michaelscott

They're not available everywhere, and the neo-banks in general are tightening their processes to a point where they're becoming difficult to use. I've had multiple situations now where Revolut denied things like a tap payment at a restaurant in NYC because it was an "unknown recipient". Of course they don't notify you of it, leading to an awkward 5 mins with the waiter while I rummage through the app to figure out why I can't use my own money this time. Fast and convenient it ain't

This happens in a number of medical systems of countries across the world and is perfectly feasible, usually through premium subsidies for stuff like planned check ups, buying certain types of food and even installing equipment to monitor your driving behaviour. The thinking is that encouraging customers to be healthy reduces premium claims in the future

Unfortunately in this game first principles requires massive resources, not "some". Building in-house on top of existing open weights is a good way to bootstrap this process, especially since there's nothing inherently magical or particularly expertise-heavy when it comes to weights themselves

OP probably means using the existing open weights as a base for further homegrown development and research, not that the homegrown models are always updated based on whatever US or China are doing in the moment

It is not a given that we should allow computers to exist, the risk of harm is too great.

It is not a given that we should allow vehicles to exist, the risk of harm is too great.

It is not a given that we should allow hammers to exist, the risk of harm is too great.

The argument, even if it weren't moot due to the cat already being long out of the bag, is recursive all the way back to the discovery of fire. As a species we already regulate things that can cause harm in ways that are commensurate with the potential for that harm. Some are regulated more, some less, depending on the region. But all these things exist regardless; you have to decide whether you're comfortable with elites and governments being the only people who should have access to this, especially given that they have a history of not keeping your best interests in mind, or whether it should be democratized and available to all (like most other tools in existence)

It does with Volvo, although I couldn't say how big it is relative to global industry. Within Europe it's a large player

But why wouldn't he start the conversation in Polish? If the priest responds, he is Polish. If he doesn't, he is Spanish.

It's actually not available to most people that live in any third world country, otherwise migration would be significantly higher than it already is.

Regardless, "voting with your feet" is an individual action. Voting at home is a collective one, representing the will of not just you but the people from the place that you come from and were born into. Only one of those reflects the ideals of democracy, if that's really the ideal being strived for

Yes, they are. Accurate prediction is rewarded in every market, for every asset or asset class. There are adjacent order benefits (I live in my home, I eat corn, I have a say in how a company is run) but these are never divorced from the impact of prediction and prediction in aggregate is just a crowdsourced leading indicator of value; oil spikes the moment a missile hits Iran not because the 2 are explicitly linked, but because the market has predicted the effect on the flow of that oil over some subsequent timeframe and priced itself accordingly

This is functionally what already happens. The market to sell into for defense is almost entirely B2G, the only exceptions being places where the existing government is not very strong or centralized

Sure, but what _is_ the practical solution to the invasion of a foreign military power on your home soil then? Do you think these systems should only be developed by the government? And if so, do you then apply the same logic to anyone working in the government?

I had a partner who was a teacher in very rough areas; the school principal was routinely called by the local gang leader to let school out early because rival gangs planned to have a shootout in the afternoon. Kids there were abused in more ways than I want to remember or recount, babies were sometimes found in dumpsters, and the whole thing had this constantly oppressive and hopeless atmosphere.

My partner did her best to help the kids in her class, and part of this included reading them stories so they at least got a glimpse of the world outside of what in my opinion was hell on earth. The stories the kids always loved most were the Grimms, the violent ones. I think they allowed them to process and in some weird way make sense of what was happening in the real world around them, if such a thing is possible in that environment. I agree, I think the environment most kids grow up in today necessitates a "sanitizing" of story content in order to make it relevant.

For me, video is the main one. Sizes from 100MB - 3GB. Getting videos from an Apple device to an Android is a pain in the ass because I need to 2FA log in or click through something relatively convoluted (Dropbox, GDrive) or deal with pulling out some hardware I use once every 100 years (external drives). Localsend is a 2 or 3 click operation and very robust.

The kind of site that makes one happy the Internet exists

Their problem space may be just fine with open weight models regardless, but yes the release of gemma 4, GLM 5.1 and qwen 3.5 (and now 3.6!) have all happened in the last 6 months

Thanks these are genuinely helpful. I also use/d Pinterest for this kind of use case and haven't managed to find many good alternatives

I guess it depends on the definition of deterministic, but I think you're right and there's strong reason to expect this will happen as they develop. I think the next 5 - 10 years will be interesting!

Nothing you've said about reasoning here is exclusive to LLMs. Human reasoning is also never guaranteed to be deterministic, excluding most correct solutions. As OP says, they may not be reasoning under the hood but if the effect is the same as a tool, does it matter?

I'm not sure if I'm up to date on the latest diffusion work, but I'm genuinely curious how you see them potentially making LLMs more deterministic? These models usually work by sampling too, and it seems like the transformer architecture is better suited to longer context problems than diffusion

Switzerland's draw is the money. It's true that a significant proportion of the population is foreign born, but the whole country is smaller than some tier 2 cities in China and many foreigners do not stay longterm. If China paid Swiss-level salaries there would be more people going for sure, but the country is so big that at a relative level I'm not sure if the proportion would change significantly

This was crafted with a subtlety that captures the continental combination of infrastructural petrification and untethered pride perfectly. Wonderful

Yes because these descriptions are meant to foster dehumanization and detachment, which is very useful in military and scientific study contexts. That's why they also sound unnatural in casual conversation

A lot of mechanisation, especially in the modern world, is not deterministic and is not always 100% right; it's a fundamental "physics at scale" issue, not something new to LLMs. I think what happened when they first appeared was that people immediately clung to a superintelligence-type AI idea of what LLMs were supposed to do, then realised that's not what they are, then kept going and swung all the way over to "these things aren't good at anything really" or "if they only fix this ONE issue I have with them, they'll actually be useful"

After using Rust for many years now, I feel that a mutable global variable is the perfect example of a "you were so busy figuring out whether you could, you never stopped to consider whether you should".

Moving back to a language that does this kind of thing all the time now, it seems like insanity to me wrt safety in execution

* old Toblerone Matterhorn logo unfortunately :( They've had to remove the mountain from branding since the chocolate is no longer produced in Switzerland. Still, I love finding the bear in the older boxes still floating around.

No, more like literal survival. In these conditions there is no mental bandwidth for things like spiritual constipation

You're conflating 2 issues here: judgement of adult attempts at a new language and the time required to learn it. The first is just a cultural thing, although it is sometimes valid for understanding a speaker (cases in Slavic languages, pronunciation in a homonym-heavy language like French, tones in Asian languages). Problem is that it's oftentimes more "cultural" than "valid" critique, which helps no one.

The second problem is more practical and it's not the only difference between child and adult speakers; the vocabulary required in most day-to-day settings for a child is considerably easier to master than the adult equivalent, regardless of language (describing symptoms to your doctor or getting through a bank or tax appointment will be much more difficult than describing the weather or what you want for lunch). Adults in general are just as good as children at learning new languages, it's just that life has different requirements from that age group.

Edit: that said, I actually am agreeing with your general sentiment

I am not young, but I have never seen a major institution (including governments) caring about citizens in aggregate in my lifetime. To me, this is an artifact of the 50s or 60s, some bygone era (which is funny, because the government did not care about citizens in aggregate back then either).

I can only imagine how the younger kids see things. They're bombarded by public knowledge of nasty things institutions did in the bigoted/ignorant past, underhanded things they're definitely doing now, an anger/fear inducing news cycle and endless social media conspiracy theories (some of which end up being true) engineered for clicks. Extreme cynicism is a logical conclusion.

This is how Cloudflare does (did?) PRNG

I think this is a primary reason why there is no real "cheatsheet" for this stuff. The application of a given algo (and even what types of inputs you provide) are heavily dependent on the detailed specifics of your use case and how you apply them

Yeah correct, the French relationship with modern English is much closer because of (among other reasons) the Norman conquest that happened long after the Indo-European split and much closer to our time