GitHub Copilot For Business Update
The A.I. software of auto-complete for coders will get a lot better in 2023 and 2024.
Hey Everyone,
More programmers are using ChatGPT and GitHub Copilot for Business now than ever before in February, 2023.
GitHub’s OpenAI-powered developer tool with new capabilities goes live with enhanced Codex
So what actually changed this week?
GitHub today announced that Copilot for Business, the company’s $19/month enterprise version of its AI-powered code completion tool, is now generally available, after a short beta phase that started last December.
Google is busy working on a competitor to this based on PaLM. I believe they are likely to announce this tool at their I/O Conference in May or June, 2023.
As you probably already know, first previewed in 2021, Copilot uses OpenAI’s Codex large language model (LLM) to turn textual descriptions into source code.
Among Microsoft’s many weird experiments around A.I. adoption of its software suite productivity, GitHub Copilot is perhaps the one I’m most bullish upon. The reviews of BingAI weren’t so great this week, if we are to believe the journalists and anti-hype PR.
Whatever the case may be, as engineers and software dev enthusiasts if the sum-total of all of these new tools allows us to be more productive, than it’s all worth it, so long as we use the tools as they were intended.
For companies and enterprise, helping Software devs be more productive with a tool that costs just 19$ a month is a very affordable rate. ChatGPT Plus is $20, last time I checked.
So the mid Feb announce comes with additional “enhancements”. Enhancements unveiled February 14 include an update to the underlying OpenAI Codex AI model to support large-scale improvements to code quality suggestions. The time required to deliver suggestions has also been reduced. These enhancements are available for both Copilot for individuals and Copilot for Businesses.
GitHub said that in addition to upgrading the underlying Codex model to improve the quality and responsiveness of its code suggestions, it also developed a new security vulnerability filter to make the tool's code suggestions more secure and help developers identify insecure coding patterns as they work.
Copilot for Business adds features like license management, organization-wide policy management and additional privacy features. Until now, you had to work with GitHub’s sales organization to sign up for the business version, but now there is a self-serve option as well.
Microsoft for Coders
GitHub said in a statement GitHub Copilot is the world’s first at- scale artificial intelligence (AI) developer tool.
I’m not sure that’s technically true but they do seem to have corned this market in the early going.
The benefits they claim are striking if true:
55% faster coding?
46% code written?
Microsoft of course thinks all of A.I. is helping us with a co-pilot in search and elsewhere. I guess that means GitHub Copilot’s branding is selling well.
Sitting within the editor as a simple extension, GitHub Copilot draws context from a developer’s code to suggest new lines, entire functions, tests, and even complex algorithms.
It’s hard to verify the efficacy of the benefits that Microsoft claims but it does sound like it has momentum.
Specifically, the technical improvements reportedly provide:
A more powerful client-side AI model with algorithms that improve the quality of code suggestions by 44 percent
Simple self-serve purchasing, where businesses can sign up for Copilot themselves and immediately assign seats -- even if they don't use the GitHub platform
AI-based security vulnerability filtering to target common issues
What do you think?
One thing that GitHub seems to be doing really well is integrating feedback to improve the product.
Better code suggestions
One of the important parts of the LLM life cycle is gathering user feedback and updating models. Since officially launching Copilot, GitHub has used feedback from millions of developers to improve its model, increasing the quality of code suggestions and reducing latency. According to GitHub’s latest report, on average Copilot writes 46% of code for developer users, up from 27% in June 2022.
Microsoft definately shouldn’t be allowed to acquire Activision and even GitHub and LinkedIn weren’t great acquisitions for the community. But anyways:
Their team is constantly refining the models and adding new features as they become available in Azure’s OpenAI Service.
Since its release, GitHub Copilot has transformed developer productivity for more than one million people, helping developers code up to 55 percent faster.
That sounds like a lot, and Twitter influencers are coming out and saying how amazing it is, which sounds like an unsolicited Ad.
Then it turns out he’s joining OpenAI again, sort of ridiculous the manipulation going on, but anyways.
How many users does it even have though? The Visual Studio Code extension has been installed nearly 3.7 million times, while the Visual Studio tool has been installed nearly 154,000 times. According to GitHub, more than 1.2 million developers used Copilot's technical preview in the past 12 months as of September, 2022. I wonder what the numbers are at now?
No ChatGPT like craze, that’s for sure.
As the model gets better, the team is also adding new features, including things like “fill-in-the-middle,” where the model can’t just complete a line but also start adding words in the middle because it knows what sits before and after the current cursor position, for example.
The PR is a bit like a Microsoft commercial and what you might expect:
“With more accurate and responsive code suggestions, we’re seeing a higher acceptance rate [for code suggestions],” Shuyin Zhao, GitHub senior director of product management
Trained on billions of lines of code, GitHub Copilot turns natural language prompts into coding suggestions across dozens of languages.
If I was just learning to code, I’d likely try all of these tools including what ChatGPT can do in various languages.
Will Coders Get Automated?
Dohmke expects that soon, Copilot will be able to generate 80% of a developer’s code. Today, that number is about 46% across programming languages — and 61% for Java.
If A.I. buddies in the AI-human hybrid workforce do more of the heavy lifting, it’s going to be a different job if this keeps up. You can almost fathom a time when the no-code revolution means less of a demand for software engineers? I don’t know enough about it but it’s worth speculating about this if you are thinking of a career in software development. On the other hand an expected supply-demand shortage may become more serious.
GitHub FIM
GitHub has also added a few new tricks to improve the Copilot experience. One of them is a new paradigm called “Fill-in-the-Middle” (FIM), which gives Copilot more context to improve code suggestions.
Microsoft is Gaining Adoption via Generative A.I. Hype
“The rise of generative AI models like GitHub Copilot has triggered widespread
recognition that the age of AI has begun. But until now, generative AI has largely
benefited the individual. That’s changing today,” said its Chief Executive Officer Thomas Dohmke.
There’s now 5 million users on the Programming Subreddit, with some interesting debates.
Tim Davis, professor of computer science at Texas A&M, has reported that GitHub Copilot has produced "large chunks of my copyrighted code, with no attribution, no LGPL license" even when the block public code flag is enabled. This is not the only controversy surrounding the tool. The tool is likely to face a lot of legal challenges and lawsuits.
Github owned by Microsoft now claims 400+ organizations are already using GitHub Copilot. That’s not actually that many considering how long this has been out and available. But it does point to how coding is changing in a world of software empowered by A.I.