Stephen Wolfram is talking more publicly about the Wolfram Language, this week releasing a video demo of the knowledge-based programming language. As he describes in the video below, the symbolic language builds in a vast amount of knowledge of how to do computations and about the world itself. “Through symbolic structure of the language,” he says, primitives for everything from processing images to looking up stock prices “are all set up to work together in a wonderfully coherent way.”
The concept of coherence – the idea that everything in the language must fit together – is in fact one of the principles that have guided the development of the language over the past decades, he explains, as is maximum automation – the idea that the language should take care of as much as possible. If you are working in machine learning, for example, and want to build a data classifier, “in the Wolfram Language there’s just one Super Function, Classify, that’s packed with meta-algorithms to automatically figure out what to do,” he says. There are thousands of Super Functions in the language, he says, which “effectively give you the highest possible level of building blocks for programs.”
These building blocks contain not only algorithms but knowledge and data, too, including knowledge about how to import and export formats and interact with external APIs and huge amounts of curated computable data – the same data that powers Wolfram Alpha, completely programmatically accessible, he says. Ask it when the sun will set today, and you’ll get the answer for your current location, for instance.