Ai Coding Assistants Don't Boost Productivity Or Prevent Burnout: Study

By Author: jamescolin
Total Articles: 215
Comment this article

Generative artificial intelligence (GenAI) has made remarkable strides in recent years, particularly in fields like software development. Tools such as GitHub’s Copilot and OpenAI’s GPT models were designed to streamline workflows by automating routine coding tasks, enabling developers to focus on higher-level problem-solving. The promise of GenAI was clear: increased productivity, faster task completion, and reduced frustration from tedious, repetitive coding work.

According to a 2022 study, 88% of developers who used GitHub’s Copilot reported feeling more productive. This tool, which integrates AI to help developers generate code and offer intelligent suggestions, was seen as particularly effective for repetitive tasks. In fact, 96% of users said they completed repetitive tasks faster with Copilot, while 77% reported spending less time searching for solutions. Many developers also found that using Copilot allowed them to get into a more productive flow, with 73% stating that they were more in the zone while coding.

Additionally, 60% of developers felt more fulfilled in their jobs when using AI tools like ...
... Copilot, and 59% reported feeling less frustrated. One developer noted that AI tools allow them to focus on more creative and intellectually stimulating aspects of their work, making coding a more enjoyable experience overall.

However, a separate study conducted by Uplevel Data Labs tells a different story. Uplevel’s analysis, which surveyed 800 developers, found that teams using Copilot did not experience significant improvements in key performance metrics like pull request cycle time, throughput, or bug rates. In fact, code generated by Copilot was found to have 41% more bugs than code written by human developers. This suggests that while AI tools may speed up certain tasks, they can introduce errors that offset the time saved.

Read More: https://www.techdogs.com/tech-news/td-newsdesk/ai-coding-assistants-dont-boost-productivity-or-prevent-burnout-study