Eagerly looking forward to sending this to the next LLM-obsessed person who breathlessly asserts ai is the only future, despite having no notable skills of their own. www.colincornaby.me/2025/08/in-t...
-NaN
-NaN
@fridgeseal.xyz
Book reader, coffee-appreciater, eldritch machine whisperer, producer of 5+ abandoned side projects. Opinions are my own etc etc.
118 followers172 following261 posts
Imagine coming up with a world and aesthetic as stunning as the new #Marathon game. And then absolutely wasting it by making a boring extraction shooter. How disappointing Bungie. youtu.be/fvbEnWLRo1s
-NaN reposted
We had a game where this happened, it was called “BioShock”. I encourage anyone who thinks this a good idea to go play it and see how it turned out for everyone.
Balaji Srinivasan suggests giving Elon Musk “control of a huge swath of land surrounding Starbase, Texas” & designate that land as “a military base or … similar, so that [Musk] can fence it off & determine at his sole discretion who can enter.” Then replicate in other states w/ other Tech Bros. 1/
> lobbies for ai “safety” regulation. > applies for an exemption from the burdensome regulation they lobbied for. finance.yahoo.com/news/openai-...
Oh look a Phoronix article about Rust! I’m sure the comments on this will be ✨normal ✨and ✨sane✨! www.phoronix.com/news/NOVA-Dr...
I just read this, and I cannot agree more! > would give my kingdom for an engineering culture in which people figured out how to build good primitives and protocols 🔥🔥 100% yes. It also links Frank McSherry’s COST paper, which might be my fav paper.
Relevant: “If your service can’t be run on a laptop, your service doesn’t deserve to be ran in a cluster” - Marilyn Monroe hazelweakly.me/blog/scale-d...
-NaN reposted
Just found that @kagi.com can automatically provide associated HackerNews and Reddit on search results. I’ve always wanted something like that. This is so, so handy.
> SWE-Bench tainted by answer leakage; real pass rates significantly lower The grift continues. arxiv.org/abs/2410.06992
Hope you’re all excited to use software where you get ~~“subtle content prioritisation”~~ “fucking gaslit” by a wobbly machine consuming more energy than a small country.
For now, real-time personalization means subtle content prioritization, dynamic caching strategies, or preloading relevant data based on user intent. It doesn’t have to be an erratic, ever-changing UI nightmare. But yeah, if done poorly, it could absolutely be a disaster.
I for one am _shocked_ who could have _possibly_ seen this coming? /s www.404media.co/microsoft-st...
Link provided without comment: gradual-disempowerment.ai
Oh look, “Open”AI turned out to be lying about benchmarks again: this time it turned out they had access to the set, but verbally promised they *totally wouldn’t misuse it*. www.lesswrong.com/posts/cu2E8w...
-NaN reposted
Call me when more than 1 or 2 MP’s can functionally demonstrate an actual understanding“homomorphic encryption” and “differential privacy”. The photos already leave my device to get backed up This is such a technical non-issue.
ICYMI Apple has opted in all users to "Homomorphic-based Enhanced Visual Search". This allows it to grab your photos & geolocate landmarks in them. They say this is being done in a way that preserves privacy but it being secretly sprung on users doesn't inspire confidence.
-NaN reposted
Replying to @culturecrave.co
Examples of AI-generated profiles: www.facebook.com/hellograndpa... This one gives relationship advice: www.facebook.com/datingwithca...