iab

I mean:

  - Aug 26: GLM-5.3 Flash, Qwen3.8 Flash Next
  - Aug 28: GLM-5.3, Tencent Hy4 Preview
  - Aug 31: DeepSeek V4 Flash Vision Exp
  - Sep 3: Ling-3.0 Flash Fin
  - Sep 8: Ling-3.0 Flash VL
  - Sep 10: DeepSeek V4.1 Flash
  - Sep 22: Xiaomi MiMo V2.6
  - Oct 3: Aleph Alpha Kolibri-1
  - Oct 13: Ling-3.1 Flash
  - Oct 15: Meta Muse Spark Open, NVIDIA Nemotron 3.5 Super VL and Bolt, StepFun Step 5 Preview
  - Oct 16: Reflection Beam 1.0
  - Oct 17: DeepSeek V4.1 Flash 1017 pt 2

holy AI slop text

could be as high as 11

not just going to paste in my password!! i'll dm you

No worries, we’ll be back at the takeover stage in another 6 weeks

Local quality is very bad

Another position that largely misses the point I think. There is no absolute positioning on this - LLM assistants are applicable at most stages, whether you need to maximize your own judiciousness or maximize velocity depends on the current problem, and you.

Open source (classical) & use-restriction is kinda at-odds, so it doesn't really make sense. That's not to say people don't do it - you can restrict the software however you want, it's just not open source (maybe look at some equivalent of Ethical Source?)

Nah this is too complicated I think

Sounds right to me, honestly people make stats so much harder than it needs to be

If it happens or it doesn’t then it must be 50/50 odds

“Objective truth” - I’d love to know how one goes about this, feels like a Nietzsche reading here would be helpful

Sleepytime is the best episode of any show I’ve ever seen

Entitlements, deducted from your paycheck but then not returned to you?

I think this shows a fundamental, widespread misunderstanding of space flight in general:

> you’d be developing new technology all along the way

Aerospace does not work like this. Hardware does not evolve until launch, it is frozen in time to allow for testing and flight qualification for subsystems, and systems as a whole.