When I visit with non-IT corporate executives and ask them about artificial intelligence (AI), machine learning (ML) and natural language processing (NLP), they tell me that they have initiatives underway. But they don't exactly know what AI, ML, and NLP are.
Trying to explain what AI, ML, and NLP are, how they work, and how they deliver results for the business isn't easy. Yet, all of these technologies have prominent roles in analytics as IT deploys them. It's incumbent upon CIOs and IT leaders to find ways to break down these technologies and their business deliverables in plain language for non-technical stakeholders.
How do you find easy ways to explain these technologies, how they work together, and why it makes business sense to use them?
Here are some plain language explanations that could prove helpful.
AI is a computer system that can perform tasks that were formerly performed by humans. It works in contexts where the tasks are repetitive, and where the data to be reviewed is vast and would take many human man-hours to process and digest. AI operates based upon human-defined rules and expertise programmed into it in the form of programmable logic and algorithms. AI cannot perform well outside of the rules that are defined for it the way that creative human reasoning can. That's because AI strictly follows business rules that users and experts program into it.
In business applications, AI is best suited for highly tailored specific use cases where human experts define clear sets of business rules.
A prime example is a medical diagnosis system that can pore through terabytes of data contained in medical journals, diagnosis histories, and other data sources. The AI software reviews all of this data in a fraction of the time that it would take a human to do. Then the AI presents four or five possible diagnoses for an elusive medical condition to a physician, who then uses his or her own professional judgement, in concert with collaborative discussions with other experts, to make the final diagnosis.
AI can also be used to predict weather patterns based upon weather history, to develop the most optimal travel routes for logistics carriers, or to predict what e-commerce website visitors are most likely to purchase next, based on their past purchasing patterns and what they've browsed on the website.
A majority of companies begin their AI deployments by using AI for analytics. As companies gain more experience, they seek to "train" their AI by introducing machine learning, which is a sub-category of AI that enables the AI to gain additional insights into data on its own by recognizing recurrent pattens of data and then drawing conclusions (and "learning") from these conclusions.
Machine learning is a sub-category of AI that enables an AI system to learn and adapt to new data and events so the AI can become "smarter." The ML component of AI learns by observing repetitive data patterns, and then applying a set of algorithms and logic developed by human experts that enable it to make decisions based upon the repetitive data patterns it is observing.
An example in a logistics scenario is a recurrent pattern at a particular highway intersection where there are always traffic delays. If the sequence continues to recur, the ML component of the AI is likely to detect the pattern and to conclude that it is better to reroute traffic another way so that the busy intersection can be avoided.
Like machine learning, natural language processing is also a sub-category of artificial intelligence. NLP is used to understand, interpret, and manipulate human language.
An example of this is SIRI on an iPhone. The SIRI NLP component of AI is able to recognize your human voice command and respond in kind in the same language.
Other NLP examples include automated phone and chat systems that recognize human languages and conduct automated conversations with you, or a home security system that recognizes and responds to human voice commands
NLP together with AI's normal data processing and analytics is capable of automating numerous business processes that involve the reading, speaking, and writing of language.
Bringing it all together
While at first glance many of these AI, ML, and NLP discussions might seem overly simplistic to IT professionals who are used to conversing in acronyms and technical abstractions, conversations like these can be instrumental in gaining and retaining executive, board, and end-user support for AI, ML, and NLP projects.
Most importantly, plain language conversations that link the technology to the business are essential for eradicating the feelings that many business executives and end users have about AI, ML, and NLP being mysterious "black boxes."
"A lot of senior executives and business leaders today are almost desperate to understand how AI may affect their businesses,” said Thomas W. Malone, director of the MIT Center for Collective Intelligence. “I think leaders are increasingly worried in many cases that if they don’t figure out how to use AI effectively, they’ll be left behind.”
For more on AI in the enterprise, read: