India poised to lead the world in the age of AI, says GitHub CEO

"I'm optimistic about India's future. There's work to be done, but you have the opportunity to become the leading nation in software and AI," said GitHub chief executive Thomas Dohmke in an interview with Mint on the sidelines of the GitHub Galaxy event in Bengaluru on Tuesday.

The age of AI

Amid a global shift to AI, India boasts over 15.4 million developers on GitHub, compared to a little over 21 million in the US. Globally, there are more than 100 million developers. 

Read This: The AI-generated population is here, and they're ready to work

Dohmke believes this large pool of developers will significantly benefit India, with more individuals leveraging AI to advance companies and society. "Last year was all about talking about AI. This year it is all about scaling and adopting AI…India will not only be a global leader but the leader in the age of AI," he said. This was Dohmke's first visit to India as GitHub's CEO.

While GitHub defines developers as "anyone with a GitHub account," India is currently estimated to have 4-5.5 million software developers out of about 30 million worldwide. 

Dohmke is confident that India's vast population and the increasing number of graduates in computer science and other engineering disciplines will ensure a sustained surge of developers. "People of all ages will learn to code in their native languages (with generative AI)…There will be no catching up with India for a long time," he remarked. 

Productivity Boost

GitHub was founded in 2007 by Chris Wanstrath, P.J. Hyett, Tom Preston-Werner, and Scott Chacon. Dohmke, who co-founded HockeyApp in 2011, joined Microsoft when it acquired his company three years later. 

Microsoft acquired GitHub for $7.5 billion in 2018, and Dohmke became CEO of the code-sharing platform in November 2021, replacing Nat Freidman. 

GitHub, according to Dohmke, maintains "full independence" regarding its product, design, roadmap, branding, marketing, and events like the GitHub Galaxy. However, its integration with Microsoft helps scale GitHub and leverage Microsoft's "copilot stack." Dohmke noted, "Developers using Copilot are already 55% faster, propelling the software economy, valued in the trillions globally, to move even quicker thanks to AI."

Students can access GitHub for free. Individuals pay $10 per user per month, while businesses pay $19 per user per month. The "most premium users can access a Copilot enterprise for $39 per user per month," allowing enterprises to customize the Copilot for their needs, Dohmke explained. 

GitHub Copilot has 1.8 million paid subscribers and has been adopted by over 50,000 organizations globally. In India, companies like Cognizant, MakeMyTrip, and Paytm use Copilot. 

More Here: Artificial general intelligence could reboot India’s prospects

For instance, about 35,000 Cognizant developers are trained on GitHub Copilot, with an additional 40,000 developers slated for training. According to Prasad Sankaran, EVP of software and platform engineering at Cognizant, GitHub Copilot has helped the company become more productive, improve software quality, and speed up software shipments.

Infosys, too, uses GitHub Copilot to help its developers "become more productive, efficient, and enabling them to focus more on value creating tasks", according to Mohammed Rafee Tarafdar, CTO at the Indian IT services firm. “Generative AI is transforming every aspect of the software development lifecycle, and using Infosys Topaz assets, we are accelerating Gen AI adoption for our clients," he said in a media statement on Tuesday.

Paytm used to monitor its expanding public cloud infrastructure with third-party tools, impacting productivity and raising security concerns. Using GitHub Copilot, Paytm launched ‘Code Armor,’ "a solution revolutionizing the way we approached security and development," according to Vijay Shekhar Sharma, founder and CEO of Paytm. 

Similarly, MakeMyTrip uses GitHub Copilot to improve productivity. Sanjay Mohan, Group CTO at the online travel firm, also present at the event, said, "The coders are spared the boredom of coding up mundane stuff, freeing up time for higher-order problem solving, which is core to our business of travel."

AI innovations and open source community

Dohmke believes the Indian open source community's creativity will thrive, leading to more AI innovations and startups. Although GitHub is not an open-source company, Dohmke explained, "Where the top of the [software] stack is closed source, everything within the stack is open source." 

Also This: How Fractal, India’s first AI unicorn, is prepping for the next race

Millions of developers worldwide collaborate using code and natural language to build the future. For instance, the Open Healthcare Network enhances healthcare delivery and management with Copilot's help. Srikanth Nadhamuni, founder CTO of Aadhaar and chairman of 10BedICU, an initiative under the eGov Foundation, noted that the 10BedICU project, now implemented in over 200 hospitals, has "significantly improved CARE EMR with the help of GitHub Copilot," enhancing productivity in coding, documentation, and testing.

To be sure, GitHub competes with similar platforms, including GitLab (in which Google's parent Alphabet has a small stake), GitKraken, SourceTree, Amazon Code Whisperer, IBM Code Net, Meta's Code Compose, Tabnine, and Codeium. Despite this competition, GitHub holds about 79% market share, according to research firm 6Sense. 

"You wouldn't want to watch a cricket game if there's only one team in the league. It's much more fun to compete when you have competitors that push you forward," said Dohmke. “We were the first to build a copilot, and so given that we have four years of experience in this, we have you know done a lot of research and have developed the product over time into what we call the Copilot-powered developer platform."

Automation and job losses?

Dohmke addressed concerns about automation and job losses, noting that while AI will write 80% of the code, developers will still be needed to write descriptions in human language. 

"We have invented programming language compilers to write 100% of the machine language, and open source runs 90% of all applications today. So, while AI will write 80% of the code, we will still need developers to write descriptions in human language," he insisted. 

Despite AI advancements, the world still relies on legacy code, such as banks running COBOL from the 60s, Dohmke highlighted, necessitating ongoing modernization efforts by developers.

Most developers will "become system thinkers -- they will tackle large problems that AI cannot handle today, like building a Facebook, or build(ing) a self-driving car, or build(ing) a car for that matter," said Dohmke. 

Understanding programming languages remains crucial in Systems Thinking, which "you can learn with the help of a copilot, but you still need to understand things like coding and databases." 

He added that prompt engineering or AI engineering will help use large language models to their fullest potential. If the output isn't satisfactory, you can adjust the prompt and instruct the model to perform differently.

However, even prompt engineering is becoming automated. "But the model still needs inputs from the user," argued Dohmke. "There's no AI company as yet that has a sentient model; and there's no AI company that is creative in the sense of the model creating by itself - it needs an operator because the neural network today does not have the skill of a human".

Dohmke emphasized that companies will need prompt engineering to reskill workers to use AI models effectively. "This will become an essential skill, helping us to learn how to use AI like we learn to use social media or build web pages," he said. Dohmke concluded optimistically: “Kids of today need not be concerned about the incredible pace at which AI is progressing. Rather, they should be more excited than ever about the future."

Also Read: Mint Primer | An OpenAI search engine? Decoding the growing buzz