RSS Feeds & Google's Role

First off, there was this big article about how Google basically killed RSS feeds back in the day. Remember Google Reader? Everyone was talking about how awesome RSS is for getting your news without all the ad crap and algorithms. One guy even runs his own RSS reader site and said it's still alive, just mostly with us geeks. Someone else brought up how the EFF (Electronic Frontier Foundation) is still pushing RSS as a way to fight 'the algorithm.' It's wild how much Google shaped how we consume content, even without us realizing it sometimes.

OpenAI's Math & CS Advances

Then, OpenAI dropped this piece about ten big advances they've made in math and theoretical computer science with AI. Super high-level stuff, like arithmetic circuit complexity. But the comments were really interesting. A lot of people were like, "Is this really groundbreaking, or just AI hype?" One person was worried it might kill off a lot of education because AI just spits out answers. Another called it "reward hacking," which is a good point – AI might just be finding loopholes to look smart.

How to Exist

On a totally different note, there was this post called "How to Exist." It was pretty chill, talking about mindfulness and being present. People in the comments were sharing their own ways to de-stress, like "forest bathing" or just going for a run without music to quiet their minds. One person mentioned how our modern environment is just not compatible with our "old brains," which totally resonates, you know?

Cursor Removed Cost Information

Okay, this next one is a bit more techy. You know Cursor, that AI-powered code editor? Apparently, they just removed all the cost info from their usage page and exports. People are pretty annoyed! One commenter said that without the cost, just seeing "token burn" isn't enough to understand what you're actually paying for. Another was bummed because they feel Cursor is actively trying to kill its own advantages, like its good permissions model, and that it's becoming just a worse VS Code without the LLM features. Seems like a bad move.

ByteDance's Seedance 2.5 (AI Video)

ByteDance, the TikTok folks, released Seedance 2.5, which is their AI video generation tool. It sounds like it's getting pretty good at "one-take creation" and "flexible referencing." But the comments were a reality check. Someone said AI image generation (and probably video too) is still mostly "slop" if you want anything specific. And another person pointed out that while it's cool, the actual cost of getting a final, perfect AI-generated clip is probably way higher than just the per-second rate because you'll need so many takes to get it right. Like, it's not final quality yet.

Canada's Surveillance Treaty

Then there was a more serious one: Canada quietly signed this UN Cybercrime Convention, and people are calling it a "surveillance treaty in disguise." The article was basically saying it gives governments a lot more power to snoop around online. The comments were all about privacy and how governments are trying to catch up to the internet, but maybe in a bad way. Someone even brought up how people just believe what they want, making it hard to have these nuanced conversations about surveillance.

AI Financial Advice is Surprisingly Good

And finally, this one was pretty surprising: MIT put out a piece saying AI financial advice is "surprisingly good," especially if you ask the right questions. It's not just basic stuff, apparently. But of course, people were quick to point out the risks. One person said that while AI might maximize expected returns, it might not factor in things like reducing risk or minimizing regret, which are super important for actual humans. Someone else shared a story about a car salesman giving terrible financial advice for a lease, implying AI might be better than some human advisors, but still, be careful, you know?

Anyway, that's the rundown! Thought you'd find some of that interesting. Talk soon!