The rise of large language AI models is changing the programming landscape, creating exciting opportunities for AI developers and software engineers alike. At a recent roundtable hosted by GitLab, Simon Dawson, head of engineering at Atom Bank, shared insights on how AI can help train junior developers. He emphasized the need for more senior software engineers but highlighted the organization’s commitment to nurturing junior talent. “We want to help them become senior developers sooner,” he said, acknowledging the challenges but also the potential of AI for skill development.
Dawson noted that while there are hopes around AI technology, it’s hard to predict all the outcomes. He pointed out, “There will be unintended consequences of AI that we don’t know yet.”
Kishor Toshniwal, the enterprise architect at Community Fibre, shared a personal experience. He recently returned to coding with the help of AI while developing scanning software for the company’s equipment. “I had no idea how to scan using a phone’s camera. Neither did my developers. But with an AI code generator, we found the right code and implemented it.” This experience reignited his passion for coding after 15 years away. “Now, I can dedicate 25% of my work time to software development,” he said, noticing a trend of senior managers in tech getting back into coding thanks to AI.
David DeSanto, GitLab’s chief product officer, echoed this sentiment, sharing that even their VP of engineering writes code. At the recent Microsoft AI Tour in London, CEO Satya Nadella talked about using GitHub’s AI tool, Copilot, to finish personal projects on weekends. “I joke that I can wrap up a project in no time,” he said, emphasizing the practical benefits of AI.
Dawson’s focus then shifted to measuring AI’s impact. He acknowledged that tracking the benefits of AI can be tricky. “It often feels subjective,” he admitted. But he suggested an approach: compare how long it takes to complete tasks with and without AI assistance. He noted that using Google’s Gemini AI for office productivity had improved engagement during meetings by enhancing note-taking and summarization.
DeSanto addressed concerns about productivity measurement. “Forget the vanity metrics, like counting lines of code,” he said. Instead, he encouraged IT managers to seek quality over quantity. He added that Duo, GitLab’s AI tool, helps new developers get up to speed faster and automates operational tasks, allowing teams to tackle more complex issues.
Despite the clear advantages of AI, Dawson raised a cautionary note about its societal impact. He worries that relying too much on technology might dull human intelligence. “We learn by doing,” he stressed. If AI replaces hands-on experiences for new software developers, they might miss key learning moments needed for their growth. But there’s also a silver lining: senior managers now have a path back to coding, filling in their knowledge gaps with AI as their guide.