The march of technology
AI is getting a lot of attention these days, and rightfully so. ChatGPT looks to take lots of jobs and to become the plagiarist’s tool of choice. Will AI replace artists, composers and novelists? Time will tell, but to look at what passes for “art” today in the mass consumer market, my money’s on AI.
Another technology that is pushing the envelope is CRISPR/Cas9 genome editing. The New Yorker has a nice piece summarizing the state of play. The tl;dr take is that the genie is out of the bottle on human germline editing and there’s no way to stop it.
I read the New Yorker article with the eye of someone who has written on the topic myself. It is very good and accurate, although it accepts at face value the claim that germline editing is “therapy.” It isn’t. There are ways for couples who carry disease-related mutations to have children without germline editing (e.g., zygote selection, adoption). But the march of technology will not be denied.
My thoughts on the ethics of genome editing are at the link below:
ethics of genome editing
A problem with AI that has surfaced in a number of contexts is its tendency, in some professional applications in law and medicine, to “make shit up”, to use the technical phrase. The need for an actual human professional to review and fact check the product suggests that the humans probably should have done the initial work themselves.
@Jack,
Call it a hunch, but I have a feeling AI is gonna get a lot better in the next five years.
I hope you’re right because it’s definitely not going away.
@Jack,
AI already outperforms human radiologists in diagnosing certain conditions.
https://med.stanford.edu/news/all-news/2018/11/ai-outperformed-radiologists-in-screening-x-rays-for-certain-diseases.html
Joel:
I guess I would ask . . . Doesn’t this have parameters to follow? In which case a computer or AI would be able to track better to the parameters than a human?
@Bill,
I’m sure radiologists have parameters to follow in using images for diagnostic purposes. Apparently, AI is better at applying those parameters. It doesn’t surprise me. Humans come to work tired, are overworked or distracted, or sometimes just sloppy. AI is fresh and rigorous every time.
They said that about self driving cars. They’re a little better than ten years ago, but not a whole lot better.
@Kaleberg,
Self-driving cars are a lot like commercial nuclear fusion–always ten years away.
But AI isn’t just self-driving cars. In fact, it’s mostly not about self-driving cars.
@Bill and Joel,
A number of studies have shown that mistakes in diagnosis tend to be greatly reduced by requiring providers to follow check lists. The providers themselves, not the lists, have to do that. AI can’t do the actual testing and examination. As to forming a diagnosis, many times that is not pure science, based on data, but art as well. That, so far, is human and not strictly computational. With legal applications, frequently the law is not clear and, thus, not subject to results flowing from computer “analysis”.
@Jack,
“AI can’t do the actual testing and examination.”
Nobody says it can.
“As to forming a diagnosis, many times that is not pure science, based on data, but art as well. That, so far, is human and not strictly computational.”
The universe of diagnosis is very large. The universe of radiology-based diagnosis is much smaller. And yet, AI still does better than humans in some situations for radiology. For now. That’s my only point.
And yes, there are many, many applications for AI besides medicine. AI is making progress in many of those as well.
This has bugged me for a while: the insistence upon 300 mile batteries. In the normal course of daily events just how often does one drive 300 miles? Yes yes yes I am sure the nay-sayers can come up with dozens of individual ‘whatabout’ examples but really, in a country of 350 million people on a ball of rock sporting seven soon ten billion people … how often does one drive 300 miles?
Another thing that bugs me is people treating AI as if it were alive, sentient, with free will, congress. I did some work in AI in grad-school; grant it was turn-of-the-century and pretty theoretic but I don’t think the concept has changed: it’s just an ever more complex computer program, only as intelligent as its’ human input. Druidic tattoos and cascading memories
Physics is everything, everything is physics ~ like everything else neither good nor bad until something, or someone, acts upon it …
A lot of pieces in the NYer this year about AI.
Here’s one: Introducing PenceGPT, from the Makers of ChatGPT
(True AI is going to have to incorporate humor, and the ability to laugh.)