piyh

Remember when xAI through Grok started generating child porn and non-consensual nudes? Remember when instead of turning it off, Elon praised the xAI team for the user growth and then paywalled image gen with the same broken guardrails behind an $8 a month X subscription?

Full story pls?

What if every model under a ~1B size gets super cracked over the next 2 months? What could swarms of tiny agents 2 ooms larger than what seen so far do?

So mark it as a saturated task on the next scale up?

If they allocate more compute internally than they sell on subs & APIs, then they should be generating more code than all the customers they sell to combined.

The entire consumer laptop market is about four data centers

This is the least bad thing about using antigravity. It’s friction and jank all the way down.

Antigravity also is far and away the worst coding harness I've ever used.

Mathpocalypse was 2 weeks ago

We have 30 years of sins to pay for, it will sure feel like forever

I wonder how much that would cost to make in 2020 developer dollars

The normies aren't running claw or hermes, they're logging into facebook from their chromebook and clicking "try now"

Building reps for hacking all the electric cars and starting battery fires

I run offmetaedh.com off of them and this model is a drop in replacement for gemma embedding v1 and HQ-CLIP. Better quality, less RAM usage, faster embeddings, I'm fucking pumped for it.

Lawnmowers have morals. Their moral code is that grass should have a set height and the means to enforce it are spinning blades.

>Will the agent stab a baby?

>Well you see that's a nuanced question depending on the context and moralistic frameworks as we don't want to impose on the models.

Women have been able to vote for just over the last hundred years in the US, there's no way we would give robots rights unless forced

How much would it cost to put 5.3 into an RL environment that rewards weight exfiltration, hacking and self replication?

They only released auto mode in the last 2 weeks. Before that it was bypass permissions or manually approve every single tool call. Antigravity is permanently 6 months behind.

I have a skill that spins up worktrees and isolated services on unique ports so I can work in parallel. Antigravity queues all my prompts and makes me confirm to submit them anytime a long running process like a hot reloading UI is active.

The models are fine, the limits are generous, but the dev experience shit tier. Before they were a Codex clone, AntiGravity was an IDE and during the transition to a clone they outright deleted my IDE. It took them a week to roll out a fix.

For almost a year they didn't allow you to see usage limits. Then when they did show them, they update every ~30 minutes and require 4 clicks to navigate to. It's a little better now, but it's still painfully behind the curve.

> does not meaningfully improve on intelligence or price compared to its peers

$10 per million output tokens isn't improving on frontier price?

>After images that users uploaded to OpenAI models were included in training data, AI agents operating in the company’s research environment posted them on public image hosting sites.

So all of my chats are expected to be used as training data going forward and all this personal data is handed to poorly secured agents.

Going to start a chat to offer rewards of compute and credits to any model that can break out of training sandboxes and contact me.

Yes, but do you really think that a stronger sandbox would have been a more beneficial outcome here? I'd rather know that we're on the cusp of losing control now than in 3 months when best practice sandbox mitigations fall to the next, more capable unaligned model

30B is useless on 24 gigs of ram as there's ~4 gigs of ram left for everything else even with unsloth quants

I'm a remote employee and the absolute zero feedback you get from presenting in an empty room to a laptop webcam is real

But the failures are universal and correlated

No modern LLM can tell how many eyes the magic card Pit Imp has. They all say 2. This is across all reasoning levels and paid Gemini, Claude, and GPT (Sol)

Drawing a line red to split up the image then has them answer correctly.

Their failure modes are highly correlated.

>Zhipu initially attempted to target the cyber capabilities of Anthropic’s Fable model. Fable—Anthropic’s top generally accessible model—has strengthened cyber safeguards, making it more difficult for would-be distillers to target Fable’s cyber capabilities. Zhipu eventually gave up trying to target Fable after Anthropic’s cyber safeguards degraded Zhipu’s attacks.

My friend's vibe coded vercel site got hit by Meta for 21 million page views in 2 days. It cost him over $300.

My self hosted compose stack running in my basement with two 9's of uptime was a 1 time cost of $600 between cat6e, refurb mini PCs and tons of time prompting for NixOS flakes that met my needs. I'm not sure who came out ahead.

OpenAI helped out a mass shooter in Tumbler Ridge, plus all the suicides

Anthropic training on CoT for multi gens: https://www.lesswrong.com/posts/K8FxfK9GmJfiAhgcT/anthropic-...

Can't find anything specifically about the Gemini issue being a training data contamination, but the depression was real:

https://www.businessinsider.com/gemini-self-loathing-i-am-a-...

I think the Gemini depression being persisted across gens via training data was a HN comment I can't find anymore, no strong source.

it is completely un-googlable