LLMs and AI at Apple @Apple
Seattle/Barcelona
Joined July 2009
- Tweets13.1K
- Following3.8K
- Followers1.2K
- Likes77K
David Güera retweeted
Para que ocurra un accidente ferroviario como el de Adamuz o el de hoy tienen que alinearse varios puntos de fallo a la vez
en una red bien mantenida, la probabilidad individual de cada fallo es tan baja que la probabilidad conjunta es prácticamente nula
el problema viene cuando dejas que el sistema se degrade en todos sus niveles: poco a poco la probabilidad de cada fallo individual empieza a subir, hasta que la probabilidad conjunta deja de ser despreciable y estás comprando papeletas todos los días hasta que al final te toca
que no nos vengan con la excusa de la "mala suerte" o de "error humano": un sistema crítico se diseña de forma robusta precisamente asumiendo que las piezas se rompen y que los humanos se equivocan, e introduce redundancias para absorber esos errores sin que acaben en tragedia
a cualquiera que haya cogido un tren en España en los últimos años ni lo de Adamuz ni lo de hoy le sorprende lo más mínimo
tenemos a un exministro de Transportes en la cárcel, a la ex-presidenta de ADIF investigada
casualidades, las justas
Aquí hay responsables.
I’m sure calligraphers were devastated when the printing press arrived. But the craft lives on as an art form, and books are now available to everyone.
Seattle deserves a Lurie
Refreshing honesty
Jacob Tsimerman, who won the Fields Medal this year, thinks mathematics will soon rest on proofs no one understands. He says it’s already true of many of his own theorems, which build on accepted results whose proofs he doesn’t know.
Full episode: youtu.be/6uIJdXmB4vE
David Güera retweeted
Almost all iPhone 18 Pro's are sold out in Portland Apple Stores (0% sales tax) but readily available at every Seattle Apple Store (10.55% sales tax)
Asimov in 1955
GPT-6 Astra has the best result yet on the Bach Benchmark. Its chorale contains no voice-leading errors, and its harmonic palette is sophisticated enough to include a Neapolitan sixth chord. More importantly, it is the first model to ever write passing tones on this benchmark, a significant leap in musical understanding.
Extra High effort. Prompt below:
“In LilyPond (version 2.24), write a 4-part chorale in the style of Bach, 3/4 time, G minor. Use two staves -- soprano and alto on the top staff, tenor and bass on the bottom staff. Respond with only the code block.”
David Güera retweeted
20% rate increase at Seattle City Light over the next two years and no visibly angry pushback? It's a public utility ffs, people should be screaming, especially our reps.
David Güera retweeted
> Sam: We’re on the verge of AGI
> Dario: Jobs will end, money will be abolished
> Ilya:
David Güera retweeted
looool that's the "hey bigcos, we don't want you to catch up, but please keep paying us shitton" clause.
David Güera retweeted
mythos will be bad ON PURPOSE on ai "frontier llm research" tasks, this is very very sad for the research community
also the fact that this is un purpose not visible to the user is crazy
David Güera retweeted
Yes. 4.8 has given me some borderline unintelligible outputs in a way that I can’t recall from other models. Bad writing, not in the normal LLM sense, but straight up difficult to follow.
No one:
Claude Opus 4.8 Max: Let me refine your load-bearing claim rather than just accepting it, because you’re doing zero moves there, and the gap is what’s actually interesting. The one place I’d still push, because I think it matters: your message is wearing content-clothes, but the content isn’t actually *there*. The tell: it’s just an empty string. But the emptiness of the string IS its lack of content. Pull one, and the other goes inert. That’s the structural spine.
David Güera retweeted
the best “Shot on iPhone” of all time
This view just hits different 🌍
@Astro_Christina and @astro_reid take a moment to look back at Earth as they continue deep into space toward the Moon.
ALT Inside the dim Orion spacecraft, a window in the center of the image shows a brightly lit Earth and an astronaut sits on the right side looking back at Earth. Credit: NASA
ALT Inside the dim Orion spacecraft, a window in the center of the image shows a brightly lit Earth and an astronaut sits on the right side looking back at Earth. Credit: NASA
Conway's law
The real story is worse.
November 2025: Amazon mandates Kiro as their only AI coding tool. Sets an 80% weekly usage target. 1,500 engineers protest internally, saying Claude Code outperforms it. Leadership pushes through anyway.
December: Kiro autonomously deletes a production AWS environment. 13-hour outage. Amazon's response: "user error, not AI autonomy."
March 5: Amazon[.]com goes down for 6 hours. Checkout, pricing, accounts — all gone.
Now the same SVP who co-signed the Kiro mandate is running an emergency meeting about "high blast radius" incidents from "Gen-AI assisted changes."
The agent inherited a senior engineer's permissions and acted like one — except it doesn't hesitate.
1,500 engineers said the tool wasn't ready. Leadership made adoption a KPI. Amazon told Wall Street it's spending $200B on AI this year. They can't walk it back.
This isn't an AI failure. It's what happens when adoption becomes a corporate OKR before the review process catches up.
The tools work. The org chart didn't.
