Ariel Ekgren

Building AI at @hybridity_ai
ex LLM researcher

[Ariel Ekgren @ArYoMo]

Started talking to codex for more nuanced prompts. Also moving towards "carving" out the code by asking questions and just talking to codex.

[Ariel Ekgren @ArYoMo]

What are the implications?

[Ariel Ekgren @ArYoMo]

Of course! Everything is compliance. We agree at @hybridity_ai . Compliance solved with AI is the future of how serious companies will work and accelerate.

[James da Costa @jamdac]

https://t.co/TeV8e3ndp7

[Ariel Ekgren @ArYoMo]

@romainhuet @reach_vb @Andy_AJT Come by Stockholm we also like to build things

[Ariel Ekgren @ArYoMo]

@flemscutum @francoisfleuret @rebyers I'd phrase it more like some things are computationally irreducible and some concepts in math (infinite work being okay) leads to contradictions.

[Starlink @Starlink]

The Starlink team is exploring ways to extend connectivity beyond our planet

[Ariel Ekgren @ArYoMo]

@snwy_me While I'm sure LLMs will shrink a big difference is language is path dependent. While just vibing around in a 3d space is path independent. High dimensional path dependence is vastly more complex to model would be my intuition.

[Ariel Ekgren @ArYoMo]

@simpsoka Account switching!

[Ariel Ekgren @ArYoMo]

@steipete Have you updated your context window size since you last shared it in your config in your blog post? Since compaction is so good and model is worse at long context I gravitate towards a smaller window but hard to properly evaluate what is best

[Ariel Ekgren @ArYoMo]

@thsottiaux Let us switch acc's ;)

[roon @tszzl]

not enough people are emotionally prepared for if it’s not a bubble

[Anton Osika @antonosika]

come to stockholm folks

[Ariel Ekgren @ArYoMo]

@teortaxesTex Neo Robespierre just doesn't have the right vibe

[Ariel Ekgren @ArYoMo]

The computer use is just. Very good. Very impressive.

the terminal grew hands

[Ariel Ekgren @ArYoMo]

@antonosika Now that's awesome. Starstruck!

[Ariel Ekgren @ArYoMo]

@tszzl

[Ariel Ekgren @ArYoMo]

@dwarkesh_sp Awesome! Great and surprising colab

[Daniel Hnyk @hnykda]

LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate. link below

[Ariel Ekgren @ArYoMo]

@reach_vb @DeryaTR_ @sama Is the playwright skill better then the MCP?

[Ariel Ekgren @ArYoMo]

@gfodor @bayeslord Possible but not required. Each day passing that indicates a "reproduction number" larger then 1 should also lead to vibes changing. So while I'm all for continual learning being cracked don't think it's that.