Is it AI or not

A radiologist assistant is not a Large Language Model.

I would expect to some extent, image analysis would be a "module" on an LLM, and not a part of the main bit.

Bare minimum, it's a neural network, trained on images, one at a time, that slosh around and train the neurons.

For example, something like YOLO_5 (You Only Look Once), can be trained to identify animals in photos. It draws a box around the presumed animal and names it (or whatever). That uses a lot less hardware than a Large Language Model, and less storage. The article had a picture with a bear in it, and indeed, the bear had a square drawn around it.

But as for whether the "quality" is there, that is another issue entirely. In my opinion, no radiologist would ever trust something as sketchy as YOLO. Radiologists are very particular about their jobs, as they hate getting sued. And I can imagine the look on the judges face when you tell him "yer honor, I didn't even bother to look at that film, the computer told me there was nothing there". Some lawyers recently, learned about what happens when you "phone it in". Professionals are still on the hook for the whole bolt of goods. The computer isn't going to get sued for "being stupid", because it is stupid.

It would take a *lot* of films, to train a radiologist assistant. Who would have a collection, large enough for the job ? It would be a violation of privacy law, for a bunch of hospitals to throw all their films into a big vat, for NN training. It's not like crawling the web and getting access to content that way.

While a lot of individuals and their jobs can be replaced, the radiologist will be "the last to go". Regular doctors are quite dependent on the radiologist taking the fall for mis-diagnosis. The doctors would be scared shirtless, if the professional that "has my back" was as stupid as a computer. The doctors would quit. Doctors do not read films. They say stuff like "the radiologists report says you have t*ts". And you can then take that to the bank. They don't use their knowledge of Grays Anatomy to figure that out. They identify the radiologist as the source of the information. The radiologist is their "God".

Paul

Personally, I'm sick of ths AI crap which seems to exist only in the minds of the tech idiots. When it devolves into the lives of us common dummies, I'll worry about it then.

By that time, it may be too late.

I promise you, people in the programming business have been talking about it for a long while.

An AI doesn't need to pass the Turing test to be considered an AI.

Here. This will get you started:

My next mammogram might be analyzed by an AI in addition to a human being.

The term "AI" has been misused by media and most non-computer scientists. The current crop "AI" tools (e.g. chatGPT) are not artificial intelligence, but rather simple statistical algorithms based on a huge volume of pre-processed data.

Which leads to

That's why I said "popular", to exclude that sort of thing.

Okay, but doesn't it have to be more than a single purpose algorithm? Otherwise, cars have had AI since computerized fuel injection, but nobody called it that.

So they use in in conjunction with AI. Why do I need to know that? I don't know squat about any medical test. I don't have to. If they want to use AI to bake bread, what difference does that make to me?

I'm sick of reading about this AI crap. I don't need to know where it's being used.

That's exactly what I thought. Yet these ignoramuses who hardly understand the Web software they use everyday keep burping about this AI stuff when they probably would fail at learning Basic.

Good Lawd! Next It'll be our turn signals that are AI Intuitive...Jeesh!

(I'm just about at the point where this AI nonsense is going to end up in my Plonk! file.)

Oh, I think it will be a while before they come up with the Killer Robots.

It sells. That's all radio and TV first think about.

I don't know. Why don't you tell us.

Perhaps not. I think it's better to know than not know.

More amorphous defintions of AI.

Here is a post made by someone else that I copied a while back which makes sense to me.

--------------------------- Subject: The Reality of The Bullshit of Quantum Comps Breaking PGP

"Building a universal quantum computer, one that can perform essentially any computation, is an extremely challenging technical problem. We're far from having solved it."

"To crack a 2,048-bit RSA key, such as the ones that today's standards require, a quantum computer will need at least a register of 2,048 entangled qubits. That's far from what's available today. And it seems very unlikely that the current rate of progress in creating more entanglement will make it possible in the next several years."

"For now, it seems hard to justify worrying about your encryption becoming vulnerable to adversaries with quantum computers. It seems very likely that NIST's effort to standardize encryption algorithms that are quantum-safe will be completed and widely deployed well before quantum computers are a serious threat to security."

Amen.

(Now everyone shut the F/Up 'bout AI, Dammit!)

From Wikipedia: "The Turing test, originally called the imitation game by Alan Turing in 1950, is a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human."

Of what good is AI if the product of it is dumber than a human?

One of your dumb Killer Robots would be dead meat after its first kill.

That's progress? We have cities run by Dumbocrats with human non-AI killers who are getting away with more killings than the combined number committed in our war zones over the past 30-40 years.

So, how dangerous can AI really be?

Already there:

How long will we have to wait for the human size version?

Not quite...

These aren't quite the GPUs in your gaming PC:

The dirty little secret is after you've trained your model with the training dataset, validated it with the validation data, and tweaked the parameters for minimal error you don't really know what's going on under the hood.