What if there was a cheeseburger that was 50% as good as McDonalds but was only a nickel? That'd be real bad for the McD's share price.
Really depends on the type of output. For content marketing that most people are just going to skim over anyway (aka is just going to be used for blogspam/content farming, which is what it sounds like the linked writer was doing for this client), yes, it absolutely makes financial sense. But a lot of output is very much winner-take-all: music, movies, TV. Only the top 1%, for some definition of top, of that stuff is actually purchased in meaningful quantities. 60% of top quality of that is worth the same as 0%, which is nothing.
> What if there was a cheeseburger that was 50% as good as McDonalds but was only a nickel? That'd be real bad for the McD's share price.
Not so sure about this; you can look at, for example, generic store-brand colas, which taste exactly the same as Coke and Pepsi and are cheaper, yet Coke and Pepsi haven't collapsed after decades.
which is because coke & pepsi uses massive advertising campaigns to ensure the brand is doing the heavy lifting. Coke is basically a marketing company that happens to sell a liquid drink.
If you look at the state of current movies and TV shows, ChatGPT could do a much better job than today's writers. Heck, some high school kids could do a better job.
I think your comment is epitomizing just what a huge blind spot people like you have here. You're comparing AI/ChatGPT with the very best stuff produced by humans, not the average. The average is usually lazy and bad, so it's really not a stretch to conceive of ChatGPT writing something better than that.
As another poster commented here, ChatGPT could have written the latest Top Gun movie. The aviation filming was great, but the story itself made no sense. I can definitely see ChatGPT writing something no worse.
Eh, nope? Like, absolutely nope
Not all errors are equal.
Sometimes 99% as good isn't even close to good enough.
But, yes, for most things 60% for 1/100th the price, is going to cut it.
Kia in the early 2000s offered buy one get one deals on their cars. The joke was always that already you would have one car on the road while the other was in the shop.
They honestly were along the lines of 60% reliable compared to Honda or Toyota in my experience. Enough people bought them though that Kia eventually got it together and a decade later could sell a car every bit as good with a better warranty than the rest.
Cars would literally be disposable. It's cheaper to throw away than fix.
You buy 5 cars and when one breaks you toss it.
This is a great deal
Everything has failure modes though so this isn't nearly as easy of a comparison...
Just as an example on the analogy, there are plenty of cases of pilot suicide bringing down planes too that one would expect would never happen with pure automation... So you can just as easily turn the analogy around and say people are the "cheap but worse" option.
FWIW, we have been able to create processes to manage risks with unpredictable humans, why couldn't we with AI ¯\\_(ツ)_/¯
No, because this is a factual question with a factual answer. You need to actually know which is more dangerous.
> FWIW, we have been able to create processes to manage risks with unpredictable humans, why couldn't we with AI ¯\\_(ツ)_/¯
Because humans and AI are different? We are much better at predicting when humans will be wrong than with AI, for one.
I don't think it's good vs bad, but quality vs volume. If your output is already able to be duplicated at low-to-zero cost then quality is important.
congratulations, you are now the new world record holder ;)
Is it only in the “AI gonna take my job” discussions that programmers insist on calling themselves “engineers” or is this just a recent phenomena?
A bit off topic (but not really) and I know I’ll get downvoted (because, yeah…) but somehow the tide has turned from “programmer” to “software engineer” to just “engineer”.
Just think it’s strange.
Not sure what your standard of “recent” is, but I’ve seen the usage and debate of the “engineer” title since I started programming.
Will it replaced a true star surgeon within the next decades? Likely not. But how many of us have access to the world's best surgeon? Or the world's top 10 or top 1000? Vast majority see one who is local and good enough (or have no choice if you land in an ER with an accident). Those are highly skilled highly trained positions earning high pay.
If you asked me a year ago I would have recommended kids to go into studying medicine and become a surgeon as that surely wouldn't be replaced by computers.
Today I am not so sure this will be a profession for the majority of cases (that don't get to see the highest of skilled) even in 10 or 15 years from now.
In much of the world they regulate to maintain perpetual under-supply and to ensure they are inserted as bottlenecks in strategic/lucrative places.
When AI comes for health care you can bet that doctors will ensure they have a key role in service delivery.
(BTW: doctors are great! I just wish we had more of them and that on the whole they were less over-worked).
Personally I would not be worrying too much for the doctors.
That won't matter. Once people find out AI/robotic solutions are better than the best human doctors, most will choose AI/robots, they will demand them. People aren't stupid.
Nurse and doctor sets almost never overlap.
I've been told though that McDonalds in Australia is higher quality than other countries so this might affect things.
It might be true in a misleading sense that the AI {product} is some large fraction of a genuine {product}, but that doesn't mean anything because the value is contained in those differences above the AI produced baseline. The value floor, so to speak, is a lot higher, but the ceiling hasn't risen an inch.
Those that move down in quality to preserve margin will soon discover there's not a there there.
Maybe we'll get a counter movement soon: "made by humans", as in, the premium experience.
Anti AI sentiment may also grow and make that even more viable, as peoplr may boycot heavy-AI shops.