Published 17. Aug. 2018
Unlocking the Potential of AI - An Executive Guide
Business Intelligence and Analytics agency Fairpeople sharing their insights on how data can be used to drive exponential business growth.
One of the most prolific buzz terms of recent times is undoubtably “Artificial Intelligence”. There is a plethora of articles, events and job titles popping up around it – just check your LinkedIn feed. It can be confusing and intimidating. And if you are a business leader, the fear of missing out on the apparent benefits of investing in AI adds to the anxiety.
What is Artificial Intelligence?
As any decent sci-fi fan can tell you, artificial intelligence is the ability of machines to perform general cognitive functions we typically associate with humans. This includes perceiving, interacting and most interestingly learning.
The ability to learn – to find patterns and connections – from experience and apply them for better problem-solving in the future is done seamlessly by the human mind. We tend to take this ability for granted: we know what pressure to apply when picking up a coffee mug, how to drive on a busy highway, or quickly realize common tricks in crosswords puzzles. Simple, right? AI researchers would tell you otherwise.
So Where Do We Stand Today – What is Aspirational vs Achievable?
We established that AI means performing general cognitive functions by machines at par with humans. The key word here is general. While you as a human have no trouble navigating a busy intersection while giving financial advice to your parents on your hands-free, imagine how infinitely more sophisticated Siri or a self-driving Uber needs to become to do both of those things at the same time. This is called Artificial General Intelligence (AGI) and it is still something to look forward to (i.e. it is aspirational).
More achievable however is Narrow Artificial Intelligence. As the name entails, it refers to a single task being performed by machines at the level of (or beyond) that of people. Narrow AI is where actual progress is being made and which has immediate applications.
These applications can be grouped into five buckets: Robotics, Computer Vision, Virtual Agents (chatbots), Language Processing and Data Insights.
Each of these groups leverages specialized approaches to crack a specific problem and there is limited overlap among them (i.e. a robot which can play soccer will not be able to recognize trends in the stock market). Within the application groups however there are rapid advances and growing number of use cases with immediate implications.
This is All great, But Why Should Businesses Outside Technology Fields Care?
In one word: insights. Bucket number five of narrow AI is the reason why leaders in any industry should consider investing.
Somewhat simplified, the job of executives is to make decisions (e.g. what products should we invest in, how to set prices, what stock levels to maintain, how to position our brands). The quality of those decisions directly impacts the success of their company, yet even leading global organizations mostly on intuition and experience to make those decisions.
There are however proven methods to generate robust insights from data to support the decision-making process in a scientific manner. Data driven insights replace intuition with facts and shorten the decision-making process. They bring what we call the “three-R-advantage”: they are rapid, reliable & repeatable.
How Exactly Can Business Leverage AI for Better Decision-making?
So, we established that there are achievable use-cases wishing Narrow AI and that Data Insights are the subset of these which should be considered by executives regardless of industry. But what are the problems that can be solved by data driven insights?
The list below illustrates six of the most common problem types AI can solve by generating Data Insights using approaches within the Narrow AI framework.
This list is not exhaustive however it does account for over 80% of use cases in Retail, CPG, Professional services and B2B Manufacturing.
What About Machine Learning and Deep Learning?
As if AI is not confusing enough, some other terms keep popping in the same context. Most common are Machine Learning and Deep Learning. Here is a simple framework to help structure the thinking and understand the relationship between these terms.
We can thus think of Machine learning as the set of algorithms and data processing approaches used to enable AI. And Deep Learning is a subset of Machine Learning with some specific and advanced types of techniques. It is then reasonable to say that Deep Learning is as closest we are to achieving AI.
How to Get Started?
Getting on a roadmap towards building data and analytics driven organization is a must. The most practical way to do that is to identify the quick-wins and commission multiple, small scale proof-of-concepts. Then evaluate results and the potential for scaling before making decisions to invest.
However, to have the proper execution of roadmaps and POCs, organizations first need to have the right talent in place – collectively referred to as Data Scientists. To integrate a capable data science team into their organization, business leaders need to answer the old familiar question “do we make, or do we buy”?
Some companies are opting to invest in their own data science departments while others are starting up with consultants and outside experts. Which one is the better alternative often depends on the technical maturity and culture of the company.
Data, Data, Data!
There is no AI without data. It is the fuel that makes all this possible weather gathered through sensors, social media or retail loyalty cards. Organizations need to recognize data for what it is: an increasingly precious resource that needs to be treated systemically and invested in accordingly. The rise of the Chief Data Officer in some of the most innovative companies around the world is not by accident.
What to Expect?
The economics of AI are clear: access will become cheaper, quality will increase and use cases will evolve. Some say we are entering the age of prediction in which AI systems will make accurate predictions for just about anything – retail prices, stock values, unemployment, weather. And acting on those predictions will be as trivial as for example using Google Maps in the present.
Predictions will matter most to companies. Those organizations which invest in the right skills, technologies and data will likely emerge as winners.
Head of Advanced Analytics at FairPeople Denmark