Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: How do you stay on top of advances in AI?

news.ycombinator.com

Post reply on HN
Print view View original

11–20 of 43 posts

Re: Ask HN: How do you stay on top of advances in AI?

#11

Post by adides »

I run a weekly email newsletter that highlights the latest industry news (company announcements, regulations, relevant research, venture capital funding, and product launches). Link here: https://astrofeather.beehiiv.com

I also do deep dives (~700 - 800 words) on broader topics such as humanoid robots in the workforce, the recent U.S. Senate hearing on AI regulation, autonomous agents, and generative AI in advertising.

You can check out my latest issue, where I take a deeper look at generative AI in healthcare and the recent OpenAI defamation case, as well as a roundup of other recent events: https://astrofeather.beehiiv.com/p/improving-clinician-and-p....

Finally, I have a site called AstroFeather (https://www.astrofeather.com) that tracks and summarizes the latest AI news headlines on a daily basis.

Re: Ask HN: How do you stay on top of advances in AI?

#12

Post by jetnew »

I am working on a LLM-powered newsletter (https://distilleddaily.news/) that aims to solve this exact problem for myself (and targeting ML engineers, data scientists and AI researchers).

I've only started last week and have launched among friends, and so only source HackerNews for now but will quickly include curated sources: Twitter popular users' tweets (@hwchase17, @hardmaru), popular blogs (@chipro, @lilianweng), popular research papers and more.

I'm currently prioritising features to build and so would love to hear your thoughts!

Re: Ask HN: How do you stay on top of advances in AI?

#13

Post by tkgally »

Matt Wolfe’s Future Tools site is a good compilation of consumer-facing AI products and news:

https://www.futuretools.io/

https://www.futuretools.io/news

He also posts videos that I find informative:

https://youtube.com/@mreflow

I like the AI Explained video channel as well:

https://youtube.com/@ai-explained-

Re: Ask HN: How do you stay on top of advances in AI?

#15

Post by alexpls »

I try to direct as much news to RSS as possible, where I can group feeds by topic, and not get distracted by a constant compulsion to refresh the page (which I feel strongly on HN and Twitter). Some tools I use:

* https://www.freshrss.org/ for subscribing to feeds

* https://netnewswire.com/ for reading them

* https://hnrss.github.io/ to get Hacker News into RSS

* https://mailgrip.io/ (project I'm building) to get email newsletters into RSS

Re: Ask HN: How do you stay on top of advances in AI?

#17

Post by PheonixPharts »

mbm wrote: post #8
PheonixPharts wrote: post #6

I work in AI for a living, my advice would be worry less about what's the latest and greatest and focus on solving your problem, and learning the things you're interested in well. Precisely because "the rate of change in the AI space is super fast", there really isn't much of a point to keeping up, even if you're an academic researcher (for those you only need to keep up with your piece of the puzzle, and you likely…

Yes of course, but let’s assume the goal is to be conversational in what’s changing/new at the level of a Marc Andreesen, without actually having to be Marc Andreesen. I work in the space as well, but because I’m not in the Valley, find it more challenging to always be abreast of the changes. While it might not be essential for the actual product engineering to know what’s new, as a founder talking to customers it’s…

> at the level of a Marc Andreesen [sic], without actually having to be Marc Andreesen [sic].

Oh, I guess we are in a AI bubble then. There are so many wonderful thinkers right now in the AI space to aspire to be like, but the only reason to aspire to be like Marc Andreessen is because you want AI to be the next crypto.

Better find a good coat, since AI winter will be here sooner than I'd thought.

Re: Ask HN: How do you stay on top of advances in AI?

#18

Post by JoshuaDavid »

PheonixPharts wrote: post #6

I work in AI for a living, my advice would be worry less about what's the latest and greatest and focus on solving your problem, and learning the things you're interested in well. Precisely because "the rate of change in the AI space is super fast", there really isn't much of a point to keeping up, even if you're an academic researcher (for those you only need to keep up with your piece of the puzzle, and you likely…

Slightly off-topic, but I am just starting to learn about the field, and saw

> GPT4 still projects information into some highly non-linear latent space and samples from that

Do you have more information/know where I can find things to read about the "highly non-linear" part of that? I've been reading some stuff[1][2][3] about smaller models, and my impression has been that the latent space is shockingly linear for those ones. But counterexamples would be informative, especially if it's something that only happens with larger models.

-----

[1] https://arxiv.org/abs/2209.15162

[2] https://www.neelnanda.io/mechanistic-interpretability/othell...

[3] https://arxiv.org/pdf/1610.01644.pdf

Re: Ask HN: How do you stay on top of advances in AI?

#19

Post by guiambros »

Answering from the POV of someone who works with AI, but not a researcher:

1) Matt Wolfe's YT channel https://www.youtube.com/@mreflow/videos (he has also the Future Tools site -- FutureTools.io)

2) Two Minute Papers YT channel https://www.youtube.com/@TwoMinutePapers/videos

3) The Neuron Daily newsletter https://www.theneurondaily.com/

4) From a business perspective -- Stratechery https://stratechery.com/ (not AI related, but naturally touches on what key players are doing in this area. Worth the subscription price)

5) On Twitter, here's a list of AI/ML/Math folks I've been collecting over the years https://twitter.com/i/lists/230704954

Re: Ask HN: How do you stay on top of advances in AI?

#20

Post by muzani »

Build stuff. There's a lot of untested stuff. There's a lot of research that goes nowhere, or things like Bard, which are really good, but will end up like Windows Mobile.

When you build stuff, it's clearer what the strengths and weaknesses are, and which innovations matter.

There was that hype with AutoGPT and similar iterative GPTs. But when AutoGPT is tested hard, it ends up getting confused on what to do, and the price tag ends up too much. It was a nice attempt, but didn't go too far.

Langchain was extremely good at first, but then the new GPT-3.5 API chat format just allowed you to include "memory" without needing a third party to track all of it.

If you follow most of the AI thought leaders, people are still talking about prompt engineering, but that's becoming more unnecessary with something like GPT-4, which can read your mind better than your spouse. A lot of people with experience are left behind because they overinvest in these things that are just transcended months later.

Post reply on HN