The Real Promise of AI Is Innovation — Not Optimization
Moving from tactical to strategic use of AI
Oct 17 · 7 min read
Let me start right off the bat: many incumbent organizations risk missing out on the strategic opportunity of AI — instead focusing on tactical optimization initiatives.
General purpose technologies like AI have the potential to change how companies operate and create value, disrupting entire industries along the way. History proves however that new paradigms are needed for new technologies to come to fruition. The good news is for AI that these paradigms are already emerging— they are just unevenly distributed.
Using AI for both short term gains and long term strategic value is not trivial and requires strong leadership skills: combining vision with experimentation and aiming for both optimization and strategic plays.
This article explains how to instill a strategic compass in AI transformation.
The game changing nature of AI: breaking productivity constraints
Algorithmic decision making, in the sense of narrow-AI to automate or support key processes, enables companies to break through 100-years old limits in productivity inherent to the established corporation as we know it.
Specialization of labor created organizational silo’s. These silo’s create boundaries in expertise, information flow and data. Increasing scale and scope put a limit on what is “manageable” by humans, and puts a cap on learning. The result is that traditional organizations suffer from diminishing returns. Beyond a point, bigger just ain’t better.
For AI, this is actually reverse: algorithms can be scaled virtually infinitely and improve with size (i.e. more data and feedback loops).
Scalability and learning flip the traditional scale curve of firms.
This fundamental dynamic creates a new breed of firms, with a step-change in productivity due to enhanced scale, scope and learning power. Certainly, digitization is the precursor. But AI creates a fundamentally new dimension by making cognitive tasks scalable while also improving them in a learning loop — rather than one that diminishes in size. The implication is not only automation of cognitive tasks at scale, but using AI to create new levels of scale.
The writing is on the wall. Ant Financial, the financial services arm of Alibaba, has a broader offering of services than incumbent Bank of America. Yet, Ant Financial’s operational efficiency is a factor >200 higher, measured in number of employees per customer.
So the million dollar question for incumbent organizations: how to harness the power of this disruptive force?
Innovation requires odd mix of vision and experimentation
Artificial Intelligence is a general purpose technology with a fluid future — it is not yet entirely clear how disruption will play out at large for specific companies and sectors. That makes it hard for managers to develop a sharp vision and strategy for how to shape their business with AI. It makes sense to just start, and learn as you go. A wait-and-see approach only postpones an inevitable transformation.
The inconvenient truth for managers is however that a bias towards action is necessary but no guarantee for success. We have history to prove it.
When the electric motor was invented in the 19th century, the dominant use was to replace steam engines as a central power source in factories. It took decades to reconfigure manufacturing plants and redesign machines to exploit the benefits of locally applied mechanical power.
The first digital camera (Kodak, 1975)
Fast forward to 1975. When 24-year old Steven Sasson invented digital photography, Kodak wasn’t exactly enthusiastic.
They were convinced that no one would ever want to look at their pictures on a television set. Print had been with us for over 100 years, no one was complaining about prints, they were very inexpensive, and so why would anyone want to look at their picture on a television set?
A more recent example is the emergence of e-commerce. When the internet arrived, many retailers that went bust were initially the first to move online. They viewed online as just another distribution channel. Successful e-commerce players realized that the internet provided more than unlimited (digital) shelf space: easy experimentation (A/B testing), removing friction from the user buying experience, building customer relationships and recurring revenues, etc. But most of all: instead of a having a captive market based on physical presence, the internet creates winner-takes-all dynamics in single categories.
Concluding, it takes both vision and experimentation to adopt new technologies for business innovation. For AI, this means it is not enough to launch a transformation program and identify a few use cases. In a world where the new technology (AI) will become a commodity, it will be the most creative and agile firms that continue to look for new ways of value creation that thrive in the long run.
A new era of business model innovation has arrived: using AI to break traditional industry boundaries and build new revenues on top of optimizing old ones.
Let AI scale your business, rather than letting your business scale AI.
Optimize AND innovate
AI-powered innovation happens along three horizons which are typically increasingly risky in nature, but also potentially more rewarding.
Three horizons of AI-powered innovation.
The first horizon is to deploy and scale AI to drive enhanced effectiveness and efficiency of processes that can be supported or fully automated with algorithms. For example to optimize pricing or apply predictive maintenance. In this stage, companies typically focus on getting algorithms in production, implemented and scaled. This effectively optimizes the current business model.
The second horizon is to let AI scale the business, rather than letting the business scale AI. The automation of human tasks removes prevailing barriers to scale. The most value creating processes, when automated with AI, can become the core capability for new business models. Consider for instance banks that automate their loan underwriting process. Once humans are taken out of the loop, that creates an opportunity for a new digital lending platform without constraints on scale or market boundaries (e.g. geographically).
The third horizon consists of entirely new types of activities, with new products or services, potentially for new customers and markets. Consider energy companies that have data on consumption patterns which can be used to offer personalized energy management or equipment maintenance.
The far majority of AI activity in incumbent organizations takes place in horizon 1 (source: my own experience). And make no mistake, a lot of value is to be captured still. There is however a lurking issue. Four of them, actually.
First, AI initiatives could end-up as tactical gains, being in a sense no different from any other business optimization initiative. The AI-wash can obscure this fact and make it seem strategic for a while. No technology has ever created a strategic advantage by itself, without the proper business model to create defensible moats (which includes patents, of course) .
Second, once the ball is rolling on AI a flurry of initiatives may arise. Getting AI solutions in production and implemented is however a huge challenge. Fragmented efforts may result in a whole lot of nothing.
Third, the tactical character combined with fragmentation reduces the vision and nourishment the most promising areas require to come to fruition. The MVP in most cases gets you 20% of the potential. The remaining 80% requires sustained commitment and focus to realize.
Fourth, while horizons 2 and 3 may require more bold moves beyond the traditional business, they represent both exponential gains and threats. The latter because value chains may fundamentally reconfigure as scale boundaries diminish. And someone else may get there first.
The resulting status quo is that the bottom-up initiatives do not live up to their top-down potential.
AI Change Leaders: rise to the challenge
The humble task for AI change leaders is to create a vision and transformation approach that delivers both short term value while creating stepping stones towards strategic initiatives. Periodically raise your vantage point and calibrate your AI priorities. My advise would be:
Make sure to get the short term right (horizon 1). Create focus on the biggest opportunities and drive the organization to create scale on those. That includes aiming for adjacent opportunities that can leverage the same data and models. This will close the gap between top-down value pools and bottom-up results and create impetus for transformation.
While focusing on execution and impact, continue to shift upwards by defining a vision on how the horizon 1 initiatives drive a competitive edge. This helps instill a sense of purpose to the organization that is bigger than tactics. For instance, AI-powered drug discovery is at the core of a pharma company, while predictive pricing of the resulting products is valuable but more tactical. Your vision also creates a reason for continued development to get the first 20% to the potential 100%.
With horizon 1 in control on both execution and vision, in parallel develop ideas and potentially first steps on horizon 2 and 3. Your horizon 1 initiatives actually may be stepping stones, as in the example of AI-powered loan approvals. It requires you to think in terms of the broader networks your firm operates in (customers, suppliers, end-customers, etc.) to determine where the unique data of your firm can add value.
No easy task, but luck favors the prepared.
“You can’t connect the dots looking forward; you can only connect them looking backwards. So you have to trust that the dots will somehow connect in your future. You have to trust in something — your gut, destiny, life, karma, whatever. This approach has never let me down, and it has made all the difference in my life. “ Steve Jobs