"AI" is a muddied term
I've seen a lot of news lately about how this-or-that thing is using "artificial intelligence" now, and I'm frustrated with not knowing what kind of artificial intelligence is being used.
Large Language Models (LLMs) are the technology behind Generative AI - things like Chat GPT, Claude Code, Nazi Grok, etc.
Machine Learning is the technology behind Large Language Models. We've had Machine Learning technology for decades. ML is definitely a form of "artificial intelligence", and I suspect many of the reports these days about "artificial intelligence" are machine-learning but are not LLMs.
One example is: Some hospitals are using Palantir software to monitor patient vitals to detect sepsis before hospital staff would detect sepsis. This has saved over 900 lives, the report said. The report only referred to it as "artificial intelligence".
So is it a Large Language Model? Is it Generative AI? I doubt it.
It is probably a smaller & more specifically-trained model. It was probably trained on loads of patient-vitals data. Its whole input is probably patient-vitals (and maybe some demographics) and its entire output is probably a percentage-prediction that sepsis is present.
But I don't know. The report - which came from Techlinked, a technology-focused YouTube news outlet - didn't specify. It just said "Artifical Intelligence".
Just that term - AI - is too large of an umbrella. And I really want reporting to start specifying exactly what type of AI is being used. I also want things like presidential (U.S.) Executive Orders to specify what type of AI they're advocating for - such as one about using "AI" to help determine foster-care placements. This is another that is almost certainly NOT an LLM, NOT a generative-ai. It is likely a predictive machine-learning model like the palantir-sepsis one theorized above.
And I'm sure I could find out, if I did additional searching. If I did additional reading.
But I just want the first-line-reporting to be more specific.
And I think it matters - socially & politically.
Many of us are ai haters for good reason. And we should be hating on Generative AI, on Large Language Models.
But the hate we have for job destruction, the hate we have for data-center proliferation, the hate we have for increased pollution ... doesn't really apply to these classic machine-learning models.
All machine learning models (to my knowledge) require a relatively large amount of computation and a large datasets to train them. Most of them don't require much compute to actually run. But Large Language Models are on a whole nother level (citation needed).
We should be supportive of targeted Machine Learning like Palantir's sepsis-detection (though privacy concerns still abound, and fuck Palantir they're a horrible company). We should be skeptical of things like "AI" in foster-care placements or that gives sentencing recommendations to judges. The datasets used to train machine learning models have bias. The way the ML models are programmed also may introduce bias. So we should resist some of these uses, and be skeptical when they're in place. We should be vigilant.
But just labeling all forms of Machine-Learning as "AI" muddies the water in a socio-political sense. It makes it harder for regular people (people who are not tech nerds) to know whats what. Hell, it makes it harder for tech nerds to know too.
And what of our legislators? We should not be broadly legislating about "artificial intelligence". We need more specific legislation that deals with specific implementations of ai. The regulations for Large Language Models should be tailored differently than rules for classic machine-learning models. There certainly are legislative overlaps - like privacy protections. But they're entirely different beasts. They should be reported on as such. They should be regulated as such.
P.S. I am no expert in the "ai" space. I'm a programmer and a consumer of programming and tech content, but have no actual experience in Machine Learning in any form. Sorry if I've misrepresented or misinformed on anything here. Its kind of like if I were an EYE doctor talking about how media should report on HEART medicine stuffs.