Putting AI Into Perspective
Author: Koen Wijnen
The Pivotal Role of Artificial Intelligence in
How We Create and Experience
How come that certain companies seem to thrive in the so-called Data or Digital Economy by experiencing exponential growth, while others are losing momentum and see their dominant market position threatened?
At university, my strategy professor encouraged me to read Michael Porter’s book entitled ‘Competitive Advantage’. Michael Porter, who is a Harvard Business School professor, wrote it in 1985. A competitive advantage is what enables a company to provide superior value to its customers as compared to its competitors. Porter identified three generic strategies to achieve competitive advantage: Cost leadership (lowest price), differentiation (highest quality of experience) and focus (global versus niche market).
After researching the causal relationship between strategy and financial performance of hundreds of companies, Porter concluded that given the state of technology in 1985, the three strategies should be considered mutually exclusive. Or in his own words: “Avoid getting Stuck in the Middle”.
But what would happen, if technological progress would cause the boundaries between the three generic strategies to become blurry? These boundaries were once considered to be strict and absolute. In that case, the rules of the ‘competitive’ game would change fundamentally, and dominant market positions once considered a simple fact of life could now easily be threatened by new players entering the market.
Indeed, that’s exactly what has been happening at an accelerating pace over the last three decades, and data related technologies such as Artificial Intelligence and the Internet of Things seem to play an increasingly important role in fueling this process.
Delivering on what customers value
The key question thus becomes: How to delight (quality of experience) your customer by providing an affordable (price) hyper-personalized (focus) product or service.
While looking for an example of how this might work, the case of Netflix seemed to be an obvious one. But having a background in manufacturing and logistics, I was particularly interested in examining a product-related case instead of a service-related one.
Let’s select a quite intriguing example: Adidas’ SPEEDFACTORY.
The iconic sportswear brand’s SPEEDFACTORY focuses on “Using advanced digital technology powered by sports data to create the most precise performance footwear offering the perfect fit, movement and comfort.” It is Adidas’ interpretation and implementation of a highly flexible and localized manufacturing and distribution system aiming at a radical reduction of lead times. The system is driven by an open source community of co-creators, all made possible by leveraging the opportunities offered by digital technology. The main idea being that there should be a “perfect fit” with the runner’s preferences.
Let’s assume that these preferences are determined by personality (looking for performance or comfort) and context (a city run versus a run in the countryside).
Adidas might opt for embedding IoT-sensors that are able to measure temperature, humidity and a pressure profile while running. The sensors might convey the captured data to an app running on an ordinary smartwatch, which might also collect some biometric data and a GPS signal.
Turning AI into a key value driver
Subsequently, AI becomes part of the equation. AI comes in different shapes and sizes. An important distinction is the one between Prediction Machines and Decision Machines.
A Prediction Machine’s objective is to maximize prediction accuracy, by for example relying on recent significant progress in domains such as machine learning and deep learning.
In the Adidas case: the prediction whether the runner focuses on performance, comfort or strikes a balance between the two. Is the runner using it mainly for city runs or for relaxing runs in the countryside?
A Decision Machine, on the other hand, combines the prediction(s) with judgment on what matters in order to choose an action. This action in turn then leads to an outcome. Having accurately predicted the runner’s preferences, Adidas can now offer and manufacture at low risk a hyper-personalized running shoe (Selecting an action.) with the desired outcome (a purchase).
Moreover, a new feedback loop is initiated enabling Adidas to learn whether the new pair of shoes is indeed performing ‘better’, and more importantly, how it can further improve its offering in matching the customer’s needs.
This example nicely illustrates how the interplay between AI-driven Prediction Machines, Decision Machines and the Internet of Things can help to create an accelerating feedback looping mechanism, accumulating ‘sustainable’ competitive advantage during every iteration.
Why AI needs to become emotionally intelligent
I would like to advocate that we distinguish between Cognitive and Behavioural AI. Behavioural AI aims at predicting human preferences and the mechanisms of human judgment when taking decisions. Combining domain knowledge about human psychology and traditional machine learning helps to keep requirements in terms of data volume, algorithmic complexity and compute effort feasible.
In the Adidas example, Behavioural AI would most likely be used to get a better understanding of whether the customer prefers performance, comfort or an intermediate combination of both, and subsequently simulate his or her response to different customized designs in order to pick the ‘best’ one. Cognitive AI in turn could be applied to translate sensor and GPS data into usage profiles (A city run versus a run in the countryside.) or to build self-learning planning algorithms to make the SPEEDFACTORY’s operations even more efficient.
Interested in AI applications?
Contact us for more information on how Artificial Intelligence solutions can mean something for your organization!