You mean this? https://account.cablelabs.com/...
Yes; that more-direct link works too. The link I suggested provides a little more context, perhaps, although to download the document it ends up using exactly the same link that you suggested.
About twenty years ago, I was privileged to be one of the authors of a security specification written at the behest of cable-based telecom companies that described the detailed design of a system for securing phone conversations that were carried over their networks. https://www.cablelabs.com/spec.... The design specifically started with the assumption that the network was penetrated, and was designed to ensure that the attacker could neither disrupt service nor learn anything useful about the traffic (for example, taken from the specification: "All media packets and all sensitive signaling communication across the network [are] safe from eavesdropping. Unauthorized message modification, insertion, deletion and replays anywhere in the network [are] easily detectable and [do] not affect proper network operation").
Once the specification was completed and it came time to deploy, all the telecom companies decided (whether in concert or individually, I do not know) that they were not going to deploy the design. When the lead security VP at one of the major telecom companies explained their decision to me: "We don't need gold-plated security like you've designed: we have firewalls"; I knew that the battle was lost. I also wondered how long it would be before the kind of intrusion like the one described in the article would occur.
Frankly, I'm amazed that it took this long; perhaps, though, what took the time was not the fact of a thorough intrusion, but, rather, the detecting of one.
When you call the store three miles from you using a local number, you won't get routed to Vidhya who's sitting in a call center somewhere in India.
Not true: I had exactly this happen to me this past week. FWIW, it was the local UPS store... and I got routed to India instead of the phone at the local store despite having called the local number.
Then not only did I have to navigate a phone tree that very nearly caused me to throw the phone across the room, but then (after hitting '0' so many times I lost count) got to speak to two lovely Indians, neither of whom -- as far as I could tell - had more than a very basic grasp of English. I say "as far as I could tell" because both the initial person and her supervisor had accents that were all but incomprehensible. In the end, I slammed the phone down, got into my car, and drove several miles to the store to talk to one of the people there in person (I should mention that they were very nice, sympathetic and apologised for the experience I'd been put through, even though, obviously, there was nothing they could have done about it).
Still, corporate UPS -- like so many companies these days -- are obviously unconcerned about the image they are projecting to the public.
The history of AI is all about modeling human intelligence, just like the models we have in natural sciences. If the model happens to be a very good match with reality, we may sometimes mistake one for the other. OTOH, they may be the same thing for all practical purposes.
I'm not sure if I have any deeper intelligence than a fancy language model. When we say that LLMs don't really understand things, then what exactly do we mean by understanding? In my personal definition, the meaning of something is simply the graph of its associated things. I consider something very meaningful if the graph has a lot of nodes and edges, and this also explains why simple things gain more meaning as we age.
Stupid question mayhap, but isn't whether this stuff is OR isn't copyright infringement still in the air, being battled out in the courts?
Well, if it's OK for a business to freely use copyrighted material for their commercial, for-profit purposes, then it throws out all arguments against non-commercial, non-profit "piracy". In other words, the court cases make a great test for the whole idea of copyright — they can't have their cake and eat it too.
Just wait until AIs start mining Bitcoin so they can buy stuff.
How would an AI mine Bitcoin? Two extremes come to mind:
(a) It uses a language model to compute SHA2-256 hashes "by hand", and starts demanding more data centers to make a decent buck.
(b) It figures out a vulnerability in SHA2-256 and takes over the network.
you were told not to (a) copy that floppy and (b) waste precious energy on cryptocurrency mining. But when big companies are building data centers for industrial-scale copyright infringement, it's suddenly OK. Because it's "busyness" done by white men in uncomfortable suits, not by idealistic young hobbyists.
This is what I did. They're basically indestructible and once I learned how they worked, I haven't needed anything else. They also seem to be easier to clean than the old nonstick pans. Oh, and they give you a little extra iron in your food.
That is energy per unit time and not equivalent to Ah, which is already cursed unit since it is meaningless without more information. Start rating things in joules.
The average Joe might not know who this Jules guy is, but he might know about calories, which is a well-defined multiple of joules.
It seems what he is doing is time average limiting his power consumption, not just blindly throttling the CPU. E.g. you may not want to throttle the CPU when it's just performing one quick task. That task itself will be shorter (and use the same energy) thus not saving you anything but slowing down the system needlessly. If on the other hand you're blasting your CPU for a while then it appears to throttle the CPU.
For years and years I've used the ignore_nice_load option in Linux cpufreq governors. If I know the task will be running longer and it doesn't matter if it takes a bit longer, I'll set its niceness level higher, so it won't trigger an increase in the CPU frequency.
I've also used my own scripts for turning applications on/off depending on electricity price for quite a while. I've only bought my electricity with the hourly market rate for about a year and a half, but the idea really started with my profitability scripts for cryptocurrency mining sometime in the early 2010s.
Besides the cost, another reason why I like to throttle my systems all the time is fan noise. I've always built my systems with low power consumption and noise in mind, starting with Mini-ITX boards in late 2003. I've always admired how laptops can be made with lower power parts, and wondered why they don't just use the same parts in "desktop" machines. Why should you consume power william nilliam just because it's plugged in?
It features the prettiest capital city in the EU and has the best goulash
That's like saying that Finland has the best mämmi in the world. While technically true, it doesn't make Finland a great country in any other sense. It's only when Finland wins the world championship in ice hockey or the Eurovision Song Contest, then we are a global superpower in all fields for a short while.
I've been to Hungary a couple of times and I have nothing against the people and culture in general, but I won't be touching the country with a bargepole while Orban is in charge.
When GPGPU in supercomputers was an up and coming thing 20 years ago, the big roadblock to acceptance by the big government funding agencies was the need to rewrite software. The scientific programs were written, debugged, and optimized over many years. Switching to CUDA or any other new thing requires rewriting the code, including debugging and optimizing.
Switching to OpenCL would give you a lot more hardware/vendor options than CUDA. OTOH, if this new company can put a lot of RISC-V cores on one die, why don't they just make a massively multicore CPU?
Tomorrow's computers some time next month. -- DEC
