Unlike in other years, this year’s Microsoft Buildsurprises — but there’s one announcement that will surely make developers’ ears perk up: The company is now using OpenAI’s massive GPT-3 natural language model in its no-code/low-code Power Apps service to translate spoken text into code in its recently announced Power Fx language.
Now, don’t get carried away. You won’t develop the next TikTok using only build an app of any sophistication.. Instead, what Microsoft is doing here is taking some of the low-code aspects of a tool like and essentially using AI to turn those into no-code experiences. For now, the focus here is on Power Apps formulas, which, despite the low-code nature of the service, is something you’ll have to write sooner or later if you want to
“Using an advanced AI model like this can help our low-code tools become even moreto an even bigger audience by truly becoming what we call no code,” said Charles Lamanna, corporate vice president for Microsoft’s low-code application platform. In practice, this looks like the citizen programmer writing “find products where the name starts with ‘kids’ ” then rendering that as “Filter(‘BC Orders’ Left(‘Product Name’,4)=” Kids”)”. Because Microsoft is an investor in OpenAI, it’s no surprise the company chose its model to power this experience.
important to note that while this makes programming more accessible, Microsoft itself stresses that users still have to understand the logic of the application they are building. “The features don’t replace the need for a person to understand the to assist people learning the Power Fx programming language and help them choose the right formulas to get the desired result. That can dramatically expand access to more and more rapidly train people to use common code tools,” the company explains in today’s announcement.
To some degree, this isn’t all that different from using the naturalquery functions now available in tools like Excel, PowerBI, or Google Sheets. These, too, translate natural language into a formula, after all. GPT-3 is probably a bit more sophisticated than this and capable of understanding more complex queries, but translating natural language into recipes isn’t that new.
The long-term promise here, though, is for tools like this to become more intelligent over time and handle more complicated programming tasks. But that’s a , though. More complex queries require more of an understanding of a program. A formula, for the most part, is a pretty self-contained statement, but a similar model that could generate “real” code would have to contend with a lot more context. These new features will go live in public preview in English to users in North America by the end of June.