TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Lessons from a year of AI research

127 pointsby peanutcrisisover 3 years ago

13 comments

andreykover 3 years ago
Good list! As a PhD student and therefore AI researcher for a few years now, a lot of this rings true. Though 100 lessons is too much and some of these are obvious&#x2F;minor, i&#x27;d distill it down to the main ones.<p>Here&#x27;s my 2 cents on the topic from a thing I wrote last year (&#x27;Lessons Learned the Hard Way in Grad School (so far)&#x27;): <a href="https:&#x2F;&#x2F;www.andreykurenkov.com&#x2F;writing&#x2F;life&#x2F;lessons-learned-from-failures&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.andreykurenkov.com&#x2F;writing&#x2F;life&#x2F;lessons-learned-...</a>
评论 #29409152 未加载
visargaover 3 years ago
The one that matters is missing: how do you get the best random seeds? &#x2F;s
评论 #29403679 未加载
评论 #29402619 未加载
评论 #29402845 未加载
评论 #29402764 未加载
joleyjover 3 years ago
That site hijacks the CMD+LEFT and CMD+RIGHT Mac hotkeys for browser BACK&#x2F;FORWARD and also has very strange behavior when scrolling with the keyboard. Why? Why? Why?
评论 #29406036 未加载
评论 #29403430 未加载
评论 #29405607 未加载
go_elmoover 3 years ago
Find peace in doing trial and error statistical black box studies after x years of formal systems studies. I couldnt do that for my part.
hutrdvnjover 3 years ago
The title here and on the blog differs, I don&#x27;t know why submiters pick other titles. I mean I wouldn&#x27;t quote a newspaper or paper with a wrong title, would I?
评论 #29404365 未加载
criddellover 3 years ago
Do AI researchers have a commonly accepted definition for <i>intelligence</i>?
评论 #29409186 未加载
评论 #29407491 未加载
评论 #29407588 未加载
tinyhouseover 3 years ago
101st lesson - 100 is a nice number but no one is going to read a blog post with 100 lessons learned. Either focus on the 5-10 most important lessons or consolidate and summarize.
mark_l_watsonover 3 years ago
The author might find that this well thought out list will help when applying for work. I would suggest that they copy it over to GitHub in addition to their public projects.
YeGoblynQueenneover 3 years ago
Original post:<p><a href="https:&#x2F;&#x2F;jetnew.io&#x2F;blog&#x2F;2021&#x2F;100-lessons&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jetnew.io&#x2F;blog&#x2F;2021&#x2F;100-lessons&#x2F;</a>
solmagover 3 years ago
How do you formally verify your neural networks and the like? Or is formal verification possible? Does this limit the areas where it can be applied?
engineer_22over 3 years ago
-&gt; 87. Don&#x27;t let yourself be too affected by the opportunity costs of doing research.<p>Why not? Is there a strong payoff later? How did you learn this lesson?<p>:)<p>Good listicle!
评论 #29426056 未加载
reliableturingover 3 years ago
Thanks for the write up! As someone starting a doctoral degree early next year, I greatly appreciate it
tubby12345over 3 years ago
Lol this is so aspirational it could only come from an undergrad.<p>Let me tell you that I&#x27;ve finally made it to the stressful part of the being a serious &quot;AI&quot; researcher, where I have a real project (as in difficult to achieve goals, not just &quot;turn the crank&quot; stuff) and real deadlines (deliverables on collaborators projects and my own conferences submissions) and the <i>only</i> thing I prioritize above doing the work itself is keeping my advisor (and other collaborators) up to date on what I&#x27;m doing so that when he reads my paper draft he&#x27;s not completely lost. Everything like organizing papers, citations, logging infra, etc is meaningless when you&#x27;re trying to piece together a solution. Like seriously somedays I barely have time to exercise and eat dinner with my wife (let alone organizing my bookmarks).<p>For example I&#x27;m trying to solve a particular compilers problem using integer programming (note that at a high this isn&#x27;t that high level because this is a small cottage industry) and so I have like 50 paper tabs open that I bounce between when thinking&#x2F;experimenting. The way it usually goes is I&#x27;ll hack, get stuck, go back to the papers, find something, hack, and on. And usually the eureka moment comes some hours later because I connect something.<p>You might say that I&#x27;m a bad researcher but I know for a fact (external validation) that I&#x27;m not. And if you look at other highly productive researchers (like TT track profs at my &quot;elite&quot; school) this is indeed how they work. All of this zotero, notion, mlflow stuff is of the ilk of productivity porn for other flavors of knowledge workers (ie a mirage and&#x2F;or snake oil). Let me put it this way: my advisor is a top 500 h-index person (the exact significance of that metric notwithstanding) and he doesn&#x27;t have a bibtex of his own papers, let alone zotero for all of the papers he reads&#x2F;comes across.<p>The only thing that matters is code&#x2F;math&#x2F;etc output (whatever your material output is) and your abilities are also highly correlated with it with the casualty flowing in t opposite direction (make more stuff and you&#x27;ll get better at making stuff).<p>But I guess conversely do do some of these things when you&#x27;re young and have the time (and I don&#x27;t mean that condescendingly). E.g. reading outside of your area is probably the most valuable (from my own, admittedly a typical, experience, since I jumped domains many times); I very frequently can outpace even my senior colaborators very quickly on understanding a problem and solution simply because when I was younger I dabbled in ... all the things (physics, math, cs).<p>The other thing that I&#x27;ll say is there&#x27;s something obviously missing from this list but only if you&#x27;ve really made it this far: collaborators and interactions with collaborators. The only thing that matters aside from the produce is getting people to make use of it. That means writing, speaking, and getting buyin from your collaborators. If you really truly want to be successful then work on your people skills as it pertains to this area - that means learn to speak the language of your research community, learn to give good (engaging, interesting, useful) presentations, learn to write well (including making nice diagrams), and learn to explain things in ways that smart but busy people will understand. Besides all of this being key to being productive it&#x27;s also what feeds you (i.e. the real #1 priority) since it gets you jobs, academic and industry.
评论 #29416822 未加载