Gen AI widens productivity gap between senior and junior developers

The productivity gains due to generative artificial intelligence (AI) tools in programming tasks may not be uniform across the hierarchy of developers. On average, generative AI implementation helps developers reduce their task completion time by about 40%. However, the percentage reduction was higher for mid-level (41%) and senior engineers (48%), as compared to their junior counterparts (35%), a joint study by Zinnov and Ness Technologies has found. 

The report attributes this mismatch to a range of factors. “Senior developers by virtue of their experience, have more contextual knowledge about both the project and the output required. They would have also typically worked in this engagement for a longer period of time – so what prompts to give to generate desired output also becomes easier,” Nikhil Kulkarni, Partner at Zinnov told TechCircle. To be sure, a separate report on productivity gains, by Copilot-maker GitHub in January found that developers who report a high degree of understanding of their code feel 42% more productive than those with low or no understanding. The report attributed that low understanding to factors such as poor or outdated documentation, lack of onboarding, or the sheer pace of innovation with AI.

This evolution is expected to result in a decline in the number of junior developers and a simultaneous increase in the impact and productivity or senior developers. “Generative AI implementation is set to reshape organisational structure by reducing the need for junior engineers and enhancing the role of senior developers. Its impact on simplistic tasks prompts a shift towards leaner organisation structures, necessitating a smaller base of junior engineers,” the Zinnov-Ness report added. 

It reasons that even before generative AI, junior developers offered ‘fewer story points’ (indicative of value-added output); however, due to lower unit cost, it was beneficial to have a higher base of such workers. Generative AI’s entry in programming tasks is expected to challenge the traditional cost paradigm and the team composition could evolve from a pyramid structure to diamond or flatter pod structures.

If indeed companies shift focus away from junior talent with the advent of generative AI, this may lead to an imbalance in the developer ecosystem in the longer run, said Dipyaman Sanyal, chief executive officer (CEO) of boutique technology consulting firm Dono Consulting. “After years of outsourcing jobs to developing countries like India, US-based contact centers now find it difficult to find leaders for their company who understand the nitty-gritty of this business. My only fear is, due to generative AI if such a trend spills to programming, we may not have enough capable technology leaders in future,” he said.