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: What angle of startups or AI isn't covered by YouTube creators?

10 点作者 scotthtaylor大约 5 年前
I&#x27;ve just started creating content on YouTube and want to know from the HN audience what you think is lacking from the videos and creators that are currently on YouTube?<p>And I say startups and AI specifically as I&#x27;ve started and sold a few startups previously. And now help lead an AI&#x2F;ML team within a large asset manager manage ~$320 billion across a number of asset classes.

3 条评论

sneeuwpopsneeuw大约 5 年前
(This is my first comment on hacker news) but I would like to see more content targeted to people who know how to program but don&#x27;t know the slang and words in this specific field. For example Sebastian Lague did a great tutorial on machine learning <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=bVQUSndDllU" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=bVQUSndDllU</a> I would love to see something like this but from the angle lets pick a random programming language and lets program this together from scratch.
keiferski大约 5 年前
In general, most content on YouTube is a bit pedestrian. I would like to see more serious academic resources referenced, longer and more in-depth videos, and better production values. Something like <i>Every Frame a Painting</i> but for other topics.
Jugurtha大约 5 年前
Everything. The state of YouTube content on ML is similar to what the state of software engineering <i>would</i> be if there were only videos on &quot;Hello, world!&quot; <i>or</i> conference talks about awesome products demoed on &quot;Hello, world!&quot; problems.<p>Practically none of the content creators seem to have ever executed a project that used ML, or worked in an ML team, and those who are doing that professionally don&#x27;t do YouTube videos, except for talks in conferences that they <i>did</i> something, not really <i>how</i> they did it.<p>I think you are in a great position and can do a lot of good by showing how <i>you and your team</i> execute a project from <i>start</i> to <i>finish</i>.<p>Seasons and episodes. One season per project. One episode every week or two that addresses the problems solved, and how. I want to look at your screen and listen to conversations and brain-storming. How do you do ML, in long form videos from the field, not some bullshit 5 minutes videos with sentences like &quot;then you only have to deploy your model&quot;. How do you collaborate on a project, how do you choose metrics that translate to the real world, how do you retrain your models, how do you do workload management, how do is your private cloud, do your ML practitioners have self service infra for training, how do you do self service deployment, manage data and access, how do you test data is not tainted, deploy and expose models, embed models.. How do you collaborate. How do you track knowledge, decisions, and assumptions and hypotheses.<p>For example, if you could take papers like <i>&quot;The ML Test Score: A Rubric for ML Production Readiness and Technical Debt Reduction&quot;</i>[^1] or <i>&quot;Hidden Technical Debt in Machine Learning Systems&quot;</i>[^2] and show how <i>your team</i> is dealing with each point, it would be great content.<p>All the thing that anyone who has spent a day in the real world knows isn&#x27;t covered by 99% of the content online because the people doing it are either too busy, consider it competitive advantage, or don&#x27;t know it even exists (had people working at FAANG try our internal platform, and they don&#x27;t even deal with these problems because they&#x27;re in &quot;research&quot; and they have teams of ops they can offload their notebook to).<p>[^1]: <a href="https:&#x2F;&#x2F;static.googleusercontent.com&#x2F;media&#x2F;research.google.com&#x2F;en&#x2F;&#x2F;pubs&#x2F;archive&#x2F;aad9f93b86b7addfea4c419b9100c6cdd26cacea.pdf" rel="nofollow">https:&#x2F;&#x2F;static.googleusercontent.com&#x2F;media&#x2F;research.google.c...</a><p>[^2]: <a href="https:&#x2F;&#x2F;papers.nips.cc&#x2F;paper&#x2F;5656-hidden-technical-debt-in-machine-learning-systems.pdf" rel="nofollow">https:&#x2F;&#x2F;papers.nips.cc&#x2F;paper&#x2F;5656-hidden-technical-debt-in-m...</a>
评论 #23049555 未加载