Hacker News Reader

Comment by dimitri-koenig | Hacker News Reader

Good point, thx! An "About" page answering your questions could be helpful.

In short, my goal is to read and understand current science papers without needing deep knowledge of the specific jargon.

Papers are chosen based on their Creative Commons license, allowing modifications. Accuracy isn't verified, but a disclaimer notes that these are AI summaries and recommends checking the original source for accuracy.

View on HN

The idea is good, but the real difficulty is that recent publications do not yet represent established knowledge but are rather part of an ongoing scientific discussion. Now without being part of the respective community it is therefore very difficult to correctly judge the publications. If you look at semantic scholar, they're trying to classify publications as "influential" or "background", based on how often they're cited.

Hm, i get your point. My take on this is the following:

Every established knowledge started of as cutting edge. Some took way longer to be accepted and respected. Sometimes decades. So I like that cutting edge science and ideas are put out there. Only time will tell what ideas are really good.

I think a way around this would be some sort of voting/ popularity system?

Papers with code (https://paperswithcode.com/) does this via Github stars sorting.

Sure it doesn't mean something is established. But at least it gives some way to filter through the firehose of papers.

Love this project btw! I think it has potential (and the timing is right now that everyone is looking for the next "attention is all you need")

This might be an overly pedantic nitpick - and it might just be me - but I found the phrase "the newest science" very confusing at first. It sounded like it was referring to the latest "thing" which has come to be called a science, e.g. "the science of sleep". Perhaps something like "Cutting edge science explained simply" ?

"Cutting edge science explained simply" - I like that. Done!

The summaries are great. What model/process do you use to generate them?

Is it all automated?

I use OpenAI GPT-4o mini and have automated everything since for this amount of articles it would be overwhelming to do it manually. AND i wanted the translations so imagine every article is present in 6 languages. That has to be automated.

Reply on news.ycombinator.com