In response to a question about the feasibility of effective code reviews for large (e.g., 500-line) AI-generated PRs like those from Claude, especially when reviewers lack deep codebase familiarity in new projects or fast-paced environments, Uncle Bob Martin and Grady Booch have contrasting views Uncle Bob Martin advocates metrics-based oversight (test coverage, complexity, dependencies) and higher-level management over line-by-line AI code review, while Grady Booch stresses manual verification for vulnerabilities, dead code, and performance factors. Uncle Bob Martin : " I don’t review code written by agents . I measure things like test coverage, dependency structure, cyclomatic complexity, module sizes, mutation testing, etc. Much can be inferred about the quality of the code from those metrics. The code itself I leave to the AI. Humans are slow at code. To get productivity we humans need to disengage from code and manage from a higher level." Grady Booch : "Unlike B...
This Week I Learned - * From The Batch - - Alignment training teaches LLMs to behave like assistants, but it tethers them to that behavior only loosely. Beyond alignment training, system prompts act as behavioral guardrails, but motivated users can bypass them. - The lengths of tasks completed autonomously by AI agents have doubled roughly every seven months, according to METR, an independent testing organization - LLMs' knowledge is still relatively limited with respect to infrastructure and the complex tradeoffs good engineers must make...finding infrastructure bugs — say, a subtle network misconfiguration — can be incredibly difficult and requires deep engineering expertise. - Research involves thinking through new ideas, formulating hypotheses, running experiments, interpreting them to potentially modify the hypotheses, and iterating until we reach conclusions. Coding agents can speed up the pace at which we can write research code. - Meta has pivoted from its ope...
This Week I Learned - * AI is influencing not only how fast code is written, but which languages and tools developers use. - Octoverse 2025 * GitHub Copilot CLI for Beginners - * Apple's Support app update v5.13 accidentally included CLAUDE.md files , which are configuration documents used to guide Anthropic's Claude Code AI coding assistant on codebase architecture and rules. Apple quickly released v5.13.1 to remove the files, exposing a rare view into their use of third-party AI tools for app development while highlighting a build process oversight at the highly secretive company. * Claude sometimes refuses or postpones work based on inferred time/context, possibly to manage token limits, avoid rushed outputs, or simulate thoughtful behavior. * As A.I. makes the production of knowledge work more and more efficient, the job of presenting, debating, lobbying, arm-twisting, reassuring or just plain selling the work appears to be rising in importance. ...
Comments
Post a Comment