Artificial Intelligence: A Primer

Artificial intelligence is a widely used, and often misused, term first mentioned in a 1956 college seminar exploring the potential of technology for problem solving. AI’s meaning has continued to evolve, which has contributed to the concept’s apparent fuzziness. Still, AI experts would agree on this basic definition, refined by The Conference Board:

» AI is technology that mimics human thinking by making assumptions, learning, reasoning, problem solving, or predicting with a high degree of autonomy.

While this definition might seem quite straightforward, concepts of AI continue to change in response to new technologies. Viewing AI as a concept whose parameters are perpetually redefined is not the same thing as saying, “AI is a fuzzy term,” or “No one can agree on what AI means.” Those statements are untrue. Most people who work in AI, evaluate AI products, or observe its social and business impacts can clearly articulate what it is and isn’t. Business executives must be able to do so, too.

Think of AI as an umbrella term that covers a variety of leading-edge capabilities. Under today’s AI umbrella, this includes machine learning, deep learning, natural language processing, text analytics, voice recognition, speech recognition, andcomputer vision.

Rather than thinking about AI as a binary (i.e., yes/no) concept, it’s more useful to imagine it as a range, or a spectrum, with assisted intelligence at one end and autonomous intelligence at the other.

The model below can help. It presents shades, or gradations, of “AI-ness.” Importantly, the model also captures what’s not AI: automation; that is, the tools and systems that can perform simple, repetitive tasks independently, although they cannot think or learn in the process. Automation is effectively off the AI grid, therefore, but we include it in our model, set apart from AI, because the two are so often conflated.

 Shades of Artificial Intelligence

A Glossary of AI Terms

• Machine learning (ML) is arguably the most common kind of AI in use today. Technology that uses ML continuously learns from new data and experience, becoming smarter and more capable over time. It does this autonomously, in contrast to analytics, where humans need to refresh the model as the data change and decide what questions to pursue. Computer vision (technology that can process visual information), voice recognition, and speech recognition use ML to analyze data.
• Deep learning is an advanced form of ML that identifies patterns and trends in data beyond what humans are capable of—for example, by analyzing a continuous stream of real-time big data and instantaneously adjusting its models as it deems necessary.
• Natural language processing (NLP) is a tool for extracting the meaning from the words that humans speak or write. Ideally, it factors in the context to capture nuances and subtleties of language. Some companies’ human capital analytics teams, for example, can use sentiment analysis, a form of NLP, to identify employees’ feelings or attitudes from the aggregated, anonymous data the analysis scrapes from online discussions and open-ended survey questions.
• Generative AI is a type of AI or category of AI algorithm that generates new outputs based on the data it has been trained on. It can perform tasks that typically require human-like intelligence, such as problem-solving, learning, perception, understanding language, and making decisions. It has a wide range of applications, including creating content—text, imagery, audio, and synthetic data—and information that is artificially manufactured rather than generated by real-world events.