Developers are attached to tools because tools encode trust

4 days ago

Traditional tools like Vim and Emacs gain trust through deep customization and muscle memory, making them feel like extensions of the developer.
AI coding agents speed up output but reduce trust; usage rises while trust falls (from 40% to 29% in a recent survey).
New AI tools require a shift in process and culture, not just tool adoption; a broken process remains broken even with better tools.
Code becomes cheaper to produce, but validation (code review, infrastructure cost, failure risk) becomes the new bottleneck.
Humans must remain responsible for code; AI contributions should be flagged, and the human pushing the commit owns the outcome.
Collaboration risks siloing—AI lets developers bypass designers and domain experts, leading to massive PRs without proper checks.
Explicit context and prompts are essential; leaving anything to chance means it may be handled incorrectly (e.g., platform-specific builds).
The DRY principle fights AI’s tendency to rewrite code (WET), promoting reuse of verified components.
Knowing when *not* to use AI (e.g., using deterministic bash scripts) is a key trust-building practice.
Building trust in AI-driven processes requires preserving human judgment, clear workflows, and continuous feedback loops.

AI has transitioned from an assistant to an essential tool for 84% of developers, with 46% of code in Copilot-enabled files being AI-generated as of early 2023.
The shift from 'all-you-can-eat' tokens to metered, corporate models is imminent, with metrics focusing on token cost per feature and trust in generated code.
Trust in AI-generated code is a critical issue, especially in non-expert areas, where developers may plant 'time bombs' unknowingly due to inability to verify code fully.

The author initially doubted AI's coding abilities, treating it like Stack Overflow and manually reviewing code to maintain personal style.
Claude Code demonstrated AI's ability to handle maintenance tasks and reduce the need to learn specific frameworks like Laravel.
Newer AI models (e.g., Fable, GPT 5.6-Sol) now outperform the author in 80% of coding tasks, leading to acceptance and excitement about productivity gains.

AI-driven development is breaking traditional code review: faster code generation caps velocity at slow human reviews.
Asynch diff-centered code review, ~20 years old, fails in an era where writing code is cheap but reviewing isn't.
Diffs lack context, requiring reviewers to reconstruct intent and design, adding expensive overhead.

Invest in upfront alignment by making implicit assumptions explicit in the initial prompt to avoid unexpected AI outputs.
Restart with better context rather than steering AI when initial outputs are far off; avoid compounding errors through path dependence.
Equip AI agents with the same tools you use (CLI, MCP, browser control) to handle setup, testing, and instrumentation tasks.

AI tools can rapidly create working prototypes, but the distance from a prototype to a production-grade system remains significant.
Building production-grade software requires judgment, system design, error handling, security, scalability, and observability—areas AI does not automate.
Learning computer science is more valuable than ever because it provides a mental model to evaluate AI-generated code for flaws like performance issues or race conditions.

AI has transitioned from an assistant to an essential tool for 84% of developers, with 46% of code in Copilot-enabled files being AI-generated as of early 2023.
The shift from 'all-you-can-eat' tokens to metered, corporate models is imminent, with metrics focusing on token cost per feature and trust in generated code.
Trust in AI-generated code is a critical issue, especially in non-expert areas, where developers may plant 'time bombs' unknowingly due to inability to verify code fully.

The author initially doubted AI's coding abilities, treating it like Stack Overflow and manually reviewing code to maintain personal style.
Claude Code demonstrated AI's ability to handle maintenance tasks and reduce the need to learn specific frameworks like Laravel.
Newer AI models (e.g., Fable, GPT 5.6-Sol) now outperform the author in 80% of coding tasks, leading to acceptance and excitement about productivity gains.

AI-driven development is breaking traditional code review: faster code generation caps velocity at slow human reviews.
Asynch diff-centered code review, ~20 years old, fails in an era where writing code is cheap but reviewing isn't.
Diffs lack context, requiring reviewers to reconstruct intent and design, adding expensive overhead.

Invest in upfront alignment by making implicit assumptions explicit in the initial prompt to avoid unexpected AI outputs.
Restart with better context rather than steering AI when initial outputs are far off; avoid compounding errors through path dependence.
Equip AI agents with the same tools you use (CLI, MCP, browser control) to handle setup, testing, and instrumentation tasks.

AI tools can rapidly create working prototypes, but the distance from a prototype to a production-grade system remains significant.
Building production-grade software requires judgment, system design, error handling, security, scalability, and observability—areas AI does not automate.
Learning computer science is more valuable than ever because it provides a mental model to evaluate AI-generated code for flaws like performance issues or race conditions.

Hasty Briefsbeta