> It is far more economically efficient for the bad guys (whoever they are in your view) to bribe a few DC legislators than dozens of state politicians in places like Montpelier and Hartford.
State politicians are much cheaper, and no one from the New York Times pokes around when you buy off the state representative of East Bumfuck, Montana.
I believe it is possible to "upload" the knowledge from future reformulations of LLM's.
Right now most of the weights are in the FFN neural nets, but I believe we can reformulate LLM's or distill them into a format where each token/character represents either a matrix or a geometric algebra multivector. If each weight becomes token/character associated, it means we could write mindless reaction speed games: have a 2D projection of token/word/concept multivector/matrix coordinates. Let the 2D projection smoothly evolve a grand tour through the space of projections, possibly only including new dimensions if user performance on the explored dimensions suffices. 95% of the tokens/words could be rendered at their correct positions, while say 5% is at their incorrect positions. So the user sees a subset of the character/word/tokens rendered, and since they are individually placed at a correct position 95% of the time, it allows a user to learn peripherally, the user just needs to click those tokens intentionally disturbed from their true positions. Consider the hypothetical scenario that eventually all coordinates are explored and hypothetically the user scores highly (perhaps 100% or 90%) then we could say that at least the high dimensional coordinates are properly uploaded. Now you could randomly render bigrams (the new matrix coordinates being the matrix product / the new multivector coordinates being the geometric product). Next we could hypothesize that with sufficient training the user can predict the correct position of bigrams (meaning the user brain is performing the same or similar computation). From this the likelihood could be computed, and so a user would start predicting the same next tokens the model would. It wouldn't be blind belief more like a user associating the claims with what the reaction speed game insinuates. Imagine having trained on random cyrillic characters and suddenly you understand the Greek ingredients on some food packaging. There is a lot of hypothesis here, but none of the steps seem impossible to me.
It would mean both kids and adults could sync up with what LLM's claim (without blindly believing in it, but when it helps predict a lot of things around you people may lean and trust more on it, so that danger is not entirely moot), and having an LLM acquaintance in your brain to consult.
> But I also think it's clear that tool design impacts quality, safety, and efficiency
Yes, to an extent. And it's also the case that it usually doesn't matter. And that's my point.
I have also been someone like GP poster who has declared that it's physically impossible to produce valid software with XYZ tool in a team environment. And yet, there are oodles and oodles of counterexamples in the real world that proved me wrong and it worked AOK for them.
Could they all have been better off using another tool? Hypothetically yes. But their business needs (or whatever) were met and thus disproves a claim it can't be done.
Looking at it from Europe, this definitely also happens. It depends on the situation. I know of ppl who were kept bcs the parting was in good faith (which was less a firing and more an agreement that parting is in everyone's interest), but I also know of ppl who had their access revoked before firing bcs it wasn't. The latter had unilateral system access as well, which added to it. It's not about humane or inhumane, it's about risk. The 3-6 months being nice is also a fairytale that I have only ever heard in a positive light from employees who are not particularly ambitious or awake or in any way satisfied with their jobs or the prospect of a future job. On the other hand from the perspective of employers it's consistently hard to effectively restructure, it's expensice and awkward to have to pretend to want to keep someone around that you or they don't want around.
It's just one of these rules that unfortunately in Europe allow people to view life purely as the time between jobs. I'd never tell that to someone's face but it's simply a fact that the world stops of people don't work and no matter what the ideal world looks like in your dreams, working is the only real way forward for anything. It's part of the reason why Europe is falling behind on everything.
This isn't a response to anything I just said. I really don't understand why people collapse into all this handwaving when people point out the obvious: the money in our system is going to providers, and, in particular, it's going to practitioners.
I always found software IP to be absurd, but this is a particularly absurd situation. We're talking here about a small utility tool implemented from scratch and open sourced, with no apparent intent to make any money from it.
Are you concerned about the "encumberence" of using "unlicensed" tools to manipulate .doc, or .pdf, or .mp3 files?! Well I'm not, and if anyone ever tried to sue me for improper access to their proprietary formats, I'll show them some old testament impropriety.
What is even more insane is I was playing Battlefield 1 on Linux for years, until in 2023(?) they backported "EA Anticheat" to BF1, half a decade after the game stopped getting support.
This broke what was otherwise a perfectly normal Battlefield experience. Battlefield 4 requires Punkbuster, although it can run on Linux with no issues. You have to downgrade to an older version though, since EA hasn't updated BF4 to the latest PB AC, which causes you to get kicked.
I switched to Linux for everything but AAA FPS PVP games last year and have had a great experience so far.
Steam+Proton makes everything I play just work: Helldivers 2, Slay the Spire 2, No Rest For The Wicked, FF7 Remake, Stardew, modded Lethal Company (using r2modman) are the main things I've been playing recently, and all worked out of the box with Proton.
My PS5 controller may have needed me to install one package or something but has been working flawlessly after that.
I keep a Windows drive around for stuff like Apex Legends, Battlefield 6, but I pretty much never boot into Windows anymore except for those.
(I probably sound like a shill at this point, having commented something like this on multiple Linux threads now, but I continue to be impressed at how well Linux performs for gaming these days!)
Researches at top AI labs don't consider EY to be a kook even though they may not necessarily agree. EY concepts/terminology appear in Anthropic safety papers. Geoffrey Hinton takes him quite seriously and mentions him in his interviews.
> Today, many of those practices have been bought up by large corporations, including hospitals, private-equity firms and even health-insurance companies. It’s a shift that not only has changed how money moves through the health care system, but may also be helping some insurers boost their profits, according to new research published in Health Affairs.
> A study from researchers at Brown University’s Center for Advancing Health Policy through Research and the University of California Berkeley found that UnitedHealthcare, the nation’s largest health insurer, pays doctors who work for its own physician network, Optum, more than it pays independent practices for the same care.
I'll accept your first sentence for the sake of argument. You are still better off with a localist / federalist approach, because state governments are much less vulnerable to corruption and bribery. It is far more economically efficient for the bad guys (whoever they are in your view) to bribe a few DC legislators than dozens of state politicians in places like Montpelier and Hartford. Centralized, unaccountable power in DC means that when big rich corrupt companies bribe the right people, they can force the entire country to followed their preferred policies. A good example is how Purdue Pharma bribed the head of the FDA to approve OxyContin, leading directly to the opioid crisis.
Software that today is overwhelmingly prepackaged and usually professional, which I think at this point the nerds should reclaim:
* Podcast apps
* Music listening apps
* Feed readers
* Bluesky clients
* Note-taking apps
* Desktop bookmarking/read-later apps
* Chat and instant messaging
* Time trackers
* Recipe managers
These are all things that you can get better-than-replacement-grade results from Claude on --- not necessarily the best, not necessarily the most globally competitive, but certainly an application more closely tailored to exactly what you want it to do for your own idiosyncratic work style.
Music.app is a miserable experience, and I can just tell as I use it that it's miserable trying to serve me. But Apple long ago factored all the meaningful bits out of Music.app into MusicKit. Why am I still using Music.app? MusicKit is the real product now. This is new.
IntelliJ was unsupported ("community supported") when I joined in the summer of 2015. I built a new protobuf editor plugin during that period as a side project, mostly for myself, which suddenly became used by thousands of Googlers when IntelliJ became the supported IDE again in ~2016?
I eventually handed it over to JetBrains and I think it ships by default with IntelliJ now.
I guarantee you that the insurance company has zero clue or consideration for any physician and hospital resource constraints.
Gating access to medical care is the job of the patient's PCP and or other doctor. If the care is truly, meaningfully rationed (like transplant organs and blood banks), there are triaged priority lists managed by medical organizations.
You do realize all those hops are what Apple decided to set up? It’s not from the regulations. They could have allowed 3rd party stores with zero restrictions but are instead trying to fake compliance
State politicians are much cheaper, and no one from the New York Times pokes around when you buy off the state representative of East Bumfuck, Montana.