During a presentation at the Collision conference in Toronto, Canada, GitHub CEO Thomas Dohmke emphasized the profound and enduring connection between AI and software development.
He stated that in a world where software permeates every aspect of our lives, developers now require an indispensable partner in the form of AI—a co-pilot to support and enhance their work.
During an interview following his presentation, Dohmke provided further insights when asked about his belief in the widespread adoption of AI among developers in the near future.
He emphasized the significant Fear Of Missing Out (FOMO) within companies, stating that they are closely observing their competitors and considering whether they have already embraced tools like GitHub Copilot. Dohmke emphasized that even if the adoption percentage varies, having such tools gives a competitive edge to those who have incorporated them into their workflows.
Dohmke further expressed his belief that there are no downsides to utilizing tools like Copilot, emphasizing its seamless integration into developers’ workflows.
He stated that it is becoming increasingly evident that Copilot is becoming a standard toolset for developers, comparable to the evolution from Cobalt developers to contemporary practices.
Furthermore, he highlighted that tools like Copilot will find integration throughout the entire development lifecycle, indicating their growing significance and widespread adoption across various stages of software development.
During Dohmke’s talk, it was revealed that GitHub’s Copilot, one of the earliest AI-based code completion services, continues to dominate the market as the most popular choice. However, competitors like AWS CodeWhisperer and Google’s Bard-based solutions have also gained some traction among developers.
As part of the presentation, GitHub shared new insights into how developers are utilizing Copilot. Notably, GitHub reported that, based on their analysis of nearly one million users, developers currently accept just under 30% of code suggestions.
Interestingly, the acceptance rate tends to increase over time, with developers accepting around 35% of suggestions after six months of using Copilot. Dohmke expressed that while these numbers may not undergo significant changes in the near future, a 50% acceptance rate would be a highly favorable outcome.
These findings demonstrate the growing acceptance and integration of AI-based code completion tools in developers’ workflows.
The gradual increase in acceptance rates over time suggests that developers become more comfortable and accustomed to Copilot’s suggestions as they gain experience and familiarity with the tool.
GitHub remains committed to enhancing Copilot’s capabilities and aims to further increase its acceptance rate, aiming for higher user satisfaction and productivity.
According to GitHub’s report, which was co-authored by Dohmke, Marco Iansiti, and Greg Richards from Keystone.AI, as developers become increasingly adept at utilizing AI for code generation and leveraging new models that facilitate natural language interactions throughout the development lifecycle, it is anticipated that 80% or more of code will be written with AI assistance.
This significant shift towards AI-powered development is expected to democratize software development, making it more accessible to a wider range of individuals.
GitHub also acknowledges the rapid growth of AI and the shortage of developers in this field. They highlight the potential of generative AI tools to enhance developer productivity.
The company predicts that globally, generative AI-powered developer tools will contribute $1.5 trillion to the global GDP by 2030.
Additionally, GitHub estimates that each unfilled skilled developer position will result in a $100,000 loss in GDP. By leveraging generative AI developer tools, it is believed that approximately 15 million additional developers can be effectively accommodated, thus accounting for the projected $1.5 trillion impact.
However, GitHub considers this estimate to be conservative, suggesting even greater potential impact in the future.