Indeed, and the only way I can see them do this by themselves, is that they need to have the ability to construct an internally consistent model about the world (or whatever it is we want them to do) from the data feed into them and then evaluate that model against new data and adjust it as required. But this is just so much more complex than what LLMs are doing. To solve any sort of real world task it needs this absurd layer upon layer of concepts about the world it is trying to interact with.

I doubt any of these companies are anywhere near this. We can probably find some shortcuts to allow this to happen for very specific tasks, but many human jobs require require, at some point, a broader understanding of the world, so that would render these things still just powerful tools.

Which is likely where all this will go. After the initial bubble crashes, out of the ashes, companies will spring up that can apply the useful bits of the tech to specific situations. Over the next few decades these will effectively automate a large range of task, but none of it will be anything close to AGI. It will all be painstakingly designed systems to solve specific problems - good money to be made but not this present craziness.

The 'surprising' thing is that stock spruikers like Altman have been going around telling us that these things might already be sentient; that they are going to be so intelligent and powerful that they might be too dangerous; that nobody really understand how they work; etc etc. Meanwhile the people who had studied these things since the 1980s were a bit bewildered about where all this was coming from. But hey, maybe OpenAI has another breakthrough after attention that was demonstrating a path to reasoning.

Now it is becoming apparent that, no, LLMs don't have some kind of magical reasoning intelligence that appeared when you scale them enough, and that Open AI doesn't have a secret breakthroughs on how to add reasoning to these things, other than putting a lot of lipstick on their outputs to make that appear to be the case.

There are some really smart people on slashdot, and many of them work from home on a keyboard

If they are not using AI to personally hold down three or four jobs then I'm going to assume that most businesses could not get that kind of productivity boost

If they just charged the advertised price, then we wouldn't need to this sort of regulation.For this reason, the problem isn't all the fees, but the fact that they are allowed to add on fees, outside of the advertised price. In addition to this, any price increase should be communicated in advance and before the minimum cancellation window.

is a face photo really suitable for age verification ? Can it reliably distinguish between an 18yo and a 19yo ? is it better than a photo of, say, the back of the hand

What are the odds that these photos were not encrypted or even cropped to be just a thin strip of image instead of the full face

It's reasonable when you consider they have to "support" those drives via their various support channels.

You put in a drive with incompatible firmware, then start asking for support because an issue with the firmware comes up, it directly costs them money.

Im not arguing the cost is valid, but if you've ever dealt with large commercial product support you would completely understand why its logical.

No you cant just refuse support to those people because
A. You will still contact them and waste resources to confirm an unsupported drive
B. Most states require vendors to honor warranty/support for modified products unless the vendor can PROVE the modification is the source of the problem.
C. Even after proof, some customers would continue to argue and add legal costs
D. Finally the customer will trash talk the vendor online and word of mouth, right or wrong ... costing the vendor even more money

Or they could just block your cheap drive and not have you as a customer and lose less money cause you're a tight wad.

You're not the customer they are interested in, you're a potential cost rather than profit.

Check out https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Funiversal-blue.org%2F and their various images depending on which suits your situation more.

I'd rather skip the Apple tax altogether and choose my own hardware and choose something that actually plays nicely with all other hardware - Fedora Atomic distros. I would never pay $1600-3000 for hardware that costs less than $800.

I'm not necessarily advocating *for* Microsoft either. It's what a lot of people are stuck with because a lot of businesses have custom built software that they don't feel like spending the money to try and upgrade. But I stand by my statement that Apple is not suited for business - at least not big, corporate business. Maybe it's fine in smaller businesses.

Combination of what cusco said, and, to a lesser degree, the second half of my other reply to you.

-Upfront and repair costs
-Apple security is nowhere near as good as the average user (or seemingly the average CIO) would think it is
-Even if the Macs are easier to manage in and of themselves (they're not), they're absolutely terrible to manage in a mixed environment
-Does not play nice, if at all, with most of the rest of the hardware and software that businesses need or use

If you want a "workstation" that is easy to manage on a network, doesn't require crazy super Windows specific software, and works on any hardware the answer is Linux. Preferably, a Fedora Atomic based image customized to the employees role. Employee can choose between GNOME (if they prefer Macs) and KDE (if the prefer Windows). Easy to roll out, easy to manage, and basically impossible for the user to fuck it up in any way, and even if they somehow do it can be solved in under 10 minutes.

I'm using Bazzite, built by Universal Blue which is based off of Fedora Atomic Desktop. And yeah, you can't really install kernel-space drivers. I mean, technically, you *can*, but you'd have to custom build a new OS image in order to do so, and then update and reboot into it. And if you accidentally fucked up super hard in some way, you can just boot right back into the rollback before you did anything.

I use a version of Linux that doesn't allow any third parties into the kernel at all, for any reason. It's completely immutable. Which is even better than anything MacOS does.

First of all, that is not a front page story from today, but from a week and a half ago.

Secondly, I don't care what businesses are actually doing. if their CIOs want to be complete retards by increasing the usage of MacOS, that's on them and their companies.