GitHub: Developers Have Embraced AI, but Companies Are Slow to Catch Up

An AI-generated image of humans playing tug-of-war with a robot.

Software developers in the workplace have embraced artificial intelligence as an integral part of helping them do their jobs. A new study from GitHub reveals that 97 percent of engineers are using AI coding tools at some point. However, the companies they work for have been much slower in adopting AI tools and technologies—40 percent indicated their employers are “actively” encouraging and promoting AI.

Despite AI’s benefits in improving efficiency and productivity, many organizations are reluctant to incorporate this transformative technology. This could be because it lacks personalization, concerns over data sovereignty, or the difficulty of measuring AI’s broad impact.

This AI adoption gap between developers and companies reflects the ongoing consumerization of IT in which new technologies are introduced through grassroots efforts. For GitHub’s Chief Operating Officer Kyle Daigle, this research is a clarion call for executives to “do the culture shift necessary to bring AI tools to their organizations.”

Helping Developers Code, But Largely When Off-The-Clock

99 percent of developers say they're using AI coding tools at work. Source: GitHub 2024 survey on AI in software development
99 percent of developers say they’re using AI coding tools at work. Source: GitHub 2024 survey on AI in software development

While this isn’t the first time GitHub has surveyed the trend of AI in software development, this year is the first to have queried a global set of enterprise developers, data scientists and software designers. Two thousand were polled across four key countries: the United States, Brazil, India and Germany, with the goal of “understanding how AI is changing, the way they work, and how their companies are changing to support AI or to find AI.”

And though “almost everyone” says they’re using AI tools, they mainly take advantage of the tech outside their 9-to-5. On average, 24 percent say they’re only using AI at work. What’s more, it is likely to be unsanctioned by the organization. Between 38 and 40 percent of respondents note that their companies welcome AI, so more progress must be made. GitHub recommends firms establish “a roadmap, a clear strategy, and policies…to ensure wider adoption happens through building trust and driving measurable performance metrics.”

Software developers report between 38 to 40 percent of companies are "actively encouraging" AI use. Source: GitHub 2024 survey on AI in software development
Software developers report between 38 to 40 percent of companies are “actively encouraging” AI use. Source: GitHub 2024 survey on AI in software development

GitHub emphasized that the intention of its research has evolved from previous years. I’m told that its 2023 survey was more about the “aspiration of AI” and what people thought the technology could do for us. This year, it’s “very much about what AI is doing right now for people, and it’s more about what they think, how they think it could do more…if they got it through work.”

“But it’s not like a hypothetical shift at the individual developer level. They are feeling it. They’re living it right now. It is table stakes. And now, I think the question is, how can we cross the chasm to their employers, to also feel the opportunity that comes from adopting this AI tool deeply,” Daigle continued. “And so, in just a year’ish…we’ve gone from ‘wow, I’ve used ChatGPT, and I’ve used GitHub Copilot, and I’m really excited about what it’s going to do.’ Now, it’s ‘oh, no. It’s saving me three hours. We’re all using it. My employer is not buying it for me, which is strange.’ But how can we keep moving? How can it help me write secure code? How can it help me remediate it?”

“Developers are finding more and more of the toil opportunities to remove. And it’s not aspirational. It’s like, ‘And the next step is…’ We have line of sight into the next level of improvement for the individual, and now we just need to help larger organizations realize that they can have this at a much grander scale.”

Thinking Beyond Table Stakes With AI

“I’m a developer by trade. That’s what I did at GitHub up through the deal, and then I led engineering teams focused on integrating with GitHub. Then, in 2018, I switched to the business side. I wanted a different challenge, and that’s what led me to the role I’m in now,” Daigle told me at the start of our interview in May. At the time, he was in the midst of a customer road tour and had stepped out from a company-hosted dinner to speak with me from his hotel room in Paris, France.

“My heart is with developers and writing code day-to-day. But now I’m taking all of the learnings of how to write good software and applying them to organizations—how [they] can both use GitHub and our platform, but also more so these days, how they can adopt AI all up and then, obviously, use GitHub Copilot to drive change inside big and small organizations.”

GitHub Chief Operating Officer Kyle Daigle. Photo credit: GitHub
GitHub Chief Operating Officer Kyle Daigle. Photo credit: GitHub

Daigle has a good perspective of the situation as a developer and executive. He shares that companies unwilling to embrace AI are already receiving “some of the latent benefits” of their software developers using these tools, such as ChatGPT. But they’re not taking full advantage of AI to “deliver the developers from toil and let them unlock that collaboration that we hear from developers so much and focus on that…innovation and creativity.”

In other words, with AI, developers can avoid mundane and repetitive tasks and instead shift their focus on being creative and working better with their colleagues and customers.

“It’s not just enough to get the [AI] tools into the hands of developers. There’s also a culture shift that needs to happen for organizations to get the full potential of AI,” he remarked.

Daigle believes AI tools are “magnifying the existing culture of an organization.” “If you have an engineering culture that’s incredibly strong and really focused on solving the hardest parts of the problems…then Copilots [is] really just going to speed you up there. If you have a company in an engineering culture that’s…focused on…getting to the next step, you’re not going to get as much of the value out of the AI moment that we’re in for those companies because AI requires you to solve the bigger problem.”

He argued that companies that embrace AI and bring it to their teams “with such fervor” will have a competitive advantage. Whether that’s simply giving developers what they’re looking for, developing faster, or rapidly bringing products to market, “the companies that don’t adopt these tools, that don’t…drive for that next digital transformation culture change moment, are ultimately the ones that are going to be left behind as employees inevitably…jump to the companies that have the better and best tools.”

The AI Adoption Tale of Four Countries

Why did GitHub survey developers specifically from the U.S., Brazil, Germany, and India? And does AI adoption vary country-to-country?

Daigle explained that GitHub wanted a cross-section of countries “semi-representative of the region they’re located in.” The U.S. was selected for obvious reasons. As for Germany, it was picked to represent the EU, India was to cover the “broader, expanding [Asia-Pacific] region, and Brazil for South America and the Latin American region.

According to Daigle, the U.S. “has the highest, the quickest amount of adoption from a company’s perspective.” German developers are “much slower to adopt,” likely because they’re not using the tools outside of work. “Places that are still these enormous centers of economic change, particularly for the [European Union], have their developers using [AI] tools, but they’re not adopting them. They’re not rolling them out. They’re still stuck in the era of pre-AI tools where we can demonstrably show these hours and hours of improvement.”

Though the U.S. is the leader in AI, the study suggests India remains competitive. “In a country that’s producing more software developers per capita than most, India will overtake the United States and the number of software developers on GitHub in the next few years. Part of that is because they are adopting these tools.” Daigle says India is producing more code, especially with open source. He adds that even though the economic incentives of India and the U.S. are different, “they still seem to be adopting the tools very regularly.”

GitHub’s survey highlights one “problem” in AI: the infrastructure. As Daigle notes, this is a significant difference between the four countries. Brazil comes ahead of the U.S. because it has more cloud-based subscription developer tools than any other country. Daigle found this interesting because one might assume that the U.S. would lead here. Germany trails in this area, but that’s because of its policy for data sovereignty.

Using AI to Raise One’s Developer Profile

Software developers say AI can help meet customer requirements. Source: GitHub 2024 survey on AI in software development
Software developers say AI can help meet customer requirements. Source: GitHub 2024 survey on AI in software development

Though this survey provides a snapshot of developer attitudes toward AI, the intended audience is the companies these engineers work for. It’s a subtle plea urging the sanctioning of AI usage in the workplace with the belief that the technology can improve the quality and security of their programming.

GitHub’s survey reveals optimism from most respondents (61 percent in Germany to a high of 73 percent in the U.S.) about AI coding tools helping to improve their ability to meet customer requirements. Those levels vary based on a company’s stance on AI use. Developers also believe using AI tools will eliminate software vulnerabilities.

Software developers claim AI will improve code security. Source: GitHub 2024 survey on AI in software development
Software developers claim AI will improve code security. Source: GitHub 2024 survey on AI in software development

Using AI tools at work can also improve one’s professional standing. After all, gaining proficiency in AI coding tools in the workplace lends developers credibility compared to side projects or non-work efforts. Therefore, if companies encourage AI tool use, these engineers can show their worth and demonstrate they’re a real asset, raising their financial value and productivity.

GitHub: Leader in AI Code Assistants

Daigle claims that GitHub’s survey is product agnostic. But it’s not hard to believe there’s a hint of product placement in the research. After all, GitHub is a popular developer environment filled with code repositories, and the power of Microsoft’s AI is baked into it. The company recently reported more than 1.8 million paid seats for GitHub Copilot.

However, GitHub isn’t the only service offering an AI coding assistant. The market is getting increasingly crowded with Amazon CodeWhisperer, Oracle Code Assist, Anysphere, Cognition’s Devin, and others. Analyst firm Gartner recently published its Magic Quadrant chart for AI Code Assistants, revealing that the Microsoft-owned firm was labeled a Leader—a high honor.

Gartner's Magic Quadrant for AI Code Assistants (August 19, 2024). Image credit: <a href="https://www.gartner.com/doc/reprints?id=1-2IKO4MPE&ct=240819&st=sb" target="_blank">Gartner</a>
Gartner’s Magic Quadrant for AI Code Assistants (August 19, 2024). Image credit: Gartner

Nevertheless, regardless of the undertones, Daigle wants to convey that companies must embrace AI tools to remain competitive in the long term. It’s been established that there’s a developer affinity towards AI, and the tech benefits those in the field. Organizations that recognize that AI is important and have experienced ChatGPT but don’t accept its use by their workforce are doing their firms a disservice.

“It’s not just going to be enough to be able to write code,” he said. “It’s also going to be about how we solve some of the hardest problems that weren’t economically viable before because now you can do it faster, more productively, and ultimately keep your developers happy, which we think will have a longer-term impact for those companies.”

▶️ Read more about GitHub’s 2024 AI in Software Development survey

Leave a Reply

Discover more from Ken Yeung

Subscribe now to keep reading and get access to the full archive.

Continue reading