Considering they likely used credit, /,, and a bunch of other sites to train the model I think the cruelty has long since been baked in...

Yes, and it requires a multi-layerd approach to how they system raises issues. One simple prompt pass is not enough, however a multi-tiered approach can increase teh success rate. It'll never catch everything, and neither will a human, the tool is merely a way to remove much of the grunt work searching though documents so teh analyst can focus on what is important and make it easier ot find where things are referenced.

There are two different types of reports to develop - one that validates meeting requirements and one that is merely reporting results or periodic required reporting. The former is challenging, the later a lot less so, and reducing the time spent on the later frees up time for the former. Of course, many such reports do not need AI since they merely compile data, and that is easy to program in any of a number of tools. AI just adds hype and market cap...

Monitor, certainly. But why would they have remote control that can set a FREEZER to 50C?

The manager would.

I guess Yandex's DR sites don't have the latest updates installed.

Given that Russia can't refine crude or deliver fuel and now it's data centers are dropping like flies, they're the ones heading for the bronze age.

Sooner or later, the oligarchs will be showing Putin the window. He's making it hard to make a dishonest buck.

Exactly this. I have experimented with using AI, but absolutely forbid it to actually check in code. I just let it show me it's proposal. It can be useful at that, but often needs to be prodded about criteria it left out and frequently the final result needs a re-write for maintainability at least. It has a tendency to use really convoluted logic which can be greatly simplified without changing the result.

It does much better as a search tool. It's pretty good at finding exactly where some feature is implemented in a large codebase.

Of course, which is why for special purpose models this can work; especially since you can train it to be more accurate once you have enough real life data. This tool is not likely using the commercial ChatGpt but a bespoke setup.

While I agree that AI is not truly deterministic, that doesn't mean you will get widely divergent results from teh same data. There is some run to run variant, but in my experience, working with very specific criteria, it tends to be in the ranking order and not a failure to identify relationships. That can be addressed in a number of ways. While positive hits are easier, negative ones can be checked in a number of ways to increase the confidence level that they truly are negative, in the sense they are not relevant to the tested data set. To me, one of teh keys is to work with a well defined set of criteria, something I think is possible with tehy type of information they are dealing with, especially since much of the input data will be well structured and in a consistent format.

A human must always be the final arbitrator, and needs to realize the AI is not fool proof and you can't accept everything at face value.

I was think more along the lines of "Here is the periodic report on X" or "Here all the similar events and corrective actions..." to basically act like a research librarian; not to review techical documentation to identify design issues.

Nuke plants, st tFA points out, are massive generators of paper. A lot of times spent reviewing past documentation for relevance, trends, etc. That is actually an area where Ai can be useful, especially since you are working with a narrow and reasonable well defined data set and requirements. Tailoring the LLM to taht can save a lot of time for an analyst/engineer/operator by identifying and collecting teh information needed for analysis and reporting. It's not like they're giving it access to the reactor control system and telling it to pull rods or increase steam flow, per TFA.

You wrap them in plastic carefully first. The cameras are perfectly safe in the protective concrete packaging.

I remember listening in on a great phone call to a network provider about sending someone to pick up their expensive Cisco router. It ended with "in 30 days, it's going in the dumpster.

I personally 'uninstalled' another router by disconnecting the fiber cables...with side cutters. 2 months later, they actually tried to bill us for bandwidth we supposedly used from the completely disconnected router. We sent them a picture.

And that is the conundrum. If there were real efective data protections to ensure private data was secure from any cop wanting to do a search and searches required teh same effort as those in the real world, people might have fewer issues with it. Collecting massive data bases of people is inherently bad beacuse it will be missued, as events have already shown.

The US culture certainly grew out of violent struggles to expand its territory across the continent, and no doubt shaped its current one to a degree; and has it roots in European culture. Most violent in history? There are plenty of cultures that can easily compete for that title, from European ones to Asian to African and everywhere between. Europeans, for example, routinely massacred or enslaved defeated opponents in their history.

The moral of that story is don't keep fighting a war you can't win, and when it is obvious you are going to lose, surrender.

"Laser light at that wavelength has only recently become possible."

Lolwut? We've been able to run multiple frequency-doubling passes on various laser sources to reach most any range of UV laser for well over a decade. The problem is you get not a whole lot of radiant output - less than half a watt output at gobs of watts input.

Also, 148nm is not deep UV, that is flat-out Vacuum UV.