Imagine you’re waiting for Monday morning’s meeting to be kicked off by your new colleague, who no one has yet met but who already has ‘The Brain’ as a nickname. The door opens and a friendly greeting robot enters.
Even though such a situation may well occur one day, the aim of this article is to shed light on the term ‘Artificial Intelligence’, or ‘AI’, and take a short look at its history. It then tackles the scope of the term AI itself and, most importantly, relevant business scenarios.
Is AI really a new phenomenon?
Contrary to popular belief, the AI movement has been around for a long time. It started in 1956 at Dartmouth College with an academic conference on the topic. As evidence of practicality was poor, it lacked funding in the 1970’s, but regained momentum again during the 1980’s thanks to the power of digital computers. Its progress then slowed again before beginning its constant rise throughout the 21st century.
Examples everyone will be familiar with are Deep Blue beating Kasparow at chess in 1996 and, more recently, AlphaGo beating Lee Sedol in Go – a game far more complicated than chess.
Too many alleged A.K.A.’s
The term AI is often confused with others: intelligent systems, analytics, knowledge-based systems, computational intelligence, context awareness and perhaps the most well-known today, machine learning.
The two levels of AI
It is important to stress there are two levels of AI: ‘weak’ or ‘narrow’ AI which focuses on one narrow task, and ‘strong’ or ‘general’ AI intelligence that could successfully perform any intellectual task that a human being can. The latter is generally considered to be still a long time away. As the focus of this article will mainly be short to midterm business scenarios, here mainly, the first category will be relevant.
Making AI work for businesses
Before addressing the most relevant question for any business, how does AI pay off, there are some general aspects to consider:
First, AI shouldn’t be thought of as a feature that is simply added to a requirement list. If AI doesn’t add value to the business, it’s barely worth it. In other words, it’s vital to define a relevant use case. A further requirement is the data needed to feed AI and how to administrate it (cloud or local, third party solution or creation of an own ecosystem). Then, staff needs to be trained accordingly and company culture adjusted to embrace this change – something which will also be reflected in the organizational structure.
However, it’s not always about reinventing the wheel, either.
A good example of using an existing technology in an AI context is a solution by Fujitsu where the computer analysis of a parking lot video feed provides real time data for a parking guidance system. The system is comparatively easy to set up, yet highly effective.
Searching for further AI opportunities, retail is a good place to start. Amazon Prime Air Delivery started using drones in December 2016. Another example is Domino Delivery Robots that were launched in March 2017. Both handle the so-called ‘last mile delivery’ and can deal with absent recipients.
At the beginning of the customer journey, retailers can use data to predict what customers want - even before the customers know themselves. Think of what facial recognition, machine learning, computer vision and natural language can do: Robots welcoming customers and, if required, acting as personal shop assistants. Other robots deal with inventory and refill shelves whilst smart price tags optimize prices and smart screens display personalized offers.
The key benefits of such processes are: more flexibility, individual delivery as well as faster delivery services and lower logistics costs.
Since the performance of all business areas would go beyond the scope of this article, manufacturing and healthcare are only mentioned as branches here.
Value through AI across the value chain
Instead, let's take a look at some exemplary links in the value chain for inspiration on how AI can add value.
At the planning stage of a product, self-learning software can develop solutions that are too complex for humans to come up with by themselves, such as new construction methods for bridges or airplanes. In such processes AI and human intelligence complement each other, as both are doing what they are good at: AI figures out new complex patterns and the human brain takes decisions.
An inspiring example of how AI can boost a whole bridge building process is the MX3D project. Collaboration between Artist Joris Laarmans and American company Autodesk, the bridge was planned and built entirely by machine, designed by a generative design technology and built by robots from 3D printed materials.
Having looked at many useful applications, the conclusion is that AI is hope not hype when a valid use case for value creation exists as well as data and organizational infrastructure.
To end this article, some suggestions that have proven to be working guidelines in the author’s work experience.
AI: the dos and don’ts
- Make products people actually want, instead of trying to make people buy your product
- Inject your product with a healthy dose of empathy
- Think twice before trying to build your own AI engine. Instead, see if you can start with an already existing one and build from there
- Don’t just throw something out there simply because you want to do something with AI
- Don’t expect AI to do wonders if you have no data to feed it
- Don’t forget about formulating guiding principles