The real shocking news here is that he was actually arrested for it. (Not holding my breath for a conviction.)

Now you’ve got a dependent group who rely in AI as a source of truth, and now they can modify “the truth” whenever you want because it’s the only source left.

Well, besides using that for political reasons, they can also use that for massive profits.

Better make sure your carpet cleaning business is paying Google for “”“advertising”“” (no matter how much it costs), or Google’s AI will just forget to mention your business when local customers search online for carpet cleaners. Or you want to shop online for a new phone? Now Google’s AI will only feed you positive reviews of Google phone products, along with occasional horror stories about other products. Want to download and run Firefox? Google’s AI will straight-up gaslight you into thinking that all non-Chrome browsers are extinct.

Seriously, though. It’s a data center. It could be built pretty much anywhere.

When you’re building a data center, you know it’s going to need massive amounts of cooling. And for that cooling, it’s going to need massive amounts of water.

So why the ever-loving fuck would you build one in the middle of a very hot desert area? It makes cooling much more difficult, and water is difficult to get there. What the fuck is wrong in the heads of these people building data centers?

Then set up a cron job that periodically tries to read a file on the drive, and if that command fails, then execute sudo mount -a

Like, set up a root account cron job to run touch \external_drive\test_file.txt || mount -a at 5-minute intervals.

Fuel taxes pay for roads.

If only other types of taxes could be used to pay for roads… But alas, only gas taxes can be used for that. Entirely different monetary system, that. Roads need gas money and regular money just won’t do.

Or … we could just not tax electric vehicles, and call that a subsidy to encourage the more environmentally friendly option.

If, at some future point, electric vehicle adoption becomes so widespread that it becomes difficult to provide road maintenance because gas taxes aren’t being paid anymore, then you can find a different funding source for it. Maybe just fund it out of the ordinary general tax fund. Or even go really crazy and raise taxes on billionaires by two hundredths of a percent.

Even if proven and convicted, none of these rich fucks will ever see any consequences beyond a slap on the wrist fine – almost certainly a fine much smaller than the profits they made through this fraud. And then, just out of spite, they’ll appeal it in a higher court and get the fine reduced or tossed out entirely.

Only two industries left in America: gambling and fraud. And AI is definitely in the ‘fraud’ category.

but there’s a tendency of focusing more on the work during the semester than in the exam itself

Perhaps this tendency needs to be reversed?

If you have one big exam (or a few of them spread over the year) that it’s impossible to use LLM help for and those exams carry enough weight to make the student fail the class if they completely bomb it … then you’ll be stopping the LLM-cheaters dead in their tracks. Sure, they can be lazy and do much of their coursework that way, but if they’re being lazy like that, they’re likely not actually learning anything, and that will show up during the big exams. And when they fail those big exams and then fail the class, hopefully they’ll learn their lesson about relying on LLMs to get them through classes.

Interviewing every student would take a lot of time.

How is that allowed?

Hell, back in my day, teachers were even very picky about what kind of calculator you could use. And if it was a graphing calculator, you had to show them yourself wiping the memory at the beginning of the test.

(Except for one algerbra teacher, who was really cool about it. He’d allow custom programs to stay on the calculator if you programmed it yourself. On the theory that if you can write a computer program that reliably solves these math problems, then you must have a very good understanding of how to solve these math problems. And, yes, I was one of the few kids who actually did that. Ah, writing my own custom software for the TI-83 on the TI-83, because that seemed easier than actually doing the math problems by hand … good times.)

Before you can punish for using LLMs, you need to be able to reliably detect the use of LLMs, including guarding against false positives.

Current AI checkers are woefully inadequate and prone to errors.

The good students are still getting A grades naturally. And the bad students are getting A grades with ChatGPT. A grades for everybody! (Until we get to the closed-book, in-person test at the end of the year…)

Yep. A human might be a poor manager at first, but if they keep trying, they’ll get better at it.

This AI will never improve. It will always be just as bad at its job as it is now.

Oh, and now you need a new fucking degree to learn how to ‘optimize your token usage

In some companies, ‘optimize your token usage’ means using as many tokens as possible.

the same reasons web browsers store them in plain text

Why one web browser stores them in plain text. Fucking Edge.

Who knows about the others, but I can pretty much guarantee you that Librewolf, for example, isn’t doing that shit.

Still too much effort.

Automate a script to send AI agents on goose chases for you.

Name a single big tech company that died in the last 20 years.

None of them entirely die … they just become hollow shells of their former selves, sold off to another company to use the IP rights. See: Yahoo.

They’ve already had a spate of serious, unprecedented outages affecting half the internet.

So I guess that means the only place to go from there is to double down, eh?

To prevent automatic re-downloading, disable the experimental flags that trigger the download uninstall it and use Firefox instead. Or even fucking Edge. Literally any other browser would be better.