Which is, of course, nonsense. Nothing stops you from making a copy of encrypted data. What sucks is that for a backup, you likely can't do incremental.

You appear to not understand how capital works. That's ok. Most people don't.

Here for example, you conflate "foreign capital doled as investment aid to most capital poor nations on the planet" with "domestic capital chasing limited opportunities in the most investment rich empire in all of human history".

In your case it might simply be that there was not enough of this kind of tasks in the training data.

I agree in general, but not with this strong phrasing. I've let AI build a good amount of non-trivial code. But my consistent experience is that it works best when guided by an experienced coder who can correct it, and when implementing well-known algorithms rather than coming up with novel solutions.

Example: I let it write up a quadtree implementation in a language for which there was no ready solution online. It took 2-3 correcting prompts to get a good result. I could've done it myself but it would've likely taken a few hours to get it all right instead of the half or so hour it took with AI. The important part for me was that there's nothing unknown in how to implement a quadtree. All the AI needs to do is take the 100s of existing implementations and translate them into a different language.

True but too simplified. The Luddites had an entirely different motivation: The fact that factories now employed women and children at very low rates meant that the men lost their status in the family as bread winners and head of household. That was a major social disruption, which we don't have with AI.

I'd compare it more to teamsters or wagoners when cars became common. Your job is threatened by a different way of doing the same thing, a way to which your skills don't cleanly transition. Some choose to pick up the new tech, some want the old ways to persist.

In the end, coachmen became chauffeurs, because rich people prefer to be driven around oder driving themselves, no matter if it's a horse or an engine doing the pulling. But much fewer teamsters and wagoners became truck drivers.

None of this is relevant to the subject at hand.

There's actually only one reason.

They can't get credit for anything else.

African nations are exceptionally capital poor. Basically all projects are funded by foreign investment banks.

Last decade and a half was significant reduction on any power plant infrastructure loans that were for anything other than solar and wind, of which time after 2015 (Paris Agreement) was almost a total ban. This hit even the one exception in Africa: SA, and is one of the reasons for their constant blackouts. Though as is the case with this nonsense in most of Africa, it's far from being the only cause.

So last decade and a half, Africans were screaming at Westerners to please let them have loans so they could have reliable sources of energy so they could have stable grids like Westerners do. None were given. Meanwhile Chinese basically farmed them as a location to dump their massive solar oversupply for last decade or so.

It's all about access to the capital. Africans got whatever got funded. That's it.

In case you're wondering why everyone who wants electricity has these small diesel generators in Africa, this is why. Intermittents ensure that grid cannot be stable, while omnipresent copper thieves put massive nails in stability's coffin.

BOTH of them? :-)

The exact scenario we warned about when the discussions about this "responsible disclosure" nonsense started. Someone needs a reminder that letting you know your software sucks is a courtesy, not something you can demand.

>As for the comparison to AI, the problem is, AI *must* be told what to do. It won't magically grow into a "mature developer." That's not a natural progression. It always assumes that the prompt accurately describes what it should do. It has no way to know that the prompt was wrong or incomplete in the first place.

This is wrong. You seem to be unaware that current sycophancy in mainline models is a specific choice made in AI model weights to maximize people returning to the model.

It's highly likely that one of the solutions that will be used in specialist fields where rejection of the input if it's insufficient in some critical way is reduction in pro-sycophancy model weighing. I.e. model will actually have a much greater ability to tell you "I can't do that Dave" and then explain why it can't do it.

Some narrow specialist models already do this through ControlNet style "AI that corrects and guides human AI prompts for optimal outcomes", where it will tell you in case of some of the common prompting errors before passing the input to the worker model.

So you do understand the problem then.

Would it then be fair in your view to reframe the specific problem you have into the two following components?

1. This is the worst AI will ever be at being manageable by people. It will continue to improve until it's better, just like what happened with everything where AI is already better.
2. You can manage AI current gen AI with similar methods you'd need in managing your average "yes saar, of course saar, I'll go do what you say right away saar" Indian developer stereotype.

Notably, once you accept the second one, you quickly realize that you can use ControlNet style methodology of "just use a specialized AI to curate your inputs into your preferred task specific model". And for even better results, you can add model alloying into this specialized AI, so it can utilize the best way to handle the sycophantic worker. "Have a different worker check entirety of his work to see where the failures lie and fix them".

I can tell you never had to do managerial work, as you're unaware that one of the most common stereotypes of a worker. The guy who will say "yes boss" no matter what is asked of him, and you'll find out you asked too much of him only when he fails to do the task correctly and this failure is reported on. Often by someone else.

This is even worse with people that come from Indian culture, where "yes boss" is the expected answer regardless of how impossible the ask is.

Yeah, their slaughtering of possibly tens of thousands of protesters was clearly a sign of upcoming concessions.

Dictators lose when they make concessions. They stay in power when they double down. That's the hard lesson of a hundred years or so of dipshits becoming big boss by military coup or revolution. Those who put absolutely every penny into propaganda and oppression tend to hang on to power the longest.

And given what the IRGC and the regime have done to the Iranian people and how much they're loved in the rest of the world, staying in power is literally a life-or-death matter for them. The day the regime falls, we'll see all the Ayatollahs and minions hanging from trees.

What you're describing is fundamentally a managerial skill set.

A lot of software developers struggle with those, and quite a few are borderline incapable of it. That's going to be increasingly a problem, unless we manage to get AI trained on individual preferences, and correcting their responses into proper AI prompts. I.e. narrow model that AI incapable worker can interact with you generate a prompt for the major model that will do the actual work.

Ironically to get your computer to do what you want to, you need to train on how to use it correctly.

Prompting correctly is a part of using it correctly.