Eh, in the american lexicon the terms have gotten pretty meaningless. To many, they are pretty interchangable and are just shorthand for 'hates god, his chosen economic system, his chosen country/culture, his chosen race, and the natural order that enforces that.'. The actual economic or political structures are not really part of the mental model and have not been since the 50s.

Yeah, people tend to forget that Trump was mostly just a TV star and influencer who used 'business' as his gimmick.

You have to remember that the big things the right didn't like about communism is (a) not allowing rich americans to invest and (b) not linlking church and state.

Once the rich get their cut and the religious can use the state to force their hierarchy on others, it magically isn't communism or evil anymore, even if everything else is exactly the same.

Is to never use AI for anything where individual results matter.

I have a feeling COBOL will outlive ChatGPT

Nah, that only applies when they are atheists. In the US, we call it god's chosen system exceptional people.

And this is one of the big problems with machine learning. The people who develop it can not answer such a question since they don't actually know how it works. They just throw a bunch of expensive GPU at the wall and hope the results 'look' right. This is why it is kinda a joke in terms of AI research.. ML is lucrative, but the backers are kinda like their users, their whole dream is to build things that make them money without having to know anything about the subject.

Given the AI industry right now... comparing fighting over data to guano actualliy sounds about right.

Not only that, but will probably be fed into their AI models.. as they run out of good relevant data, the demand for ANYTHING has been growing.

Hating the conveniences of others? That is an.. interesting.. take on things. So which straw man in that rooted in?

The big question will be how this rule change dovetails with liability.... which we can guess the answer to that... people who make money should not be bothered with the cost of their actions after all.

Which sounds terrible. Replacing a predictable formal language with a deterministic complier that you can validate the results with something that has a quasi conversational english interface and unpredictable results? I imagine after the hype wears off, we will see a wave of locked down versions with documented language specs, people realizing how horrifically complicated it is to use, and rediscovering formal languages. Kinda like every other 4th generation language that promised to make coding 'easy' for 'idea people' and finding that once you need to worry about details, it is a nightmare.

Gotta love 'prompt engineering'.. all they have done is take well defined formal languages, and replaced them with undocumented quasi-natural languages. I can not imagine it scaling well, and by the time you develop anything of even moderate complexity, you are probably dreaming of having something as reliable and precise as a formal language again.

On the other hand, any app AI is actually good at making likely represents a segment that is so oversatured that it isn't actually worth writing.

As soon as I started hearing LLM experts describe computational linguistics as a 'hack', it highlighted how nervous and insecure they were. They don't understand their own models, they don't understand their own domains, take that lack of understanding as a necessity, and just hope that with enough VC the computers will figure it out for them so they don't have to do any of that stoggy 'research'.