GitHub COO Kyle Daigle urges software developers to look beyond coding

Developers are expected not only to learn coding but also to oversee what the AI agents are doing

Software developers of the future need to be armed with skills such as problem-solving, clear communication, and a growth mindset at a time when a lot of coding is getting automated with generative artificial intelligence (GenAI), said Kyle Daigle, chief operating officer (COO) of GitHub.

 

Developers are expected to not only just learn coding but also oversee what the AI agents are doing — popularly known as human-in-the-loop — to ensure that the codes are correct, without biases, and are reliable.

 

“Learning how to code provides me with the ability to write codes, and it’s also teaching me (many) more important things. That is how the system works,” Daigle said in an exclusive interaction with Business Standard in his maiden trip to India. “Tools will make me more powerful as a developer. For open source, it allows us to solve problems that were financially impossible before,” he added.

 

Problem-solving, while sounds simple, is a lot more complex skill to acquire. Some of the critical questions include what is the need to develop the software, how to do it, and how we can fail. Besides this, open communication between developers is also important. And developers need to imbibe a growth mindset, which means having the eagerness to learn something new, not being afraid of change. “This technology is changing so fast that it is difficult to predict three months in advance,” Daigle said.

 

The GitHub COO said he does not believe automation will reduce the number of developers in India, which has the second largest developer community for GitHub after the US. The US company expects India to be its largest developer base by 2028. India currently has about 18 million open source developers with about a million being added every quarter.

 

“With our audacious goal of one billion developers, the only road to it is through India,” Daigle added.

 

Asked of the reasons why GenAI adoption has not been on a par with expectations, Daigle said a lot of it can be attributed to the AI use cases. Experts have attributed it to a host of factors, including change management and the problems that were being aimed to be solved using GenAI.

 

“It becomes an issue when you try to bring GenAI in these workflows. It requires me to change the way I am working or it forces me into a chat user experience. And I think that we underestimate the pain of forcing someone to think differently, AI has to help us in doing the work we are already doing and not trying us to conform to it (sic),” he said.

 

GitHub has integrated AI models from Anthropic and Alphabet’s Google into a coding assistant used by millions of software developers. The models will be integrated into the main part of GitHub Copilot assistant, which can churn out codes with prompts.

 

The biggest gap, according to the COO of the Microsoft-owned company, in AI is user experience and a vast majority of AI experience are still chat windows. “How could that possibly be the best way to engage with these enormous Large Language Models (LLMs),” he asked.