Hacker News Reader

TechExpert2910on AI models may be capable of self-replicating without access to a datacenter

this is ridiculous IPO bait. they're assuming:

> It appears that GLM-5.2 is able to propagate Llama3.1-8B well, which suggests that future small models of the same capability level could also complete the network enumeration... by the end of 2026

so an 8B model will be as good as a 744B one in a few months!?

---

their table of potential devices the botnet can use:

- endoscopy AI units - portable DNA sequencers - smart shopping carts - surgical robots

---

that these tiny models with limit 32k context windows can even carry out the network scouting + exploitation + inference (they never tested all 3 together)

that these models would have complex goals without prompting them to explicitly do such a thing, and that these can run autonomously. they can't.

self-replicating and doing what with what hardware?

---

a 26B Gemma model running across 15 LOCAL $55 Android TV boxes at 0.39 t/s! wow.

plus who has 15 android TVs in the same home (wth fast local LAN)!?

TechExpert2910on AI models may be capable of self-replicating without access to a datacenter

exactly. and in their table of potential devices the botnet can use:

- endoscopy AI units - portable DNA sequencers - smart shopping carts - surgical robots

this is ridiculous.

TechExpert2910on Why does Opus 5 feel worse to work with?

i think it's because they've fine-tuned an old opus 4.6 base model more and more with code

to the point where it's really great at coding, but lost its knowledge on how to write well!

if instead opus 5 was a full fresh training run, i don't think it'd be this trash at writing.

really great coding perf, but now the weights are adapted to so much code fine tuning that it writes like a weird engineer that's trying to sound smart.

TechExpert2910on Show HN: Sentient OS – On-device intelligence layer for your entire digital life

Hey! Really great question. My plan is to charge a ~$2 a month subscription if you'd like to analyze more than the last 6 months’ worth of your data (I'll also allow a one-time lifetime license option!).

And I can let you use it on this 6-month window for free because it costs me nothing per user! Your own device does all the processing :D

TechExpert2910on Show HN: Mercury – No-code orchestration for human and agent teams

looks interesting... where do you draw the line between human in the loop and agents going all in?