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。

Methodical Banality

42 点作者 CharlesW8 天前

5 条评论

vunderba8 天前
&gt; They internalised the qualities of their models’ prose as much as possible, and also kept a sort of verbal rainy-day fund in something called a commonplace book, which is to say that, as they read, they transcribed a stockpile of salient words, metaphors, turns of phrase and clichés – usually organised thematically – that they would then draw upon while writing.<p>I&#x27;m reminded of an excellent passage from &quot;Lives of the Great Pianists&quot; by former New York Times music critic Harold Schonberg when speaking on the strength of Beethoven&#x27;s musical extemporization:<p><i>&quot;How much Beethoven prepared his improvisations we do not know. Most pianists did prepare, knowing full well that sooner or later they would be called upon to supply an improvisation on &#x27;Batti, batti&#x27; or a similar well-known tune. And all pianists had at their command a thorough supply of passagework by the yard, which they could snip off and use for any possible contingency. But when Beethoven improvised, prepared passagework or no, it was evident to his hearers that after a while he was on his own, idea pouring after idea. Then he would get carried away, pound the piano, and the strings of the delicate Viennese instruments would pop, or hammers would break.&quot;</i><p>Speaking on a purely personal level, I am mostly unconcerned with the amount of purple prose a writer chooses to embellish their work with - because I&#x27;m always internally stripping away the literary wrapping paper to get at the heart of the idea itself.
Calwestjobs8 天前
Similar problem with law. Rules based order is not ONLY about laws, it is about ethics which is represented, imperfectly, thru law. That is why most professions which deal with humans have ethical codes, oaths, so even people who do not have same morals ( or do not care about them, but want the job ) can do job in accordance with ethic frame law is based on.<p>And if people follow only laws, without understanding of ethics behind those laws, we get very strange and oppressive society. Rules based order is system of rules which helps to make meaningful, better life for everyone while minimizing harm we do to each other.<p>So fascism, as journalist use it currently, is a lack of effort for minimization of harm we do to each other. Never heard anything scarier than that.<p>Law is not a goal but a means.
Calwestjobs8 天前
This is why communist &#x2F; soviet countries were declining. Most people get confused how can anyone label what happened in those countries as decline, they build factories, they catapulted peasants, serfs into city dwellers, they made huge scientific progress. Those people do not realize that soviet countries did it after it was done in west. For example Large Panel System building so pervasively used in communist block was invented and used 50 years before in England. Inventiveness was limited only into areas that the regime deemed necessary. And even those areas were not free to pursue scientific method, having go back and forth between photos, books and letters from relatives, from west + back and forth between &quot;government ideologist&quot; responsible for &quot;right&quot; way of thinking made any effort to move human endeavor into higher levels impossible.<p><a href="https:&#x2F;&#x2F;www.socialismrealised.eu&#x2F;normalistion-everyday-life&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.socialismrealised.eu&#x2F;normalistion-everyday-life&#x2F;</a>
评论 #44010540 未加载
jongjong7 天前
I feel like some articles tend to downplay LLMs a bit too much. It&#x27;s not just about form, LLMs do bring some substance as well.<p>When you ask coding questions it can often generate code which refers to your unique variable names. It can take what you give it and solve a unique problem which it has never seen before. Like I could give it many functions I just wrote with some names it has never seen before doing some processing it has never seen and it knows how to call those functions correctly and can even assemble them correctly, even though this particular flow is unique to this situation. There is some intelligence involved in rearranging logic and filling logical gaps which goes beyond just rephrasing ideas.<p>I feel like the main issue with LLMs is that they don&#x27;t have long term memory so it&#x27;s like talking to an extremely intelligent goldfish. The lack of ability to internalize context is what makes it hard for it to meet goals. You need to keep track of the changing state of reality in order to achieve stuff.
tim3337 天前
&gt;Today there is an easy fix: we have large language models (LLMs) to write our letters for us...<p>I&#x27;m not sure that works very well. I&#x27;ve only had people send me LLM letters a couple of times and it normally pisses me off and produces a rude reply.