Let’s start at the beginning.
Artificial intelligence refers to the ability of a computer (or a robot controlled by a computer) to mimic behaviour we would typically associate with human intelligence such as mimic an action or solve problems. Think of AI as the entire house (a blanket term that covers all intelligent machines).
The kitchen (or where all the magic happens) is machine learning. ML gives the computer the ability to learn from the past – put simply, a machine “remembers” what had happened and extracts how best to “act” in the present based on historical outcomes (without being told how to act at all). Now, I know that sounds like we’re five minutes away from accidentally creating VIKI (I, Robot is a must watch for any aspiring nerds) but fortunately (or unfortunately) the processing power required for complex decisions does not exist… yet.
The interesting bits come about when you tailor the algorithm (recipe) to make more accurate decisions for a specific problem – the same way that varying the amounts of flour, liquid and sugar can be used to create a variety of treats. Each of these algorithms are limited to solving one very specific type of problem based on very specific input data – if you asked a machine what flavour ice cream your dog might like (after having input data on what ice cream you enjoy), the machine might suggest your dog would enjoy chocolate (which would annoy your dog because, like all dogs, she loves strawberry).
Speaking of, we also need to talk about where machines lack intelligence. All these algorithms are stored in a virtual prison (which can be dissected into ones and zeros). Unfortunately, we need these algorithms to make sense in our physical world of cells, atoms and the Shire. This leads to some issues with machines forming connections where there are simply no connections to be made. A machine could (and certainly would) develop an equation relating the cost of oranges in Johannesburg today to the amount of snowfall in Tokyo three years ago. A human understands that although, mathematically, a relationship between those two events is possible, it’s highly improbable that they are related. This is where we come in.
The Afrobots @ Afrobotix are embracing both human and machine intelligence to create tools that could never exist without the pair. Using these two types of intelligence together in data science is called augmented analytics. Machine learning is used to compute and execute tasks (that would take humans months) in a matter of seconds. Human intelligence is used to guide the machine to a logical, probable solution (so the machine won’t be left to compute for months). The confluence of machine and human intelligence enables problem solving that has never been achieved before. Augmented analytics is the key to discover possibilities.
Follow your curiosity,
The Afrobots @ Afrobotix.