fredoliveira

Well, two things: it's the last sentence of the film; being on hour into something you're calling propaganda is brave.

Anyways. I thought the documentary was inspiring. Deepmind are the only lab that has historically prioritized science over consumer-facing product (that's changing now, however). I think their work with AlphaFold is commendable.

Did you watch the documentary? Would probably fare better if you did, because it'd give you the context for the film title.

One of the few people that has a voice (written and otherwise) so distinctive that even reading those lists, I read them in his voice. I miss that guy.

True - but I think that might make the adaptation to a movie easier, not harder.

Like GP said, I think the trick to this book is in the relationship between the 2 main characters, so hopefully they nail that. Judging by the trailer they made it all quite humorous.

Didn't it literally take someone's life this week in Washington?

> Anthropic have consistently shown they don’t know shit about anything but training LLMs

On what grounds?

What makes you think they lack engineering acumen?

Something happened today as I was closing my car door, where I had the Airpods (Pro 3 as well) make this low but extremely loud rumble-like sound. Something related to pressure and seal, perhaps?

I could replicate it immediately at will. ANC on, nothing playing, sat on the driver seat and closed my door with a little bit of force.

I guess I'll ask: what's wrong with fiction?

What is helix? Haven't heard of it before, I don't think.

Too far away to scroll: https://www.whereisroadster.com/

Let's be honest: he had no other credentials when he decided to run.

Fair! And yep, I too wish Ruby was stricter. I just happen to value it slightly less than some of the things I do get with the language.

> Once the model is out, no one would pay for it, right?

Well who does the inference at the scale we're talking about here? That's (a key part of) the moat.

I think that the performance unlock from ramping up RL (RLVR specifically) is not fully priced into the current generation yet. Could be wrong, and people closer to the metal will know better, but people I talk to still feel optimistic about the next couple of years.

A couple of counter-arguments:

Labs can just step up the way they track signs of prompts meant for model distillation. Distillation requires a fairly large number of prompt/response tuples, and I am quite certain that all of the main labs have the capability to detect and impede that type of use if they put their backs into it.

Distillation doesn't make the compute moat irrelevant. You can get good results from distillation, but (intuitively, maybe I'm wrong here because I haven't done evals on this myself) you can't beat the upstream model in performance. That means that most (albeit obviously not all) customers will simply gravitate toward the better performing model if the cost/token ratio is aligned for them.

Are there always going to be smaller labs? Sure, yes. Is the compute mote real, and does it matter? Absolutely.

> I bet all of them wish they weren't using Ruby.

I'd take that bet. At scale, (and those 3 are the definition of scale) you can mitigate some of the downsides of Ruby (i.e. speed), but you can't recreate the upsides (i.e. developer satisfaction, learning curve, flexibility) elsewhere.

> Go and Typescript seem to have taken its place, which makes sense because they're both much better languages.

Again: depends on the metrics you're considering. I would certainly consider Go much better than ruby on some metrics, but most definitely not all - and importantly, if I put all of it on a scale (and this is where bias comes in), I still give the edge to Ruby over both of those.

> he could have stopped at any time.

Obviously, clearly untrue. You go ahead and try stopping a behavior that reinforces your beliefs, especially when you're in an altered mental state.

Intel macs sometimes went crazy with fans. The leap from those to Apple Silicon hardware was (to me at least) as impactful as HDD to SSD.

Groq (the inference service) != Grok (xAI's model)

It does! Thanks for linking this - a quick browse revealed a few things my kid will be into.

Very thoughtful of you to have written a comment that makes me laugh as a response to a website that made me cry.

In what ways?

You can't possibly think that using an em dash is exclusive to AI-generated output.

My interpretation was that Yudkowski simply doubled Christiano's guess of 8% (as one might say in conversation "oh it's at least double that", but using the actual number)

Regardless of how you feel about SBF and FTX, claiming an early investment into Anthropic is "luck" rather than being ahead of the curve feels off the mark.

that looks fantastic - well done!

I have used Bookplayer in the past and it is great - but I don't think it has native ABS support, correct? You just download the files from bookplayer whenever you want to play a new book?

You can. The author made the source code of the app itself available: https://micahflee.com/heres-the-source-code-for-the-unoffici...

Not that exactly, but I just saw this on twitter a few minutes ago from Stripe: https://x.com/jeff_weinstein/status/1918029261430255626