Let me rephrase: AI is better at human beings at the sort of pattern-matching work that radiology entails, so the majority of radiologists will be unemployed (I'll say in 20 years, if you want to check back down the line a bit and see how I did). This will both improve patient outcomes (fewer mistakes) and lower costs. Some radiologists will remain in supervisory/research/needing to have a human to rubber stamp type roles.
> It may be crystal clear to you that AI will bring the cost of popular education down, but it is not clear to me at all, and adding "clearly" and "obviously" to some future claim does not make it any more convincing. He's trying to work with existing trends and data, you are saying "clearly the trend will reverse just because".
I'm not saying anything "just because." I'm saying that technological changes in the form of AI will cause the trend to change, as technological changes have caused numerous economic upheavals. Nonetheless, I will be more specific - AI will give everyone a completely knowledgeable, fully personalized tutor. It will be able to very specifically identify places where a student is having difficulty and create custom exercises that are tailored to not only the subject, but also the student's optimal methods of learning. It will be able to custom tailor curriculum to fit a student's specific knowledge and career goals. The relative inefficiency of the large-scale lecture model of higher education will be so significant that the value it'll retain will just lie in the brand cache of the institute of higher learning (effectively, you passed the Stanford admissions AI's muster, so you're smart).
I want to see a GPT-based implementation of A Young Lady’s Illustrated Primer-like open world that constantly teaches. We have a lot of the precursors needed to attempt it and give it a shot.
In place of paid mechanical Turk tutors I want to create an epistemological engine consisting of 3 main things: an inventory system of composable pieces of collected knowledge about the game world, a set of story writing design tools to create vignettes and connect pieces of narrative… a kind of narrative procedural generation system, and finally a set of original ui elements to showcase how the “knowbits” can be used to interact with the games narrative. AI will be used for character actors.
If it is obvious to all, then the word is meant to point out how stupid the listener is.
If it's not obvious to all then the speaker is just trying to shut down a conversation.
Clearly I'm right about this. It's obvious.
But, i get the gist of it, we can probably find many more niches that might suffer.
My take on the bigger point is, I wouldn't be so sure about the economy-wide apocalypse. I agree with what he said about the outsourcing scare of 2000's, and if pure cost+customization of education, for example, were an issue, you could conceivably hire a very good personal tutor (professor even) from a low-COL English speaking country, even like 15 years ago, or build yourself your own full custom curriculum for free from MIT videos and plentiful other information that has been around forever, mostly for free. Yet, colleges haven't been abandoned, people are taking on more and more debt every year just to be a part of that system. This is probably not technological in nature. Same for healthcare, there are so much money (and technology) getting poured into that but the system keeps getting more expensive and not too much more efficient, will probably start regressing at some point.
My personal point of curiosity was that autonomous driving, which 10 years ago was promised to have fully arrived 5 years ago, which among other things would have turned 2,000,000 not-very-retrainable-in-such-numbers truck drivers in the US into basically near-homeless people. Well, it hasn't really happened:
> from 2020 you will become a permanent backseat driver
https://www.theguardian.com/technology/2015/sep/13/self-driv...
We've been, too, promised that AI will revolutionize healthcare, e.g. that Watson will replace your doctor (at least, according to the IBM and friends' pitches). Watson turned out to be less than a rounding error in the healthcare equation (i think they got rid of parts of it by now).
The point being, to me it's not very clear cut, and some of what MA pointed out seems really to be caused not by inferior pre-ChatGPT technology but for more societal/economical reasons and incentives (perverse or not).
While it is possible in theory for AI to handle these tasks, they are complex and difficult problems to solve. There is currently no evidence to suggest that AI will replace radiologists in the two decades. Therefore, it is important to recognize the valuable role that radiologists play in patient care and not underestimate the challenges that remain in developing AI technology that can effectively replicate their expertise.
“clearly TV is going to open up educational opportunities in a way that brings the costs way down”
… and that is exactly what happened. Since 1960 every kid in the world “clearly” gets the absolute best education imaginable thanks to the economy of scale afforded by TV broadcasting. It’s basic economics. Oh wait.
Books also increased productivity in the same way.
Would remote learning reduce costs? Yes.
Will AI systems reduce cost, yes.
What cost did the school save themselves here? The paid for televisions, so TV increased costs.
It appears to me that e.g., HS graduate standards have been gradually lowered over time, particularly the further back you go. Maybe this is independent of the screens.
But there are absolutely numerous use cases where being able to display graphics on a screen (or even better interactively) is a massive benefit. Think of any sort of science that's better shown with animation than just a static image on a page for example.
The actual net impact of screens in classrooms, however, appears to me to be extremely negative, and I think learning outcomes bear that out.
> Hospital services? AI is clearly going to be involved, at least in things like radiology
This is very far from the main point: accuracy of the claim that AI will wipe out (in this case health-related) jobs (and/or make health-related services cheaper). "involved" here is so vague as to be meaningless.
> clearly AI is going to open up educational opportunities in a way that brings the costs way down
It may be crystal clear to you that AI will bring the cost of popular education down, but it is not clear to me at all, and adding "clearly" and "obviously" to some future claim does not make it any more convincing. He's trying to work with existing trends and data, you are saying "clearly the trend will reverse just because".
I realize the original post was written by somebody who has a reputation for saying one thing and doing something completely different, etc. etc.