As a business stakeholder – no matter if you are CxO, middle manager or product owner – you might be confronted with buzzwords like Industry 4.0, Artificial Intelligence, Machine Learning and IoT. Even if you understand them, you often can’t get a clear example showing how the technology would help your business, solve your most pressing challenges, or differentiate you from competitors.
Among the stakeholders you interact with, there might be data specialists: data engineers, data scientists, data architects. These people can transform tons of data into precious insights. The usual problem is to help your two worlds communicate.
As a consultant and business analyst, I have always found myself in this «translator» position. Today, I wanted to share a very simple, yet quite efficient technique I often use in my role as a consultant and business analyst to help data science and business to communicate.
Business stakeholders understand business cases or solutions to relevant challenges, preferably with a price tag attached to it. Data specialists need requirements in terms of wished outcomes.
To allow these groups to come closer, business stakeholders need to understand what solutions are available thanks to technology, and data specialists need to focus on what will help the business make a step forward (be more profitable, leaner, create value, etc.).
In this blog, we will help business stakeholders understand machine learning through examples, much like we would explain complex concepts to fifth graders. This doesn’t mean I associate business stakeholders to kids, but the technique is quite powerful! It might come in handy when your kids want to know what you do at work.
Machine Learning as a superhero: meet «Captain ML»
We use this «superhero» with clients in ideation workshops: after pain-points have been listed and prioritized, we inspire participants by showing them potential applications of machine learning, relating to them as the powers of Captain ML.
If you were a superhero with these powers, what would you do to solve your current challenges?
Predict the future
ML is mostly known for its ability to «predict» as it anticipates scenarios based on data history, patterns and their correlation to business events. It allows the business to take measures in a timely manner. A very simplistic example is a feature showing the predicted train occupation when you book a ticket. The output helps the user book the best train and uses several inputs to be able to make a prediction (past occupation, weather, time of the day, etc.).
Machines are certainly able to react faster than humans and can help automate processes. Not only operations (like processing audio and transcribing it to text) can be done better and faster, but algorithms can be used to automate more complex processes too, increasing the speed of human operations even more. Think of a drive-through bot that would help an agent while a client orders food. The machine can translate the conversation to text (speech-to-text) and then to an actual order (correlation to available products) by understanding even people with weird accents or speaking very fast. This helps the agent do his job better and faster.
There are several use cases in operational efficiency that can be solved by voice bot technologies.
Thanks to deep learning and transformer models, it is nowadays possible for machines to be infused with something similar to human creativity and generate original content, like an artist would do. The most famous applications of these models fall in the category «fake news» (you might have seen videos of Obama spreading fake news, or Mona Lisa dancing in synch with Gandhi and Einstein).
However, there are also very useful applications for this: generating object photos out of a pencil sketch could help architects or designers speed up the rendering process. Turning a bunch of roughly written bullet points into a properly written e-mail could help professionals be more efficient (spending less time writing e-mails).
Imagine being able to multiply your hands and eyes with sensors capable of detecting noises, vibrations, or other indicators. In the field of predictive maintenance, that's exactly the role of IoT (Internet of Things) sensors, which capture and store vibration data (e.g. of a roller-coaster). An algorithm then correlates anomalies (strange patterns of vibrations) to potential outages and triggers an alarm which allows to plan maintenance and prevent failures (e.g. preventing a roller-coaster to break with people on it ...).
Think of this as the ability to understand objects or their characteristics when a human would not be able to by only looking at them. For example, a flying drone can recognize and pick only ripe apples out of a tree by using a camera and ML to recognize patterns in the skin, color or shape of the fruit. Insurance companies can benefit from algorithms which recognize damages in pictures and help automate claims processing.
ML can assist you in real-time almost like reading your mind and predict the next best action. As an example, think of the autocomplete function of Gmail, or any recommendation engine such as Netflix’ for example: by using some data as «context» (e.g., an e-mail history), the algorithm can anticipate what you would write and help you complete your sentences like it was reading your mind.
The same approach could be applied to banking and financial services by building a product recommendation engine to advise clients about relevant features on their e-banking or propose investment products they could be interested in.
How can ML super powers help your business?
If you have found this way of explaining ML effective, let us know! We’re looking forward to talking about your challenges and how we can help you boost your business with machine learning.