business model and value network in which the new technology is deployed, as that allows you to understand its impact on the competitive dynamics of an industry—that is, whether it is helping sustain the improvement trajectory of the dominant players or disrupting that trajectory by bringing something to the world that does not offer better products to existing customers in established markets, but instead delivers a new value proposition centered around relative simplicity, accessibility, or affordability.
The faulty conclusion around AI’s uniqueness stems from the way Clayton Christensen originally framed the phenomenon of disruption (and misses how Clay revised and improved the initial theory as anomalies, or outcomes the theory couldn’t explain) helped him do so.
In The Innovator’s Dilemma and his initial writings on disruption, Clay referred to “disruptive technologies.”
But the anomaly that emerged in industry after industry was that many new technologies showed up both as (1) innovations that market leaders did adopt to improve their existing products and (2) as innovations that new entrants brought to market that led to them eventually toppling the market leaders.
This realization led Clay to reframe the phrase and, starting with The Innovator’s Solution, call them “disruptive innovations”—because an enabling technology was just one component of what made something “disruptive.”
Anomalies continued to help him refine the theory in successive books and articles. And in May 2013, Heather Staker and I coauthored a white paper with him titled “Is K–12 Blended Learning Disruptive? An introduction to the theory of hybrids.”
The theory of hybrids observed that industries often experience a hybrid stage when they are in the middle of a disruptive transformation. A hybrid, we wrote, is a combination of the new technology with the old technology and therefore represents a sustaining innovation relative to the old. And we showed that we saw it in industry after industry, such that we developed a four-part test to identify whether something was a hybrid:
- It includes both the old and new technology, whereas a pure disruption does not offer the old technology in its full form.
- It targets existing customers, rather than nonconsumers—that is, those whose alternative to using the new technology is nothing at all.
- It tries to do the job of the preexisting technology. As a result, the performance hurdle required to delight the existing customers is quite high because the hybrid must do the job at least as well as the incumbent product on its own, as judged by the original definition of performance. In contrast, companies that succeed at disruptive innovations generally take the capabilities of the new technology as a given and look for markets that will accept the new definition of what’s good.
- And it tends to be less “foolproof” than a disruptive innovation. It does not significantly reduce the level of wealth and/or expertise needed to purchase and operate it.
For example, the automobile industry has developed several hybrid cars along its way to transitioning from gasoline-fueled engines to engines with alternative power sources. That’s because the leading car companies initially wanted the virtues of both internal combustion engines and electricity, so they developed a sustaining innovation—hybrid cars that use both gasoline and electricity.
Hybrid stages have appeared in many other industries, including:
- earth excavators (in the transition from cable-actuated mechanical shovels to hydraulic-powered ones);
- ships (from sail to steam);
- photography (from film to digital);
- injection molding (from large machines supporting large batch sizes to smaller ones fit for shorter run lengths and faster turnarounds);
- retail (from stores to online);
- banking (from branches to mobile).
Industries create hybrids for predictable reasons, including because the business case for the new technology enabler is not compelling at first to industry leaders (see the initial extreme limitations of electric-powered cars, for example), whereas implementing a hybrid as a sustaining innovation allows incumbents to satisfy their best customers.
We enumerated other implications of hybrids in the paper—including anomalies where disruptive innovation seemed not to apply to certain industries—but for this purpose, the central insight is that having a technology that has both a sustaining and disruptive impact on the field is far from unique.
AI isn’t special at all in this regard—which is why we were able to predict that companies like Google and Meta would, unless there was significant managerial intervention, naturally use AI as a sustaining innovation for their business models and customers, but some entrant firms would seek to position themselves disruptively.
It simply makes no sense to ask whether “AI is disruptive or sustaining.” What matters is the use case, business model, and surrounding value network.
Why does all this matter? It helps show which theories can still help us predict the impact of taking certain actions in the age of AI—long before the data about the past materialize to tell us what has actually unfolded before us.
