Let’s face it, the terms “artificial intelligence” and “machine learning” have become ubiquitous – as have the myriad of companies and organizations that claim they have harnessed these capabilities to the benefit of business and society. I happen to work for a company that is at the forefront of the AI revolution. At IBM we use the word “cognitive” to describe a solution, system, or technology that leverages AI. Innovative companies are exploring ways they can leverage cognitive systems to provide insights that accelerate business value to their customers, partners, and suppliers.
There are scores of articles and blogs that have outlined the incredible benefits of “cognitive systems”, but the question that often goes unanswered is what makes a system cognitive? In my line of work, it seems to be a term that is thrown around quite a bit. Label it “cognitive” and it garners interest. “Cognitive” is transformational and promises greater impact and a higher price tag. It is for this reason that we need to come up with a definition of the term. Once defined you will know when the term has been used properly – and when it has not. This newfound understanding will serve you well as you navigate the current and future AI landscape. So here goes – a cognitive system is marked by its ability to understand, reason, learn and interact.
Understand. A cognitive system can understand vast amounts of information in all forms – the same way that human beings do. Picture a wearable such as an Apple Watch or a Fitbit that collects data about your activity level, sleep quality, and caloric intake. A cognitive system can gather all that data from you as an individual, combine it with other systems of record and those of a greater population, and then build a model that understands the correlation between your daily activities and your overall health.
Reason. A cognitive system can grasp underlying concepts, form hypotheses, and provide inferences to extract ideas and recommendations. Imagine a solution for marketers that recommends prioritized target audiences based on key behavioral drivers culled from vast amounts of historical and real-time data. Other recommendations could include cross-sell/up-sell opportunities and next best action.
Learn. A cognitive system connects each data point, interaction and outcome to help continuously sharpen its expertise. It literally learns by experience; that is, based on specific training, the system will adapt to new or unexpected events. I’m a big fantasy football fan. Imagine a cognitive system that sorts through a wealth of data: football statistics, news reports, social media comments, weather reports, and more to provide real-time insights that change based on future events and results.
Interact. A cognitive system can see, talk and hear – allowing it to interact with humans in a natural way. No longer dependent on code or a programming language – cognitive systems understand natural language. We call this natural language processing – or NLP. This is not an easy endeavor. Human speech is not always precise – it tends to be ambiguous and its structure can depend on many complex variables such as slang, regional dialects and social context. Thanks to NLP, cognitive systems not only understand the words being said, but they understand the meaning of these words in context. They simulate the human ability to understand the everyday language that people use to communicate. This natural interaction can extend beyond speech to include visual inputs as well. A good example is IBM’s Watson Visual Recognition service, a cognitive system that understands the contents of images. A visual process tags the image, finds human faces, approximates age and gender, and can find similar objects in a collection. This service can also be trained to create custom capabilities like detecting clothing type in retail, identifying defective parts on an assembly line, and much much more.
So there you have it. Now you know what makes a system cognitive – it’s ability to understand, reason, learn and interact. When you combine all these components, the resulting system can be quite impressive. A futuristic use case could be a cognitive car that:
- Monitors a driver’s health (understands).
- Identifies an emergency based on subtle changes in driver vital signs (learns).
- Initiates a conversation with the driver (interacts); and
- Makes recommendations for next best action (reasons) – one of which could be self-driving the car to the closest hospital.
As crazy as that sounds – this future is closer than you think – thanks to the power of the cognitive system!
For more information on the topic: The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World
Title Image copyright: racorn / 123RF Stock Photo