I’m sure we are all familiar with the incredible power of futuristic computers in sci-fi movies. I myself have always been fond of their vision for Artificial intelligence (AI). These beautiful works of fiction have peaked my interest towards computing technology for many years, and clearly I’m not the only one. Even one of the world’s most well know companies, Google, is actively pursuing their goal of creating the ‘Star Trek super computer’ (Star Trek computer John Giannandrea interview).
A few years ago I was reading into philosophical papers about the potential of AI. I was intrigued by the idea that with the right algorithm, we should be able to accurately predict the future. Being able to do this certainly sound promising, however there are some strings attached. This theory is based on many assumptions about the way we will be able to collect and process data. However, the first steps seem to have already been taken and I believe we should be very enthusiastic of all the possibilities this might bring.
As most of us are already aware of, AI can already be found all around us. Try shouting out ‘Hey Siri’ or ‘Ok Google’ at work tomorrow, I’m sure you will be greeted by many phones asking you to make requests of them. As AI is not a new concept, a lot of technology in our homes and workplaces are filled with intelligence trying to make our lives more easy and efficient.
Describing the true meaning of AI can be difficult as the perceived definition of AI changes as our insight of technology progresses. AI consist of a very broad spectrum of algorithms capable of replicating the cognitive functions of humans. In turn these can be very simple, such as detecting heat or cold. Or very complex ones, such as reading and responding to emotions.
What do we perceive as intelligent?
For the purpose of this blog, and to avoid a very complex string of mental functions, we shall define intelligence as “the ability to react on the environment based on memory and experience”. The addition of the word Artificial tells us that the intelligence is computer based. Within the spectrum of AI we have defined two terms which guide our perception of AI, we call this ‘strong’ and ‘narrow’ AI.
Strong AI is described as an AI which is able to reason, solve problems and might even be self-aware. Basically depicting human behavior. This type of AI is very hard to create as current technology is not advanced enough to provide this level of intelligence. There are a few AI’s which have many combined types of ‘narrow AI’, which creates potential for a strong AI. In my opinion this suggests that a strong AI is a very complex combination of narrow AI systems.
Narrow AI is able to mimic behavior which seems intelligent but actually isn’t. Before we mentioned programs like Siri and Google. These systems are great examples of narrow AI as they work within a limited perimeter of data and have no self-awareness. We are easily confronted with the limited capabilities of this AI as they often tell us “I’m sorry, I don’t understand”.
Let’s educate our computers
To create a robust type of AI which is able to work autonomously, it needs to be able to learn.
A big leap for AI research is the method of “Machine Learning”. Machine learning refers to the ability of a computer to teach itself the most efficient way of completing a task. The computer is able to do this by continuously analyzing performance data. Machine Learning is extremely important when attempting to create a strong AI. Being able to learn from experiences is one of the key aspects of what we see as intelligence.
Another term which is often prompted nowadays is ‘Deep Learning’. Deep Learning is a method which makes use of artificial neural networks. These neural networks are comprised of very complex algorithms which are able to determine correlations between input and output. Basically this means the neural network is able to calculate the best possible answer based on the given input. Basic forms of deep learning are already widely spread across the digital landscape. Techniques such as speech-, text-, image- and biorecognition make use of artificial neural networks.
How should i start with AI?
The best way to start with AI is to look at the processes which are running at your organization. Many processes have not been changed for many years and seem to be embedded into the organization culture. To my experience this proves that the core process is very well constructed and that there is no direct need to change the way we do things. I’m however certain that many of these processes could benefit greatly from AI. Any existing process has a set decision structure behind it. The most challenging aspect of creating an AI is figuring out this structure. Right now you might realize that a lot of these structures are already present at your organization. In this you are correct. The only things which rests is transcoding this structure to an AI application.
AI and Apollo
Many of processes we run daily could be assisted using algorithms. The algorithms can be made very easy to use in the form of a desktop or mobile application. Imagine visualizing Big Data and IoT (Internet of Things) and the automation of your organizations processes through use of narrow AI. At Apollo Journey we work together with our customers to find and enhance their business decision making processes using AI. We involve technology to the level which neatly fits in your organization culture and enables the efficiency gains technology can provide.
Apollo is here to help you discover the potential of technology.
EGI CUIJTEN, MANAGING DIRECTOR