One does not have to search very hard to find articles about artificial intelligence (AI). Despite the plethora of sources available, it might surprise you to learn there are few, if any, universally accepted terms associated with AI. Doug Black explains, “AI is the compelling technology topic of conversation du jour, yet within these conversations confusion often reigns — confusion caused by loose use of AI terminology.” Black explains confusion is almost inevitable because “AI comes in a variety of forms, each one with its own distinct range of capabilities and techniques, and at its own stage of development.” He continues, “Some forms of AI that we frequently hear about — such as Artificial General Intelligence, the kind of AI that might someday automate all work and that we might lose control of — may never come to pass. Others are doing useful work now and are driving growth in the high performance sector of the technology industry.”
Some working definitions
Black observes trying to define AI terms is somewhat of a fool’s errand because “the industry is growing and changing so fast that terms will change and new ones will be added.” It’s because the field is growing and changing so fast that some clarity is required. Borrowing from a number of sources, below are some working definitions explaining some of the various forms of AI now being developed.
Artificial Intelligence. Pegasystem defines AI as “a broad term that covers many sub-fields of computer science that aim to build machines that can do things that require intelligence when done by humans.” Similarly, Eric Knorr, Editor in Chief of InfoWorld notes, “Artificial intelligence is the umbrella phrase under which all other terminology in this area falls.” When the term was first coined in the 1940s, there were two broad categories of AI: “Weak” (sometimes called “narrow”) AI and “strong” (sometimes called “broad”) AI. Wikipedia notes, “Weak artificial intelligence (weak AI), also known as narrow AI, is artificial intelligence that is focused on one narrow task. Weak AI is defined in contrast to either strong AI (a machine with consciousness, sentience and mind) or artificial general intelligence (a machine with the ability to apply intelligence to any problem, rather than just one specific problem).” These definitions were probably sufficient in the early days of AI since weak AI systems were being trained to do one thing, like play chess. At the very least, current discussions should distinguish between weak, strong, and general AI:
- Weak AI: The Wikipedia definition of weak AI remains relevant: “Weak artificial intelligence (weak AI), also known as narrow AI, is artificial intelligence that is focused on one narrow task.” In other words, weak AI developed to handle/manage a small and specific data set to answer a single question. Its perspective is singular, resulting in tunnel vision.
- Strong AI: As noted above, strong AI originally referred to General AI (i.e., a machine with consciousness, sentience and mind) “with the ability to apply intelligence to any problem, rather than just one specific problem.” Today, however, there are cognitive systems that fall short of AGI but far surpass weak AI. These systems were developed to handle/manage large and varied data sets to answer a multitude of questions in a variety of categories. Cognitive AI can deal with ambiguities whereas weak AI cannot.
- General AI: The AGI Society notes the ultimate goal of AGI is to develop “thinking machines” (i.e., “general-purpose systems with intelligence comparable to that of the human mind”). Black adds, “AGI is ‘the singularity,’ another futuristic concept around the idea that AGI will trigger runaway technological growth, a runaway reaction of self-improvement cycles resulting in a powerful superintelligence that would, qualitatively, far surpass all human intelligence.”
Occasionally, pundits discuss AI in terms of General and Applied AI. The AGI Society notes, “The mainstream of AI research has turned toward domain-dependent and problem-specific solutions.” In applied AI, the emphasis is on solutions. When discussing weak and strong applied AI, a number of AI techniques are likely to come up. They include:
Machine learning. Ryan Young writes, “Machine learning is a subdivision of AI that involves machines deciphering data and learning for themselves. It’s used a lot throughout the businesses of today as is very efficient when used in areas such as speech, object, and facial recognition, translation, and other tasks. Programs that use machine learning can learn to recognize patterns on their own and make predictions based on what it’s learned.” Steven Norton asserts, “When people talk about artificial intelligence, they usually are referring to one of its subfields: machine learning. While AI concerns itself with making machines think like humans, machine learning has a narrower purpose: enabling computers to learn from data with minimal programming.”
Neural networks and deep learning. Some experts try to make a distinction between Neural Networks and Deep Learning; however, the terms are often used synonymously. For example, Norton writes, “The hottest field in artificial intelligence today is a branch of machine learning known as deep learning. It uses complex algorithms — essentially a set of instructions for solving a particular problem — to perform more abstract tasks such as recognizing images. A well known deep learning tool is the neural network, which roughly tries to mimic the operations of a human brain.”
Computer vision. Black writes, “Computer vision is the ability of computers to identify objects, scenes and activities in images using techniques to decompose the task of analyzing images into manageable pieces, detecting the edges and textures of objects in an image and comparing images to known objects for classification.”
Natural language/speech processing. Natural language processing (NLP) is involved in both inputs to computers and outputs to users. As the name implies, NLP is capable of understanding and using common language to find solutions to problems. Since much of the data being created today is unstructured (i.e., it comes from sources other than structured databases; such as, social media, newspapers, videos, etc.), NLP is essential to analyze it. And since many computer users are not data scientists, NLP can be used to explain results in terms understandable to non-technical users. The analysts at Nanalyze explain, “Natural language processing, as defined by aitopics.org, ‘enables communication between people and computers and automatic translation to enable people to interact easily with others around the world.'”
Robotic Process Automation (RPA). Black explains “RPA is computer software that is configured to automatically capture and interpret existing applications for processing a transaction, manipulating data, triggering responses and communicating with other digital systems.” In its simplest form, RPA is more of a rules engine than an AI application. In discussions with clients, I’ve found many of them want to move beyond RPA and make their processes smarter not just automated. They want Cognitive Process Automation™ (CPA). Ankur Kothari (@ankur_786), Co-founder and Chief Revenue Officer at Automation Anywhere, notes, “When integrated with cognitive capabilities, RPA can move beyond simply automating standardized business processes to understanding the context for — and making — important business decisions.”
Cognitive Computing and Strong AI
As noted earlier, today there are cognitive systems that fall short of AGI but far surpass weak AI. The Enterra Enterprise Cognitive System™ (Aila™) — a system that can Sense, Think, Act and Learn® — is among the strongest cognitive systems available. There are, however, a number of approaches that fall under the cognitive computing rubric. I define cognitive computing as a combination of semantic reasoning (i.e., the use of machine learning, natural language processing, and ontologies) and computational intelligence (i.e., advanced analytics). The analysts from Nanalyze write, “Cognitive computing is one of those terms that has fairly recently entered the AI lexicon. One of the definitions for cognitive computing … that seems to be prolific around the inter-webs sums it up thus: ‘Cognitive computing involves self-learning systems that use data mining (i.e., big data), pattern recognition (i.e., machine learning) and natural language processing to mimic the way the human brain works.’ Katherine Noyes writes in Computerworld that cognitive computing ‘deals with symbolic and conceptual information rather than just pure data or sensor streams, with the aim of making high-level decisions in complex situations.’” Jenna Hogue adds, “The types of problems [involved in cognitive computing] … tend to be much more complex and human-like than the average non-cognitive system. These problems tend to comprise multiple different variables included, shifting data elements, and an ambiguous nature.” That’s why cognitive platforms should be considered strong AI systems.
 Doug Black, “AI Definitions: Machine Learning vs. Deep Learning vs. Cognitive Computing vs. Robotics vs. Strong AI….” Enterprise Tech, 19 January 2018.
 Eric Knorr, “Making sense of machine learning,” InfoWorld, 6 March 2017.
 “Weak AI,” Wikipedia.
 Ryan Young, “Artificial Intelligence, Machine Learning, and Deep Learning and How they Differ from One Another,” TrendinTech, 25 March 2017.
 Steven Norton, “CIO Explainer: What is Artificial Intelligence?” The Wall Street Journal, 18 July 2016.
 Staff, “An Artificial Intelligence Definition for Beginners,” Nanalyze, 12 November 2016.
 Ankur Kothari, “Get Ready for the Cognitive Era,” BW CIO World, 28 June 2017.
 Jenna Hogue, “Cognitive Computing: The Hype, the Reality,” Dataversity, 12 January 2017.