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。

Ask HN: Given advent of AI driven coding, would you still recommend the field?

1 点作者 lifeplusplus超过 2 年前
I feel I&#x27;ve been underestimating the progress of A.I. in last few years. But that all changed when I interacted with text to image generators. I wrote few prompts, and it exceeded my expectations, and as someone who used to be in painting&#x2F;drawing, I can see how many hours of hours it&#x27;d taken me or even a professional to accomplish this.<p>Sure end result wasn&#x27;t 100% logical, but it was MORE than enough. And creativeness and variety more than made up for it.<p>I&#x27;m thinking in near future you could have 1 senior engineer vs whole team. And this engineer simply writes very detailed prompts. Generates 20 copies of code, and pick one from them. Edit them a bit here and there, and boom work of 11 developers done in a day.<p>I know it sounds far fetched but I didn&#x27;t think A.I. could take few words I typed and generate an image even better than I had imagined, with more details that I hadn&#x27;t even mentioned automatically added.<p>Given this, is software engineering viable career for people starting out right now. I think we are ok for next 10 years?

1 comment

rossdavidh超过 2 年前
In the late 90&#x27;s, I did my master&#x27;s thesis on neural networks, and I have been checking in on the machine learning (which is a better description than &quot;artificial intelligence&quot;) field every year or two since then.<p>I&#x27;m not worried.<p>In addition to the many, many other reasons, the fact is that the most important and difficult part of programming, now and for the last few decades, is deciding exactly what it is you want the program to do. Nothing is more common than doing exactly what the customer asked for, and they don&#x27;t like the result. Programming is largely about helping the customer to figure out what they actually want the program to do for them, and we are many decades away from machine learning being able to help with that.