I think that’s the most likely scenario. So far, no one really knows if AI will be able to be truly creative, coming up with new ideas for things that people actually want done.
I recall a story that was like that. The AI could do almost anything it was told to do, but couldn’t come up with ideas about what should be done. One example was a character saying, “Hey, with all the advancements in medicine we’ve seen, why do I still have to brush my teeth every morning to get rid of bad breath?” Once the idea had been formulated, it was trivially easy for the AI to fix it, but the notion of fixing this would never have occurred to the AI in the first place.
So we’ll be The Idea People, while the AI does the grunt work of putting the details together.
I think there’ll still be some value for human interaction type jobs where there’s a fair degree of interpretation or teasing out what someone’s really looking for.
I keep thinking of my old job where a big part of it was being able to understand both the IT landscape, the customer’s business processes/model, and then knowing that, try and figure out what sort of solution they were actually asking for (rarely is it what they start out asking for), and how to determine whether that solution was successful or not.
An AI isn’t quite there. It’s the follow-up questions based on the first round combined with the business
model/IT landscape that are the tricky ones. It’s not even an easy job for an educated human to do well.
I imagine that other jobs like that- psychologist/psychiatrist might be one that’s really hard to replace with AI, and so would a lot of teaching-type positions that require a lot of intuition as to why a little kid is acting like they are, or not learning with that particular teaching style.
What the AIs are really good at is identifying patterns that aren’t easy for humans to see and generating results based on those patterns. Like say… going through a million patient histories and figuring out how to diagnose diseases. Or maybe looking at a bunch of customer profiles and purchase histories and predicting what sorts of things a particular person might want for Xmas. Or even something like being trained on a million pieces of garbage, and being able to sort recyclables from straight up trash at the solid waste processing facility. And of course, being trained on a colossal number of images and being able to generate images that combine the user’s criteria like DALL-E.
There will be problems here, too. Who gets to decide what ideas AI gets to work on? We’ll be fighting over that, for sure. Will we ask AI to solve the climate crisis, or ask it how to destroy the nation next door? Or will someone ask it how to mitigate humanity’s inbred savagery toward one another?
Now this is something I’d rather see computers/robots do. I’d rather not depend on some minimum wage, over worked rest home employee performing this task on myself or upon my loved ones. I’ve heard too many stories about incompetence and neglect in this area.
Bring on the AI ass wiper I say! Hopefully before I need one.
This is exactly what’s happening right now. They are designing cockpits now for them to be “optionally piloted”. Whatever comes ff boeing’s or Airbuses design teams next will be intended to be non-pilots from teh git-go. With human pilots as interim helpers.
UPS, Fedex, & USAF are actively planning on flying airliner sized airplanes with one pilot + an AI helper before 2030 and with zero humans before 2040. In normal commercial / military service. Not as experiments on a closed course do not attempt at home.
When building elevators first were being automated, the general public said nobody would dare get on an elevator that did not have a human at the controls of that wildly dangerous machine. Nobody. They were wrong.
You can learn a lot from reading, especially if what you’re reading is the entire Internet. But I still maintain that you can’t learn everything that way. The first truly general-purpose AI won’t be able to just train itself by consuming already-existing data. It’ll have to be taught. Which will take the same amount of time, more or less, that it takes to teach a general-purpose human.
And then once it’s trained then taught as you suggest, we make 10,000 copies of it overnight. That’s the hard part for humans, but is trivial for computer programs / data sets.
formula + bottle (c’mon, bloke, who the fuck hires a wet nurse even today??)
one donor = however many thousands of viable sperm, so okay, that’s like a .0001% chance of job security
ETA: I suck at math, the sperm security job (why is that a phrase I’m ever typing in my life?) number was pulled out of where the, uh, sperm don’t shine, but I stand firmly by the general thrust of my argument.
I’m told artificial porn is already a thing. It suffers a bit from the uncanny valley Polar Express effect, but it’s out there.
Of course porn being porn, the modeled characters tend to be even more … superhuman … than the carefully curated subset of humanity that are live-action porn actors of either sex. Soon enough they’ll know what sells best: wildly superhuman, slightly superhuman, or just top 1-2% youthful hotness.
The example of elevator operators speaks to the reason why I’m deeply skeptical of the idea that we’ll genetically engineer humans to adapt to new environments and fit niches such as in outer space. Brave New World, published ninety years ago, postulated a future centuries from now in which a class of idiots were deliberately bred to fill vocations that would be unbearably boring to people of normal intelligence- including elevator operator. Aldous Huxley was able to envision of future of artificial gestation and customized humans, but the concept of automating elevators was beyond his imagination.
I’m reminded of a short story set in the far future: true general AI exists but requires so much intensive training that except for niche purposes old-fashioned meat bags or human minds that were uploaded are more economical. Then someone gets the brilliant idea of using nanites to repair the freeze-damaged brains of early cryonics pioneers, which gets around laws against “wetware” on the grounds that the brains are legally dead. Thus one poor sod finds himself revived as a brain in a vat expected to work as a slave, until the story’s protagonist rescues him.
