At the recent Microsoft Build conference
, the software giant announced it had created a feature in PowerApps which allows computer generation of code using the English language.
The company is using an AI engine called GPT3, OpenAI’s large scale language-generation model.
About a month ago, I was lucky enough to be approved to access the beta for OpenAI’s models. I have been developing an application that communicates with our end-users based on support tickets and online interactions.
My results have been nothing short of amazing, and there is a significant amount of potential in this solution. To give an example of its power, I asked GPT-3 to provide me with some ideas of articles for the CRN ISV Hub
and here is the entire interaction (Items in bold are my text).
Here are some article titles.
- Seven steps for ISV’s to work with Partners
- What could go wrong with an ISV
- ISVs take note as AI takes a crazy leap forward
- What you need to know about AI or how you could be replaced
- How to make the most money from your ISV
- Don’t let your ISV die
- Five reasons why ISVs are going out of business
- This is what we do to ISVs
Whilst I’m not sure about 7 and 8, the other few are excellent titles that I might use in the future.
OpenAI’s models are nothing short of awe-inspiring, and we’ve used them to respond to helpdesk tickets, communicate with end-users, assist in writing blog posts and much, much more.
The current licensing terms are quite restrictive, so I can’t give you examples of everything it can do. Still, hopefully, this serves to whet your appetite.
ISV’s however, need to take note. I’ve already got the AI to generate compilable code from English instructions, and Microsoft is using it in PowerApps. One user even wrote that he got the AI to deliver a functional REACT web application from English instructions.
Such instructions are:
This is a REACT application that shows a table and input boxes:
GPT-3 will then proceed and write a REACH application with input boxes and a table.
For me, the staggering item was how GPT-3 is “programmed” using the English Language
(or other languages too).
Once we have an application approved and in production, I’ll show you more. Still, for
now, I’ll paste this article into GPT-3 and see what it comes up with as a summary:
The following text was generated using GPT-3, OpenAI’s large scale language generation
model and was unaltered.
“This is a great time to be an ISV, or maybe you’ll be the next Uber, the next Airbnb, the
next Facebook, the next Google, the next Amazon, the next something else. You could be
the next somebody, but you could also be the next nobody.”
All of the above is an example of what is possible with OpenAI’s GPT-3.
Keep reading the ISV Hub to find out more.