The speaker stopped using Copilot due to concerns about skill dependency, reduced enjoyment of coding, declining code quality, and privacy issues related to data sharing. They are considering exploring self-hosted, open-source models as an alternative to AI-driven tools in the future.
The speaker stopped using Copilot after redoing their neim config without adding it back. They realised that not using Copilot made them more conscious of their coding skills deteriorating if not maintained. The speaker noticed a dependency on Copilot, which they termed the “Copilot pause,” where they would wait for Copilot’s suggestions instead of writing the code themselves. They felt that relying on Copilot was similar to copy-pasting from Stack Overflow, hindering their creativity and problem-solving abilities.
The speaker found that using Copilot made coding less enjoyable for them and took away the satisfaction of learning and problem-solving that coding usually provided. While Copilot aims to increase developer productivity, the speaker felt less fulfilled and creative when using it. They mentioned a study by GitHub showing that Copilot increased developer satisfaction for many, but the speaker personally felt more frustrated and less engaged with coding while using it.
Another reason for not using Copilot was the decline in code quality over time. The speaker found that Copilot sometimes made mistakes, and its suggestions became outdated quickly as software development evolves rapidly. This led to the speaker often having to refer to documentation anyway, defeating the purpose of using Copilot. Research also showed that code generated by AI like Copilot often needed rewriting, raising concerns about the maintainability of the code it produced.
Privacy concerns were a significant factor in the speaker’s decision to stop using Copilot. They highlighted that Copilot constantly sends code snippets to a remote server, potentially compromising data privacy. Additionally, the default setting saves these snippets unless explicitly opted out, which the speaker found unacceptable. This privacy issue was a major drawback that solidified their choice to discontinue using Copilot.
The speaker concluded by stating that they haven’t decided on a new tool to replace Copilot yet, as they are considering taking a break from AI and exploring self-hosted, open-source models in the future. They expressed fatigue and disillusionment with the current hype surrounding AI technology, suggesting a desire to return to a simpler era of coding without AI assistance. Overall, the speaker’s decision to stop using Copilot was influenced by concerns about skill dependency, reduced enjoyment, declining code quality, and privacy issues, prompting them to seek alternative coding approaches.