> you'd get a robust C-FFI style implementation of those data structures which has been tested a lot more than some random library that has almost no existing users or active maintenance.
What are these mystical random libraries you're talking about? There is a solid C implementation of every data structure on this planet.
This comment made me think of pencil sharpening speedruns in the context of "preservation". Had a good chuckle. Quite a different mindset: the point is to sharpen a pencil (or 10) as fast as possible.
Call me a cynic, but I don't believe this is a genuine change of heart at all. It feels much more like a panicked response to something that might undermine their IPO.
Even if you trust Anthropic today (which I don't), they clearly don't want competition and there's no telling what other shady moves they'll pull in future.
The only sustainable way forward is to support open models. I was already on the fence about whether or not to keep my Max subscription (the extra cost over something like DeepSeek V4 didn't really feel justifiable). This is the tipping point for me, I'll be cancelling my sub before it renews at the end of the month.
I said I wondered if the models were going to start poisoning distillation and I got downvoted to hell. It’s interesting to me that they are now downgrading ML research too in this model, I would argue this implies the terrifying and impossible to reason about self improving AI doom loop is coming sooner rather than later. Bit worrying.
To be fair, nerfing Claude on frontier research tasks is consistent with Anthropic's stated beliefs. So in that sense you can trust them to always behave consistently if strangely. But this launch was done very poorly with the lack of transparency on when the frontier research policy was violated.
I have had multiple managers and the thing is I wouldn't expect a truly honest answer from my manager ever, so to me it is also just all performative ritual that I have to go through and would rather avoid. So it does feel like waste of time to me. And I also feel like I have to be careful, because honesty has most of the time just yielded me useless work in the long term. It is easier to pretend everything is good, than to point out issues as it will i variably lead to a rabbit hole of nonsensical work.
If the city used eminent domain to snatch it away from the farmer this wouldnt even be news.
If the farmer deeded it to someone else, like a neighbor, and said they could only use it to build a park, no one would expect that to apply once the government had yoinked it by eminent domain.
It follows then, that government entities really aren't bound by deeded restrictions. If you made them jump through some hoop, where they had to gift the land and then eminent domain it back, that would probably be more wasteful than just letting them do whatever they think is best.
The oddest outcome I can imagine is the government being able to compulsorily acquire other peoples property, but being permanently stuck with a fixed use asset that they cant do anything with that they actually own. Thats bonkers.
MacBook Neo for $600 is the only laptop under $999 worth looking at nowadays. Apple used to be the expensive premium option and they’re basically the cheapest today. What a time to be alive.
Completely agree. This is what Hofstadter means by a strange loop. Our current LLMs have no attentional autonomy by design. The recursion is superficial and without its own Now. Adding attentional autonomy is The frightening alignment issue.
I don't want to be cynical, but I assume a third party we can trust has verified this model is actually this good?
I would think it would not be Anthropic, out of all the players, that is selling a lie hidden behind "I am sorry, I can't do that; it's too dangerous."
None of those models are Fable's class. There are definitely many cheaper options for LLM agents, but people are paying attention to Fable for a reason.
Billions use HTML+CSS+JavaScript. Who really uses WASM? There are of course users, but very, very few in absolute numbers. Many projects are not web-based really. For Autodesk Fusion, as one example for many, I have some mega-slow application that takes forever to work with in some cases on my laptop (it is not the fastest laptop, but I recently tested this on a faster desktop computer with 32GB RAM and it is still slow to no ends; using it all WASM based would be even slower I bet. That's not winning anyone over ...).
Apple could bridge the gap between Mac Mini Pro and HomePod Mini for the Neo to try experimental local AI configs in need of more memory and storage. The Vision Neo completes the three part environment system for students in biology or morphogenesis and robot design.
A couple of months ago, I experimented with this - took tsgo and ported tsc to go with Claude. The main issue why this still didn't happen yet is because tsgo doesn't expose plugin API externally, but it's still there, so you can just co-locate your plugin as extra Go module and compile everything together. Managed to get my fairly large Angular app to compile and even run unit tests. Cold compile time went down about 2x - so the benefits are there, but not as dramatic. I think this would still need architecture level optimizations that enable build parallelization, but that also requires making some changes to framework API so components can be isolated-compiled or something.
