Into the balls! And spair the girls. They are probably good at sewing and knitting.
(I would not hope for cooking, though)

You do not put business rules into the browser/app.
You put them into the backend.

The reliability is on the App/Web site developer.
How did the AI do it?
Simple.

Normally it would use the HTML interface to click just like a human.

But the drop down only offers (let's say) 45 days advanced booking.

So it looks at the HTTP endpoints and how they are called.

And simply sents the HTTP request directly.

So: without authentication to "be person X", the AI could cancel person X's appointment.

Has nothing to do with liability of the user or the AI, or manufacturer of the AI.

It is a standard problem of web site / app designers to "allow" forged requests, because they trust their own HTML/JS in their web site's backend.

There are plenty. They are called Mac Mini.
A hated fruit company sells them.
Actually: pretty cheap.

It as well runs in ordinary RAM.
GPUs and TPUs are more important for training, less important for using an ANN/LLM.

NVIDIA is not an AI company. It is a hardware company. They need the money to expand production.

Money is made at the moment by companies that offer access to LLMs like Google's Gemini.

And to a lesser extend money is made by companies that use those LLMs for their own business.

And similar, money is made by companies that do their own machine learning - but need to rent GPU time in a data enter.

Not everyone has the know how and wish to run a GPU farm in his basement - or the power connection and Internet connection to even try.

Because the camera comes by default with it.
I have a few security cameras for $3 each, size of 2EUR coin, just a bit thicker, which open their own WLAN.
You connect with an app via Wifi with them, and configure them and tell them into which WLAN they should connect.
The camera they use is probably similar. Is connected with the ship computer via Ethernet and the boat itself connects via starling or similar to the home base.

With "standard radio technology" you do not have the bandwidth to reasonable transmit videos.

Do you know if Tesla or Pascal came first?
Yes I do.
But perhaps only because I studied all three: Math, Physics and Computer Science :P

You have some funny examples, lol.

Well,
my point is simple: I doubt this is happening in the relation between birds and windmills.
And: you can not proof it does.

So we are even.

Natural selection means: there is a filter that prevents/reduces "unfit" members of the population to breed.

There is nothing particular unfit in a bird behaviour to crash into a windmill. And there is nothing particular more fit, to not crash into one. There is no "magic gene" that would favour one and kill the other one.

As I pointed out in previous posts, perhaps not to you: the amount of birds dying to OFFSHORE wind mills is basically zero. I did not check onshore wind mills, though.

There is most likely a "data center bubble".
But not an AI bubble.

As soon as the data center builders realize they can buy optical processors, this "hype" is gone.

Yes, I know about those studies.
And as I gave you some rough numbers: it does not make any sense.

You neither have enough birds nor enough cats for the ca(t)rnage for such numbers

You would need some (some squared or cubed) monster cats that kill a thousand birds a year to compensate for all the cats who are to lazy to even try.

I did not feel attacked.

Well, we do not have MD here.
And if I write "BigDecimal", someone comes up and asks: what are you quoting here.
As plenty of people are not aware about the various usages of "quotes".

They do especially good in the cold down to -30C.

Well,
I never really got the naming schemes of Apple OSes as well.
Tiger, Panter, Cogar - ok.
The Mogave as desert, okay.

In general I do not know what my machine is running, I have to look into the system, and I prefer the version number.

Regarding LLMs, I would use philosophers or science persons and would pick a figure which personalizes the strength of the LLM. And then add the quarter of the year when it is released. Like Pascal 1/2026 for a math LLM or Einstein 2/2026 for a physics one and a Renaissance 3/2026 for and al purpose one.