Not sure if it's 1:1 with this new matter thing (I think it's their third push in this direction, at least), I can only recognize the door/window open sensor (even that is SLIGHTLY different), anyway I have a bunch of door sensors, leak/water sensors, and movement sensors. They all work over zigbee, went into Home Assistant with no fuss, I think they can do even firmware OTA updates. Best design ever for each specific use, batteries last forever (and work, even better, and report well the capacity for NiMh), super-happy. In some threads I follow people reported repeatedly that they had problems with the batteries draining quickly, but mine will go over a year for sure (I recharge them just in case once before the winter together with the ones for some thermostats and such) but it might be that there are some quality control issues, OR some environmental (radio) factors that might play a role here.

With a president who solves an existential threat by finding the best expert he can find, and using his own formidable political skills and charisma to run interference for that expert.

Seriously, Comacho was a meathead, but Iâ(TM)d vote for him.

No comment.

Of course we know, because we don't see the source tree to the next GTA game posted anywhere, do we?

No. Like any software, AI requires maintenance, and that maintenance costs money, lots of money.

It does not. Models need nothing more than the storage of some gigs of weights, and a GPU capable of running them.

If you mean "the information goes stale", one, that doesn't happen at all with RAG. And two, updating information with a finetune or even LORA is not a resource-intense task. It's making new foundations that is immensely resource intensive.

Can you integrate it into your products and work flow?

Yes, with precisely the difficulty level of any other API.

Can you train it on your own data?

With much less difficulty than trying to do that with a closed model.

Has it been determined whether the IT situation was related to the theft that occurred?

Obviously it sounds like basically no bad option was left unchosen when it came to their IT config; but I'm curious whether this was a situation where the perps were actually sophisticated enough (or unsophisticated at traditional smash-and-grab/balaclava-when-on-camera techniques) to incorporate the bad IT into the heist; or whether the entry was more or less pure physical access control failure that happens to put the general state of the system in stark relief?

Obviously if it were a heist movie there'd be a hoodie kid using the power of fast typing to haxx0r the cameras and guide the operatives while using a precociously cobbled-together AI to selectively delete them from the surveillance footage; but if the overall physical security posture was bad, and the building is largely accessible to the public, it seems entirely plausible that someone just cased the joint and walked in much as they would have 50 years ago; though a different interested party is probably hosting a C2 server or some exploitation payloads on their DVR.

This is the part in the bad summer blockbuster where it turns out that there was some sort of alien organism involved in the collision; and now people are quietly scrambling to ensure it doesn't make it back to earth, right?

Steam offers developers the option to sell their own Steam keys. In theory, that lets developers use all of Steam's infrastructure services without paying - Steam doesn't take a cut of those sales. Essentially the only condition is that they don't offer those keys on better terms than for Steam buyers.

That's basically the exact opposite of abusing your market power. To the degree you can do better than Steam, with help of a third party or not, they let you keep Steam's share accordingly.

The small game developers I've talked with have all said that Steam is by no means a replacement for a marketing budget. Random indies on steam don't get noticed, as a rule.

Remember when Windows 95 wasn't installing the IP stack by default? Better yet you had to buy it as Trumpet Winsock for Windows 3.x ?

Not all Nvidia exports to China are banned by America.

Even older Nvidia chips are superior to what China can produce domestically.

And my point is that AI wouldn't just stop being used even if the bubble imploded so heavily that all of the major AI providers of today went under. It's just too easy to run today. The average person who wants something free would on average use a worse-quality model, but they're not going to just stop using models. And inference costs for higher-end models would crash if the big AI companies were no longer monopolozing the giant datacentres (which will not simply vanish just because their owners lose their shirts; power is only about a third the cost of a datacentre, and it gets even cheaper if you idle datacentres during their local electricity peak-demand times).

Your scenario is impossible, so try again.

Because we're discussing a scenario where the big AI companies have gone out of business, remember? And the question is whether people just stop using the thing that they found useful, or whether they merely switch to whatever alternative still works.

It's like saying that if Amazon went out of business, people would just stop buying things online because "going to a different website is too hard". It's nonsensical.

They believed you could mimic intelligence with clockwork, etc. Why do you only count if it if it involves computers?

If you want to jump to the era of *modern* literature, the generally first accepted robot in (non-obscure) modern literature is Tik-Tok from the Oz books, first introduced in 1907. As you might guess from the name, his intelligence was powered by clockwork; he was described as no more able to feel emotions than a sewing machine, and was invented and built by Smith and Tinker (an inventor and an artist). Why not electronic intelligence? Because the concept of a programmable electronic computer didn't exist then. Even ENIAC wasn't built until 1945. The best computers in the world in 1907 worked by... wait for it... clockwork. The most advanced "computer" in the world at the time was the Dalton Adding Machine (1902), the first adding machine to have a 10-digit keyboard. At best some adding machines had electric motors to drive the clockwork, but most didn't even have that; they had to be wound. This is the interior of the most advanced computer in the world in the era Tik-Tok was introduced. While in the Greco-Roman era, it might be something like this (technology of the era that, to a distant land that heard of it, probably sounded so advanced that it fueled the later rumours that Greco-Romans were building clockwork humans capable of advanced actions, even tracking and hunting down spies).