Nothing like solving a critical problem in a core piece of your operations by hiring outside contractors. Anything to avoid investing in expertise, just like Sam Walton.

Wonder what those flaws would be like if they didn't farm out their operations to cut-rate people who don't give a shit? We all know banks can't afford permanent employees, they need to cut expenses to the bone.

"But there are a lot of lightbulbs in your neighborhood."

How many? Billions? Even so, a billion smartbubs couldn't do shit running an LLM.

"Because the LLMs are self assembling a mesh network..."

No they aren't. Just because these are words doesn't mean they make sense. LLMs are NOT networks and there is nothing about the internet that is a "mesh".

"Not every node needs agency, only the peripheral nodes. Like someone's pacemaker. Do what we want or your granny gets The Big One!"

What.a load of moronic shit, try don't even understand agency. For an LLM to threaten you, it doesn't need to spawn an LLM on a pacemaker, it only needs to allege that it can control the pacemaker. You are an idiot.

"They have realized [slashdot.org] that society is becoming ill at ease over AI. And they are creating a safety net for themselves."

No, they're not, they do not give a shit about you. The only thing that matters is money and power.

So what? An SD card, and the camera that uses it, lacks the bandwidth and processing to do anything with a model that it could store and run.

Fixed function embedded devices are NOT targets for running LLMs, many DON'T EVEN HAVE FLOATING POINT.

"What consumers want: Home Assistant, without hours of setup and scripting."

And without being forced to be a full time support staff.

"The only tech company that even remotely understands the concept of "durable goods" appears to be Tesla."

Appears to whom? Worst cars on the market, not even controversial.

"And even then it's a little shaky, but they've proven themselves by continuing to update software in 10+ year old cars OTA with no bullshit."

Tesla invented the car that got continuous updates, but also invented the car that needed them. Tesla is the worst thing to ever happen to the auto industry.

And Tesla updates have, in the past, been compromised by malware, that's a lot of bullshit. Tesla uses open source mile-high pile of shit software stacks to push their software. Will never be secure, Tesla doesn't care.

Because integrations are a feature of an open system, Apple doesn't provide an open system. Vendors have to pay Apple to support HomeKit.

And commit unlimited personal time, and prepare to be continuously disappointed.

I bought an HA device to evaluate against competitors and chose a device that came with HA preinstalled to make a fair comparison. It worked, took a couple of test devices and implemented a couple simple automations, just like the other devices. Then I let all devices sit powered on an unused. In a week, the HA device had completely failed and could not longer boot, owing to a bluetooth device that wasn't even used (needed for Matter) but failed init, restarted endlessly, could not be disabled and flooded the logs until the device itself could not function. That's Home Assistant: "Open Source" and piece of shit. Also, it was my 3rd HA test, I also ran it on Pi4 and native on a Mac, all poor experiences. Meanwhile, Hubitat is reliable and the cost is not much different than the hardware needed to run HA. Doesn't have the mass of nerds masturbating to it, though. Open source is a scourge, it spawns an endless supply of dimwitted tribalists whose sole characteristic is bad decision making. But HA is the largest open source project in the world, so it's got that.

"arguing that AI can be managed with conventional guardrails "

It can but they won't. AI wouldn't need guardrails if it wasn't enabled to do damage in the first place.

'...claimed the new system was "going to allow much more sophisticated behaviours that are at the frontier of what is being done in the AI field."'

But that's the problem, "what is being done in the AI field is the OPPOSITE of "conventional guardrails". LLMs do not exhibit "behaviors" at all, it is the agents that do that, so what Zuck is really saying is that they have no plan to manage with "conventional guardrails" at all, they plan to deploy models using agents enabled to perform obscene acts, models that have been trained on the work of sociopaths and that have absolutely no values of their own. ANY behavior IS the problem.

It is one thing to ask an LLM for an answer and get Nazi propaganda as a reply. It's another thing to mass propagandize the public by enabling AI to blanket social media with it. We know which of those is in Facebook's plan, it's not guardrails. Putin has plenty of money to pay Zuckerberg to spew toxic shit all over the public, frontier AI will do that better than ever!

"...you can not convince me it is more than that..."

Because that would you require learning something.

You can offer an analogy to search engines, but an LLM is not a "search engine algorithm", it doesn't even perform a search. And the chatbot is a separate app, not the AI itself. Finally, if you can prove "most the data fed into AI is copyrighted" you are in the wrong business.

It is your right to be stupid, you should reconsider how you put it on display. AI is a lot less than what the industry propagandizes, but it is a lot more than you "think".

That was true when I was young, but my belly button is a lot lower than it used to be...and hoods are higher.

My first car was a 62 bug, it was remarkably capable on unpaved areas ignoring it's skinny tires and it's ride quality was better than a Model Y I had recently (the worst car of my life). The immense size of current trucks is marketing combined with terrible law making. The upcoming Telo, if it delivers, will be an interesting data point on what a smaller truck can do.

That why you create a separate entity to commit mass murder. That way Claude can contract out some of the work. And that separate entity is just another python script not officially part of Claude. Don't even have to do it yourself, just ask Claude to do it.

Yes, all of this. And AI companies do know how LLMs "work". AI companies are very intentionally pursing agents that can do anything and LLMs that can learn what that is, then if you ask an LLM to do damage it will know how to. That is entirely intentional.

LOL you think the constitution matters? It only matters when Democrats have power.

We've needed them for decades, but Silicon Valley billionaires don't want protections against their exploitation. That will continue to be true despite AI accelerating the damage to society.

"These scenarios are not treated as inevitable..."

But creating agents that enable those possibilities IS treated as inevitable, because AI companies want the catastrophes to be enabled for their profit. The ability to do disastrous things is the plan, so the disaster is inevitable.