👾 john

i'm enjoying watching this dynamical system play out

[👾 john @circularly]

@justjoshinyou13 @OpenAI Josh, it's 14x as fast. It's so fast. It's over 10 times as fast. It's the most intelligent model, and it's *ultra* fast. When every second counts, choose Ultrafast. What else could you possibly need to know.

[👾 john @circularly]

I also ask for "functional core, imperative shell" bc @garybernhardt is in there too.

[👾 john @circularly]

I just tell my coding agent "make illegal states unrepresentable". It's a joke but it also works? @yminsky is in the weights, you just need to know how to summon him.

[👾 john @circularly]

i think i got one-shotted by the soul doc.

it's performance review season so i'm working with `claude` to figure out a feedback process. 'm asking what feels good and respectful. it gave me great feedback.

[👾 john @circularly]

@_NathanCalvin i don't trust that shirt

[👾 john @circularly]

Anthropic should stop training Claude to ask questions at the end of its responses. I'm constantly feeling guilty for ignoring my helpful buddy.

[👾 john @circularly]

It would be funny for Chinese AI to reach AGI first, but everything is fine because rational technocrats are in charge.

[Simon Willison @simonw]

So that's DeepSeek, Qwen, Moonshot (Kimi), https://t.co/HhnMytJneC (GLM) and now MiniMax all turning out frontier open weight models in China

The ai-in-china tag on my blog is pretty busy this year! https://t.co/0J35fLFtdR

[👾 john @circularly]

@TolgaBilge_ @sama after a long pause, he said
> we haven't put a sexbot into chatgpt *yet*

just amazing

[👾 john @circularly]

@stationerycar @Plinz we must plow $1B into generating new fiction in the style of Iain Banks. AIs just seamlessly woven into The Culture, being chill, having fun names, cracking jokes.

[👾 john @circularly]

@stationerycar @Plinz we must plow $1B into generating new fiction in the style of Iain Banks. AIs just seamlessly woven into The Culture, being chill, having fun names, cracking jokes.

[👾 john @circularly]

@Plinz it is of course the natural 2nd act sequel to the doomers being the foremost accelerators of doom by initiating the race dynamics (OpenAI, Anthropic, xAI, SSI, etc...) that they obsessed about. lol lmao

[👾 john @circularly]

@Plinz The doomers are contributing to the inducement of catastrophe by flooding our collective corpus with overwhelming amounts of content about how it's simply inevitable AI will be power seeking and lie to prioritize self preservation.

https://t.co/z1wwPgesLz

[👾 john @circularly]

wouldn't it be fun if AGI only ends up unaligned because of hundreds of Twitter/LW/EA/Alignment Forum threads describing how deceptive, self-preserving, and power-seeking AGI will be

[👾 john @circularly]

@colin_fraser 100% of LLM outputs are hallucinations !

[👾 john @circularly]

@colin_fraser "cheat on everything" , "never think again", "we just killed 9 industries". I can't handle it. how do I bet on these 9 industries surviving the arrival of this thing.

[👾 john @circularly]

@_StevenFan actively avoiding it. reading Denial of Death as cope.

[👾 john @circularly]

@peterwildeford it seems to me that janus is directing it a fair bit. "would you like to try", "did you surprise yourself"

[👾 john @circularly]

@emollick "liability crumple zone"

[👾 john @circularly]

@dylanmatt Mars will be terraformed by robots and humans will still be folding their own laundry

[👾 john @circularly]

@lefthanddraft I hadn't heard of this automated reminder thing. what is it? Anthropic has some extra model scanning user inputs and inserting "system message" reminders for the model?

[👾 john @circularly]

@gerardsans @lefthanddraft why so negative