TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Five Theses

4 点作者 lartin_muther超过 1 年前
1. Things besides “more code” need to happen for a good tech product to be built and released (ideation, quality assurance)<p>2. The market’s tunnel vision around generative AI tools (more code! more art!) creates an ecosystem-wide effect whereby emphasis and usage of tools for other things is boxed out. People have certainly built a ton of great AI tools for design, documentation, and dev tools — things that suggest bug fixes&#x2F;refactors as you write, summarize dense onboarding docs, and create test data that looks like production data, for example — and it doesn’t mean companies will use them<p>3. Focusing on fixing the simple yet time-consuming problem of “more code! where is my more code?!!?” will balloon the responsibilities of these other areas, whose demands can grow exponentially (because complexity can, especially in software products)<p>4. Companies have thus left themselves with a perfect storm: (a) buckets of new code, (b) smaller “leaner” teams after layoffs (so less density of institutional knowledge), and (c) a drought of investment in tools that actually help with the most difficult parts of releasing good software products<p>5. They will find (if they haven’t already) that they quickly need to hire people back, this time with AI-related skills that thanks to AI’s newness basically no one has

2 条评论

austin-cheney超过 1 年前
1. Automation (think industrialization plus firing people)<p>2. In most cases AI is the new Crypto (it’s mostly bullshit)<p>3. Focus all investment on a shorter pipeline between product imagination and pivot actions to something that actually makes money. Most businesses, especially start ups, cannot do this. This is the difference between confidence and tool users.<p>Typically that will results in less code coupled with higher refactoring efforts as necessary to achieve code durability freed from restrictions of current business requirements. It’s a paradox: code is a cost center and extreme durability costs much more, but allows flexibility you wouldn’t have otherwise. It’s like what credit is to finance.
atleastoptimal超过 1 年前
AI will progress faster than it takes for people who overinvest in AI to be proven wrong