The No-BS Guide to AI - Part 1: What is Artificial Intelligence?
Mini-series: How to increase your competitive advantage, cut costs and create added value through the use of AI.
Over the course of the past years Artificial Intelligence (AI) has regained interest in the broader public and more specifically the interest of businesses. Most people associate AI with Killer Robots from the future or sneaky algorithms from Facebook trying to keep you scrolling and liking. But this is not what Artificial Intelligence is about and not where its value lies for Business.
What we aim to do in this mini-series about AI is to help you understand what the business relevance is for AI. We will talk about the elements involved, educate you to spot opportunities in your organization to apply AI in order to create value and overcome the trivial fears of AI taking over job or society as whole.
The buildup of the mini-series will be as follows:
Part 1: What is Artificial Intelligence?
An introduction on the subject to familiarize you with the terminology and background of A.I.
Part 2: Machine & Deep learning
How we can teach computers how to think and make this synergetic with our processes and strategies.
Part 3: Natural Language Processing
Making computers understand our language and communicate with us in ways that feel natural and intuitive.
Part 4: Robotics in business
On how our physical bodies limit our productivity and the role robotics can play in optimization and efficiency.
Part 5: Implications on Business & Society
How in time we will shift from narrow to general AI and what this means for us as humans and businessmen.
Part 1: What is Artificial Intelligence?
Defining AI
Artificial Intelligence is not easy to define, the two words individually already create enough room for interpretation to have long lasting discussions. Artificial could be defined as “something that doesn’t occur naturally.” Intelligence could be defined as “the ability to solve problems.” Putting them together one could create the definition of “Activities that are hard for computers to do as opposed to simpler activities computers do routinely today” (e.g. calculations, file sharing and Microsoft Paint).
More important than trying to make a scientifically correct definition, the symbolic value of AI should be discussed. In the field of AI there is an important distinction to make between “narrow AI” and “general AI.” Narrow AI is a machine-based system aimed at addressing a specific problem (e.g. playing chess). General AI, by contrast, is more of a research topic on how machines could solve a variety of problems on their own, like humans can. Though most people think of AI (or fear it) in the general definition, the examples and applications of AI to date are all forms of Narrow AI. General AI is regarded to be at least decades away from becoming a reality, but more on this in the last module of this mini-series.
A brief history
Artificial intelligence in itself isn’t something new. Though the attention it has gotten in the past few years, the birth of AI was in the mid fifties by renowned names such as Marvin Minsky and Alan Turing (yep, the guy from the test). In these starting years AI was aimed at figuring out how we could make a computer learn as apposed to solely executing simple commands based on simple rules.
In the early 1960s a MIT student called James Slagle wrote the first program to have integrated symbolic expressions which initiated the first AI wave. Sagle’s program and its successors became the corner stone of modern AI elements such as neural networks, but we won’t go deeper into this for the sake of readability.
The most important take-aways are that 1) AI isn’t something new under the sun. People have gained and lost interest in AI due to failed over-hyped promises and ignorance of the mainstream media on the subject multiple times before. Though we feel this isn’t the case in the current hype or trend because of the technological (hardware) advancements that have recently been made and more importantly; the business value it is starting to create. This last argument should guarantee the adoption of AI in mainstream business and education. 2) AI is primarily a tool for knowledge, process and business optimization, not world domination and killer robots. Though Hollywood is partly responsible for the rebirth of interest in the fields of AI, it should not be mistaken for the true intentions of AI research and application.
A more in depth view on the history on AI can be found here and here.
Where are we now?
Depending on when you define a software application or algorithm Artificially Intelligent, modern examples of AI can already be found all around us. We know the cliché examples of money-making Wall Street algorithms, autonomous driving cars and chatbots who help us get information we need, it has already more hidden heroes than we think.
There are for instance callcenters that let AI programs monitor their conversations in order to suggest the right tone of voice and choice of words to satisfy and comfort customers. Or virtual assistants that help us plan our meetings and go beyond simple command and execute structures.
In businesses there are four ways AI can be used:
- As a tool
- As an assistant
- As a peer
- As a manager
AI as a tool
Most known and applied are ways in which AI is used as a tool. For instance in large complex databases AI algorithms can search and process these volumes more efficiently and with less error than humans can. An example of this is the Google search engine. Google indexes websites based on their content and ranks them in search results based on their relevance for the terms used in the search. Though Google does the legwork for you and way faster than we could ever as humans, we as humans still need to select the best result and filter out the rest.
AI as an assistant
Over the last years we have seen a rise in various AI assistants. From the iPhone’s Siri to the newer Amazon Alexa that is purely speech-based. Though these assistants are primarily used as tools to set reminders and order an Uber, the underlying software, if it has access to your personal data, can also think autonomously without your command. It can tell you that you might want to take an umbrella with you when rain is expected or that you might want to leave earlier for an appointment due to heavy traffic. These are proactive exercises that are pre-programmed, but seem nonetheless more intelligent than the pure command based AI tools.
AI as a peer
When we speak about AI as your peer at work, we don’t mean the general form of AI in which a robot with human intelligence sits next to you at the lunch table. The cases of AI as a peer are found in AI applications that run autonomously on the background without the need of human command. The system only contacts its peers when it is confronted with a request, process or prompt that it doesn’t recognize or hasn’t been trained to perform. An example in this field is fraud detection. AI algorithms monitor millions of transactions every second and are trained to seek out suspicious activity. These programs run on their own and train themselves at becoming better detectives of fraud, but once a suspicious transaction has been found, it notifies a human peer about the situation who in turn will use the suggested data to figure out whether or not the anomaly is fraudulent or not.
In these and other cases, AI programs are operating autonomously without human command and acts as a fellow employee within the organization.
AI as a manager
Don’t worry, we’re not talking about robots becoming your boss. When AI acts as a manager it’s because it’s trying to accomplish tasks that for humans would otherwise be too complex to delegate. You can think of traffic lights as an example of this. To optimize the flow of traffic certain conditional rules are given to a system that then monitors how to best guide traffic through intersections.
One of the main areas in which we see this form of AI already taking place is in logistics. Amazon recently introduced its Mechanical Turk that in essence allows people to participate in the Amazon operations but performing small tasks that AI programs appoint to them based on their skill set, location and or experience. There are no humans here delegating tasks to other humans, it’s a smart network of AI peers communicating with AI managers to optimize the flow of work through an open platform.
A Collective Intelligence of Humans and Computers
When we look at the true value of AI, we hope you too will see that it lies in balancing the best of human capabilities and the best of computational power. AI isn’t here to take over your job, it’s here to make your job easier and your results better. By finding this optimal balance we can start to grow and exploit the collective intelligence of our organizations.
This might sound like a fuzzy definition, but a collective intelligence is nothing new. When two people or more meet to discuss a topic, it’s already a form of a “collective” coming together to use their intellect to solve a problem. In the current market we still rely too heavily on human-human interaction in using the Collective Intelligence of our organization.
The problem with that is that we can only physically be in one place at once, sleep 8 hours every night and maintain a social life. When we start to implement AI programs into our daily business processes we can start to optimize the collective intelligence within an organization beyond humans and become more agile and adaptive to market changes, giving us a competitive advantage and increase in productivity and innovation.
Next up: Part 2
In the next part of this mini-series we will take a deeper look at how computers “learn” and how we can teach them to understand patterns, processes and in this make our businesses more efficient and effective.