LET’S GO!!

Hoist by his own petard

Headline is a neuron activation lightning rod. Torvalds used the model to debug, the code is a one word fix he probably wrote himself.

This is the actual commit:

https://github.com/torvalds/linux/commit/818bebeb63dd6bf5f4e07e145f6cdbace520a34c

It fixes a goof where Intel Xe drivers allocate a few kb of memory that is used by the system during boot that causes the video to stop working. The fix is round_down() the memory allocation instead of round_up() the memory allocation. That’s pretty much the entire code change. round_up() -> round_down(). The commit message makes a lot of sense, and this bug is something you’d be spending a lot of time on manually debugging due to how insidiously subtle it is. Torvalds used the AI model to help debug the problem, but I’m pretty sure a one word change in code is not an AI generated code… and if changing one word is what the AI really did in the end, where do you draw the line at copyright? Challenging the copyright of individual words is impossible to enforce, because the word is used everywhere.

Of course.

There is a lot of government / intelligence agency interest in making sure devices can be hacked into… see the whole FBI vs. Apple thing from some time back.

Most device manufacturers are trying to keep costs down… adding security features adds R&D money they have to spend and pass on to the devices as sold.

Most people don’t know or care their devices are easily hackable… until it affects them personally.

Some people have to use GOS because enemy nations are targeting them and their devices in particular for hacking into, and normal devices are easy to break into… see Pegasus.

GOS is designed for maximum security at its base. It uses all the hardware features it requires to minimize its attack surface against adversaries that have infinite resources to attack the device and get at the infirmation inside.

Lowering those standards means opening a vulnerability, which means someone with that vulnerability can have their data stolen from the device in particularly that way, since that vulnerability is well documented.

amazing… I know of smaller models that specifically avoided that sort of thing because even negative training could go wrong so easily… they could have avoided that thing entirely, but nope…

I could understand if they were training a safety classifier (a model that learns what danger stuff is so it can recognize it when it sees it) but I doubt this is what they were doing here. That and handling such a radioactive dataset makes handing the demon core seem safe.

Ah, yes, someone’s made a business out of hunting for dmca violations… using an ai based watchdog… and because dmca is “shoot first ask questions later”, that makes disruption easy and remedy hard.

Saw that coming a decade away

COLON THREE

Your clutter has purpose. There is a difference.

A lot of AI images involving tech desks and workstations look “cluttered” like this, with no purpose or reason to the set props. Apologies to any artists that draw messy dystopian desks, but this style is super common with generated images in this theme

Knowing that something rolled up just behind your car can be a saver in many ways… not running over surprise debris, for one, not hitting the neighbor’s anti-cut boulder on the edge of the grass that you can’t see when looking behind you for another…

We don’t need no thought control

Unified memory hasn’t been the norm for a long time. The stranglehold is the bus connecting the memory to the graphics. SoC designs can bypass the strangle since they put the memory and gpu on the same die or literally next to it.

Having unified memory means removing unnecessary copying of data between memory segments, saving both memory usage, cpu and I/O time, and energy, especially as integrated systems (SoC) use the same literal memory chips for both things.

Basically, memory should not need to be dedicated to any task at all, it should be available to be allocated and referenced as required by the task that needs it, but since historically gpu memory and system memory are different memorys that require data to be loaded in either or the other, we are stuck with that paradigm in apps writen with that divide in mind.

A PCI Bus… last time I saw one of those, Chester was still alive…

needs more jpeg

Voice recognition? In a lift? In Scotland?