GitHub GitHub Universe 2022 - AI, 21 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: How GitHub Copilot can impact developer happiness - Universe 2022

Description

Presented by: Seth Juarez

GitHub Copilot represents a huge leap forward in developer productivity, but did you know it also contributes to developer happiness? Join Seth Juarez and some special guests to learn what Copilot is, how it works, how it makes developers' lives easier, happier, and more productive.

As always, feel free to leave us a comment below and don't forget to subscribe: http://bit.ly/subgithub

Thanks!

Connect with us.
Facebook: http://fb.com/github
Twitter: http://twitter.com/github
LinkedIn: http://linkedin.com/company/github

About GitHub
GitHub is the best place to share code with friends, co-workers, classmates, and complete strangers. Millions of people use GitHub to build amazing things together. For more info, go to http://github.com

A

Oh hi I didn't see you there, but I'm glad you're here, because today there's something very important and very exciting that you are going to want to know all about. As you may already know, GitHub is obsessed fascinated, nay, consumed by two things: developer happiness and productivity in that order, so they built an AI, assisted coding tool called co-pilot to help improve both. But what is GitHub co-pilot I'm glad you asked to truly understand what co-pilot is you've got to understand the history of devtools, hey Marty, cue, the flashback.

A

Silicon is revolutionizing the way we live in the 1950s recognizably modern computers appeared, but for developers this was still the Dark Ages physical wires, Punch Cards wool, pants EW by the 1960s things had improved a little now we could use monitors and keyboards. In addition to that, fancy new basic programming language- something referred to as technology- has made this possible, but even though our ability to interact with machines was getting better coding was still in a word tedious. The 1970s brought us the first useful, interactive, debuggers and non-ironic disco.

A

Now, for the first time, we could stop our programs to see what they were doing. Oh.

B

That's pretty come on. Let's get that productivity.

A

Leaped forward and that leap led us to the 1980s and the dawn of the internet. Sure the internet was arguably a mixed bag when it came to productivity, but with these breakthroughs came, color monitors and the syntax highlighting that made it so easy to see which part of our code was doing.

A

What the 90s finally introduced code completion, which meant those heavy reference books, could go back on the Shelf where they belonged, and this brings us to the modern era, the 2000s and Beyond, where we've enjoyed the super powers of semantic refactoring, remote pair programming edit, and continue and countless other breakthroughs that made it seem as if our tools finally understood what we were trying to do. Another development that has taken the World by storm is machine learning, which is a very interesting subset of AI, and it has completely changed the way we develop.

A

Software AI is awesome fast forward to now the future and GitHub co-pilot your AI pair programmer the latest, the newest, the most next Leap Forward in this Rich history of developer tooling. But even if it feels like magic and trust me, it does. It's really. Just a bunch of elaborately sophisticated machine learning protocols running on the Cloud's most powerful gpus.

A

As you know, machine learning is really just a whole lot of fancy. Math, you know math now, I, don't want to be a name dropper, but Mike Krieger he's the co-founder of Instagram, maybe you've heard of it. He said about co-pilot. This is the single most mind-blowing application of machine learning. I've ever seen, co-pilot is trained on billions of lines of publicly available code. This means it can turn natural language prompts into code suggestions and it works in dozens of languages.

A

This is pretty exciting stuff right now, let's check in with all of you watching at home, uh hello, caller who's. This hi.

B

Big fan, this is George Michael from Scranton Pennsylvania. Well,.

A

Hi George Michael, thanks for calling, what's on your mind, I'm.

B

Curious if co-pilot can match my coding style in conventions when I'm writing code great.

A

Question- and the answer is yes: co-pilot can anticipate what you're trying to build match your styling conventions and let you focus on solving bigger problems rather than writing boilerplate code, it's like being able to write code with the help of the world's Collective knowledge about software development. It's like having the world's best senior developer as your coding, buddy wow,.

B

So it's like you're taking the cumulative knowledge of the world software developers and cramming it into your workflow.

A

And somehow it fits great call thanks, Michael, but just having co-pilot wasn't enough. Github is a company of Engineers and scientists, so naturally they wanted to understand and to measure the impact of GitHub co-pilot on the community and find out whether developers were indeed happier and more productive. So GitHub did some research. In particular. They wanted to discover three things: how does co-pilot impact developer productivity? What does the quantitative and qualitative data say about the experience of copilot users and how does using copilot impact day-to-day work to talk about these findings?

A

I'm joined by irini from github's research, team, hi Rini? So what did.

C

You find thanks Seth in short, there's more to productivity than just speed. Tell.

A

Us more, we.

C

Surveyed, more than 2 000 people and found that between 60 and 75 percent of users reported they feel more fulfilled with their job. They also feel less frustrated when coding and they're able to focus on more satisfying work when using GitHub copilot copilot can also help conserve mental energy. 73 of developers reported that GitHub co-pilot helped them stay in the flow while 87 percent said it helped preserve mental effort during repetitive tasks. So.

A

Co-Pilot can take away the boring and repetitive.

C

Work exactly now, obviously, there's a lot of technical detail that goes into each of these questions. There's a lot of process that goes into defining the terms and Gathering the information, and you can read all about our methodology here. Well,.

A

Thanks for that irini, when the team built co-pilot, they expected that it would help developers complete tasks faster. They didn't expect to see 90 percent of users agreeing that it helped them to do just that. Let's check in again with the callers Sam from Lake Arrowhead California thanks for calling. How are you hi.

D

A

Sam, you are, how are you.

D

Yeah: okay, I am a loving co-pilot I.

A

D

So I'm wondering if, after you learned that 90 of users were finding co-pilot really helpful, did you do any additional research about how much more efficient it can make developers.

A

Great question: actually we did so we set up a second experiment and oh I can.

C

Jump in here Seth we randomly split 95 professional developers into two groups, and we timed how long it took them to write an HTTP server in JavaScript, one group used copilot and the other one didn't we control for as many factors as we could, but even so, what we found was the group using copilot had a higher rate of completing the tasks than the group that didn't 78 versus 70. That's.

A

Really cool yes,.

C

But what really stood out was the group using copilot completed the task so much faster. On average they finished their HTTP server, 55 more quickly, finishing in 71 minutes versus 150 minutes for the group without copilot wow.

A

That's that's really significant thanks, Irene, no.

C

Problem, okay, I'm, actually going now me too.

D

A

Exciting thing about this data is what it tells us about how co-pilot helps developers and it does it in three really big ways: it helps them complete tasks and projects faster. It conserves mental energy and it helps devs stay more focused on the work that is more satisfying and interesting.

A

The field of AI assisted code is new, but it's growing rapidly and there are lots of exciting research exploring this top to learn more about this research or to start using copilot right now. Visit copilot.github.com.