And while there are some negative effects of these advancements, generative AI has had a great positive impact on software development thus far.
Generative AI Revolutionizing Software Development
Since the launch of OpenAI’s chatbot, ChatGPT, back in November, the tech industry has seen the release of many generative AI coding tools including Microsoft’s GitHub Copilot and Amazon’s CodeWhisperer.
These coding companions perform a lot of the monotonous tasks involved in creating new apps and services using a large language model (LLM) trained on open-source projects and technical documentation.
Generative AI coding tools are powered by large language models (LLMs). Today’s state-of-the-art LLMs are transformers, which makes them adept at connecting tokens, big-picture thinking, and scaling. The results are coding and content suggestions that are contextually relevant.
— GitHub (@github) April 20, 2023
Software for code completion and coding assistants are not new concepts since for well over a decade, developers have had the ability to repeatedly iterate over properties and functions using well-known IDEs.
This time, however, there is a key distinction, LLMs, which have the ability to produce syntactically sound, idiomatic code by understanding your meaning and inferring context from the previous lines of code including comments in addition to being able to predict the next line of code.
In an interview with All Things Distributed, Doug Seven, GM of Amazon CodeWhisperer said:
“The concept is to support that with a large language model, allowing a coding assistant to grasp what you’re doing. Based on that understanding, it can guess what you might want to do next and suggest it to you, possibly by completing the line of code you’re working on. If you’re writing a method signature, it could provide the parameters you need to fill in.”
In addition, these AI-powered tools have greatly simplified repetitive and time-consuming tasks like developing unit tests and converting code between languages resulting in enhanced workflows, productivity gained, and time saved.
Joe Welch, principal and technology leader of Launch Consulting says:
“By incorporating GitHub Copilot into VS Code for a recent project, we saw programmers reduce ten-minute tasks, such as writing a small function, down to the 30 seconds it took to simply write out a comment that explains the function.”
A Calculator For A Developer
Even as the development of these assistants moves forward, developers are already making significant profits from AI automation in two main areas: code creation and code editing. The writing produced by ChatGPT resembles the AI-generated code. It is useful, created rapidly, and requires human review to confirm its accuracy.
Nevertheless, AI-generated code speeds up development and lowers the risk of human error. The speed with which developers can detect and fix faults with AI aid in code editing programs accelerates the software development life cycle.
The introduction of these coding companions has raised a lot of questions about the future of software development roles as well as the requirements to become one. Will developers still require computer science degrees?
Seven equated this situation to math class where students still have to learn addition, subtraction, multiplication, and division after which they then learn some basic algorithms and some basic algebra capabilities.
“And eventually you get to a point where your teacher says, okay, you can bring a calculator to class now, and you’re going to use that to speed yourself up in doing the things that you already learned how to do by hand. And that’s what Code Whisperer is. It’s the calculator for a developer,” he added.