arrowleaf

Not even a year ago people were dragging Anthropic for not committing enough to compute buildout. Important to note that most of these contracts are a decade-long, by their reporting they're able to cover this easily with their current revenue.

Tangentially related, has anyone done comprehensive labs on their own (via Function or others), or even genetic screening like Galleri, and had that prompt you to get something looked at? I've never been impressed by my annual physicals and how few biomarkers they check, but don't want to go overkill and find out I have a million things I could treat.

Curious, why do you say the model is mediocre? I haven't tried it, so I can't pass any judgement... I've learned to distrust benchmark rankings. Are benchmarks and Artificial Analysis the yardstick you're using?

I'm reading Sagas of the Icelanders as prep for a trip there, but I can feel it pulling me down into a rabbithole of older Scandinavian medieval writing. I've already got an order placed for a saga that's not in this book, along with the Poetic Edda and Havamal to learn more about Norse mythology.

And if margins are decreased, then... (what happens to new supply?)

That's not the point. They're never going to be the ones stopping and letting others race ahead. The only reason to strum up discussion around how dangerous AI is, is so that they can introduce legislation framing AI advancements as scary and that they are the only ones who can do it responsibly. It's all a regulatory capture play.

Package the product with a low denomination paper currency, lookup based on the banknote's serial number.

No, there are many mission-driven companies from enterprises to startups, but in the USA most software engineers work for Fortune 500 (or Russell 2000, or aspiring) whose underlying mission is extracting profit from revenue.

"- it would cause someone to lose a job they didn’t know was under threat."

Most of my life's work has been in areas of increasing productivity. Is that good or bad? On the good side of the spectrum, it reduces the time hundred of people need to spend on mundane tasks, or reducing the amount of mentally taxing complexity. On the bad side, based on this definition, there is less of a need to hire someone with specialized talent when the software can handle what their job function would be.

I've always felt kind of bad helping companies automate people out of jobs, that's the core of what most of us writing software are doing. Maybe I am a bad person.

Why the 'but'? The article is about becoming intentional in your thoughts and not letting them be hijacked by algorithms.

Why would AliceGG need to reference the paper? Anyone who is vaguely aware of the LLM architecture knows about the paper title and it being a huge advancement in machine learning. This is an article about recommendation algorithms hijacking our attention and LLMs ruining the internet... it's pretty clear the article title is a nod to the paper?

Really? While reading I couldn't help but think about how I wish this article on writing well were... written well. That block quote loses me in the first sentence, I had to re-read it a few times.

"Good teaching requires alternating between two views, going up and down levels of detail. Sometimes you zoom out and give students one word for a big idea, calling a whole chain of events a "collapse." This 'Nomiinalization' lets them hold the idea in mind instead of the whole story. But summary words alone can lose students, the abstract loses meaning as details are forgotten. So you zoom back in with examples, stories, pictures, concrete cases. Those reinforce the idea of the summary word."

Same here, first thing I need to know when considering a distributed system is how consistency is handled. If it's eventual consistency, what is the replication lag like? If it's strong consistency, can their network handle that? What happens when a node goes down?

The first thing I want to know when looking at a distributed system is how consistency is handled, ctrl+f for "consis" on both the blog post and neki.dev has zero hits. So I can only assume it will have eventual consistency with a considerable replica lag for real-time systems.