And? Please scroll the start of this thread. Robotics is a long way off, we need better models and more compute. I am not sure what you are arguing for.
In my experience who live and work with smart people, and make smart decisions about their personal lives and families are pretty happy .
You can have a high IQ and make really stupid personal decisions.
Albert Einstein was notorious for this . He made catastrophic personal decisions that destroyed his families. His various wives and numerous mistresses ended up despising him and his son ended up in an asylum .
Well, as a user of LLMs I also find it annoying, the agents gather information on my behalf, they save me time and money. Are we splitting humanity? Those with info accessible to agents and those who only use their human sense directly?
This whole distinction is futile imho. And no, I'm not Sam Altman.
I mean training is one thing, you should honor people's licenses, but browsing and gathering information? Why force me to do it with my biological neural net?
My biggest fear with AI is that it basically devalues the production of knowledge. They’re a continued march away from actually showing people the original information, and the people doing that writing get less and less credit.
The companies themselves know they depend on this knowledge but have no plan to fund or replace the work done on sites like Wikipedia, stack overflow, or blogs like acoup.blog
Honestly, I'll take it a step further: what's wrong with training?
Having principles means applying them uniformly -- even to large entities or those you hate (it's fine if the principles themselves have size bounds in them, though -- versus them being implicitly glued on -- but then you need a universal justification for why that size. Which is possible and valid.)
I think that one should use the best algorithms, the best information, the best knowledge they have access to -- period. I don't like using gimped machines, I don't like making gimped machines, and I certainly don't like being sold them.
So I'm not going to turn around and say LLMs need to be gimped via arbitrary restrictions on their training data.
I agree. As soon as I understood the gist of how modern models do what they do, I found it unreasonable to apply any stricter standards to their "learning" than we do to humans. Humans learn by reading ideas and looking at images, then go on to have ideas and draw images based on that training.
We have definitions of plagiarism and copyright infringement that apply to people based on what they put out into the world, not how they trained themselves to get there[0]. Artists literally trace art and study specific examples in detail to learn, and that's not a bad thing. Humans can also accidentally plagiarize or make things identical to past works - it's easy to think a great guitar riff just came to you when it was really based on a song you heard years ago that stuck in a part of your mind but you don't even consciously remember the influence, for example.
It would certainly be nice for AI output to provide citations if it realizes it's using a significant chunk of an idea from its training that has a clear source (or many), though this is difficult in the same way it would be difficult for me to cite where I learned about the Towers of Hanoi. The AI frequently does web searches for specific resources to get ideas from these days, which are easy for it to cite.
However, it would have a chilling effect on progress as a whole if we all started jealously guarding our ideas so close to our chest that machines couldn't read them and only a select few humans who passed some kind of gate (or even paid us) were allowed to see them. We would be nowhere close to where we are if we had always had such a mindset.
---
[0] We use proven absence of viewing certain material as a legal shield against copyright infringement e.g. clean room engineering, but this isn't strictly necessary, and possibly even discouraged with modern precedent: https://reactos.org/forum/viewtopic.php?t=21740
Litigation and legislation against Big AI may be the only things that work .
We will probably have to wait Europe and China to do this since the US executive , legislative and legal systems have all been bought by Ames oligarchs .
Linux is so very predictably stable from a desktop user perspective . It is an absolute pleasure to use , particularly when combined with TUI applications like yazi, lsd, fzf , neovim etc
Been using it since the early nineties .
At work I use Windows Server 2025 as my desktop ( the alternative would be the dreadful AI infested Windows 11).
While Windows Server 2025 is not dreadful , the Linux experience is so much nicer .
"In 2012, Apple joined the PRISM data collection program run by the National Security Agency (NSA)."
"The firm added that between 1 December 2012 and 31 May 2013 it received between 4,000 and 5,000 requests from US law enforcement for customer data, involving between 9,000 and 10,000 accounts or devices. It did not say with how many it had complied."
just as a point of comparison, google had 25,879 requests involving 42,500 accounts in approximately the same time period (january 1 through june 2013)
google complied with ~65% of the requests
("fun" tidbit: the number for the last half year are 304,915 requests for 748,478 accounts, and google complied with.... 98%!)
The question is, how much data NSA accesses in a single request. The official rules have changed but the "two hop" rule permits the agency to recover metadata from about 25,000 phones with a single request, according to 2016 Stanford study.
Google's privacy ain't main/one of main selling point since we all know how things work in ad business.
If apple was principally against sharing those things they wouldnt be joining the very organization made for that.
Plus any public statement from anybody, unless forced by court, shouldnt be taken as facts. We have no way of knowing if it was 25k cases or 25 million cases, and they have no reason to tell such non-public truth if its harmful to them.
I didn’t find this in the article where did this come from? I remember when Apple suddenly shutdown the developer network to deal with a security breach in 2013.
Credit to Steve Job's, he wasn't too interested in joining PRISM. Reasoning was never given but I would suspect it was part his anti-establishment bent, part trying to not damage the brand if it came out.
It was only once he passed than Apple went all in on it.
Definitely, he may have been the architect of digital jails but he did seem to care about his prisoners. Tim Cook is just a free for all but with some nice marketing.
Whatever number they give publicly on how many US law enforcement requests they received you can be sure that number won't include anything they turned over the NSA which is probably everything in their cloud all of the time. We know they backdoored the devices (https://www.cultofmac.com/news/nsa-spyware-allegedly-gives-b...)
from NIST [0] to AT&T [1], the NSA has always had a backdoor into huge swathes of data created or transmitted through whatever means along with having Congress laws passed to retroactively make all of this collection legal [2]
it really makes you wish that the lesson the Ruby Ridge far-right took away was simple anti-authoritarianism and not the antisocial tendencies of racial separatism and a Judeo-Christian theocracy
I think the exploits of the Inquisitors across the millenia especially post-Cathars when torture and mass slaughter were recommended tools for religious indoctrination might be a bit a counter-factual to this (but religious history is not my most strongly studied area)
You understand that there is difference between US law enforcement and the intelligence community? And that all service providers are required, by law in almost ALL countries, to respond to legal requests from their governments? What you are citing is not any serious evidence of some shadowy conspiracy between apple and the "government".. Not that I believe Apple's bullshit about protecting users privacy, but you're grasping at straws there.
Prediction: Chris Lattner will be off chasing some other shiny thing within a year. Without a strong champion, the company will be subsumed into Qualcomm and stop publishing meaningful updates.
I have this pang too, but for a different reason: everything is moving to the edge, training and inference. Who needs a complicated heterogeneous setup like this when you’re running everything locally?
Chris Lattner is now Qualcomm's VP of AI Software and Tooling, so he is now responsible for all of Qualcomms AI software. Mojo wouldn't be doing any dying
Still slow compared to humans, but Chinese robots will be as successful as Chinese EVs, phones and solar panels.