Читать книгу Enterprise AI For Dummies - Zachary Jarvinen - Страница 49

Back in the day, a virtual assistant was really virtual, as in digital, not a term to describe a remote clerical worker.

As the personal computer gained popularity, it migrated into the homes of users, who had widely varying degrees of computer literacy and aptitude. Software providers rushed in to fill the void between computer capabilities and consumer competence.

Anybody using computers in 1997 will remember Clippy, Microsoft’s ill-fated virtual assistant. Officially named Clippit, it was a stylized paper clip with googly eyes and Groucho eyebrows standing on a sheet of yellow legal-pad paper like Aladdin on a magic carpet. Watching from a corner of the window, Clippy would monitor what you typed and jump in when he thought you might need help. For example, if you opened a new document and typed “Dear” followed by a space, Clippy would jump to the middle of the screen and say, “It looks like you’re typing a letter. Would you like help?”

In 2010, three years after Clippy officially died, Apple launched Siri, followed in the next six years by Google Now, Alexa, Cortana, and Google Assistant. This new wave of virtual assistants uses voice recognition and expands the scope beyond help with a specific computer application to help with almost every aspect of life. Like chatbots, virtual assistants can be deployed in the enterprise to enhance internal or external customer service.

Clippy provided assistance based on Bayesian algorithms, a family of probabilistic classifiers. Modern virtual assistants use NLP to interact more like a human.