Submission + - I Canceled My Cursor Subscription (linkedin.com)
BitterEpic writes: I prefer Vim. Not VS Code. Not IntelliJ. Not Atom. Not Sublime Text. Not whatever tool is popular this year or being pushed as mandatory. I've seen so many come an go.
Vim has been my primary editor for decades.
My first professional role was at Slalom. I used Vim with no completion while surrounded by IDE users. Despite that, I consistently delivered strong results and high client satisfaction.
I've found that when you get surrounded by words like "Intellisense" and IDE and other jargon in your working environment, it's secret speak for the source code not being maintainable by humans without special proprietary tools. In my experience, the more advanced an editor becomes, the harder it is to understand what's going on. And yes, AI is becoming another layer of that. So why am I not all-in on AI tools for engineering?
When using Copilot-style completion, it often feels like rolling dice and hoping the model guesses what you want next. It might impress you occasionally, but in practice it does not reliably save time. That limitation is not entirely the AI’s fault. It simply lacks sufficient context to predict intent at that level.
Conversely, Cursor should, in theory, have enough context if you provide detailed instructions. In practice, I rarely use its output. Too often it is incorrect, and I still end up consulting official documentation to fix the result.
So where is the time savings?
My preferred workflow is grounded in primary documentation sources, where behavior and expectations can be verified. AI adds another layer of indirection between me and the source of truth. Unlike a human collaborator, the tool does not meaningfully learn from repeated corrections in a way that improves trust over time.
This raises a broader question about the growing stack of abstractions and tooling that require increasingly specialized roles to manage. There is a cost.
There is also an ethics dimension. In formal software engineering education, ethics are a core topic. Engineers sometimes work on systems where failures can cost lives. When AI is part of a professional workflow, where do we draw the line on responsibility? At what point does engineering ethics become important again?
For those curious
One of my primary languages in Vim was C with GTK. I would not choose that stack again for a project because of the heavy boilerplate. Selecting the right language and libraries is often the most impactful decision for reducing complexity and time to delivery. Understanding what is busy work is one of the most important parts of engineering.
Slashdotters what do you think? Do you think there is a responsibility issue with engineers using AI tools?
Vim has been my primary editor for decades.
My first professional role was at Slalom. I used Vim with no completion while surrounded by IDE users. Despite that, I consistently delivered strong results and high client satisfaction.
I've found that when you get surrounded by words like "Intellisense" and IDE and other jargon in your working environment, it's secret speak for the source code not being maintainable by humans without special proprietary tools. In my experience, the more advanced an editor becomes, the harder it is to understand what's going on. And yes, AI is becoming another layer of that. So why am I not all-in on AI tools for engineering?
When using Copilot-style completion, it often feels like rolling dice and hoping the model guesses what you want next. It might impress you occasionally, but in practice it does not reliably save time. That limitation is not entirely the AI’s fault. It simply lacks sufficient context to predict intent at that level.
Conversely, Cursor should, in theory, have enough context if you provide detailed instructions. In practice, I rarely use its output. Too often it is incorrect, and I still end up consulting official documentation to fix the result.
So where is the time savings?
My preferred workflow is grounded in primary documentation sources, where behavior and expectations can be verified. AI adds another layer of indirection between me and the source of truth. Unlike a human collaborator, the tool does not meaningfully learn from repeated corrections in a way that improves trust over time.
This raises a broader question about the growing stack of abstractions and tooling that require increasingly specialized roles to manage. There is a cost.
There is also an ethics dimension. In formal software engineering education, ethics are a core topic. Engineers sometimes work on systems where failures can cost lives. When AI is part of a professional workflow, where do we draw the line on responsibility? At what point does engineering ethics become important again?
For those curious
One of my primary languages in Vim was C with GTK. I would not choose that stack again for a project because of the heavy boilerplate. Selecting the right language and libraries is often the most impactful decision for reducing complexity and time to delivery. Understanding what is busy work is one of the most important parts of engineering.
Slashdotters what do you think? Do you think there is a responsibility issue with engineers using AI tools?
