Has AI Been Overhyped? Study Shows Minimal Impact On Jobs And Pay
For the study, 25,000 workers belonging to occupations believed to be susceptible to disruption by AI, across 7,000 workspaces, were analysed.
adoravke
I seriously do not get how people don’t understand this. The absolute dependence so many have with AI, so quickly, is utterly baffling. They’ve completely given up all thought, all pursuit, all effort because it’s the Cool New Thing.
We will NEVER be able to look at content online any more without thinking “Is this AI?”.
From funny voices and videos of famous people playing CoD, to fake smear campaigns, product placements and scams using those very same voices and faces.
How about academics? Why bother writing a paper on a subject when AI can do it for you? Teachers not bothering to create course work because it’s easier to have an AI plan a lesson for you. Doesn’t matter if the information is incorrect because the person checking the work isn’t real! It’s yet another AI, working off of the same bungled data that yours spewed information from.
People are giving up the very idea of THINKING, of CREATIVITY and CONNECTION to a machine purely for the sake of convenience.
Tech-bros stopped reinventing the train for once, instead they created The Torment Nexus
We all thought it was going to be some over the top, horrid machine of endless pointed blades. It’s not. It’s a recursive river of information, slowly poisoning itself on it’s own bad data while people unknowingly or without care, drink from it deeply.
People will suffer, knowledge will be lost and those terrible few will use it to obscure their crimes more so than they already do.
Please, don’t use AI.
Be human. Struggle. Grow.
THINK.
Last night on my playlist two ‘na’ songs popped up in a row.
First was Roxette - The Look
And she goes
Na-na-na-na-na
Na-na-na-na-na-na-na-na
She’s got the look!
Next up was J Geils Band - Centerfold
My blood runs cold (Na, na, na-na-na-na)
My memory has just been sold (Na-na-na, na-na-na-na-na)
My angel is the centerfold (Na, na, na-na-na-na)
On a whim, I decided to fire up a couple of AIs and ask each one which song had more 'na’s and by how many. Meanwhile I wrote a script to ingest the lyrics and count them for the definitive answer.
Personally, my guess was that it would be that it would be pretty close, but The Look, which is both faster and longer than Centerfold, would win. It certainly felt like more na’s.
I was wrong. Both AIs were correct. They were way way off on the numbers, misjudging counts by a whole order of magnitude. But they both got the right answer anyway.
So I asked them why they chose that answer and they both admitted that they didn’t count the 'na’s, but rather looked at the song structure (verse, bridge, chorus, outro) and guessed based on that.
But looking at the lyrics, I still don’t really know how it got that. They both have basically one chunk of 'na’s in the middle and then a lot of them in the outro. What are the AIs seeing that I’m not?
https://genius.com/Roxette-the-look-lyrics
vs
https://genius.com/The-j-geils-band-centerfold-lyrics
In case anyone’s wondering, the counts were:
The Look: 122
Centerfold: 192
Relative popularity of the song in their training data?
Just woke from a dream where I was at a really dreary office party. It was a sort of potluck where all the food was made from AI generated recipes. There was a bowl of green and yellow macaroni (no cheese or other sauce) labeled “execute population all the time.”
So you know that asking an AI why or how it did anything always produces a plausible-sounding but utterly meaningless answer, right? There’s been studies and research papers published on this stuff.
It sounds like an elder Millennial scoffing reflexively at the idea that several decades have passed since high school and then having to confront their own mortality as they actually do the math.
I keep hearing my colleagues complaining that the AI “assistants” that have been foisted on them insert whole discussions into their notes that never happened. It’s not just wrong, it’s fraudulent documentation. And potentially dangerous. As it happens, the one they are using is only compatible with Apple phones, and I have an Android. Bummer, dude, guess I am just left behind. I just cannot express how sad this makes me. I do have the advantage of being a senior physician, nearing retirement age, and in a position where they know they have precious little leverage. So they just let me do my Luddite thing. But, as I watch what is happening to my profession, I despair of where it will be after I am gone.
I’m sure this expresses it adequately
Pretty damned close, I must say.
There’s also Land of 1000 Dances…
TVs are getting worse all the time, and I absolutely hate the user interfaces on most modern TVs that I’ve used. But now they’re shoving AI into them like everything else. Yuck.
On TV, Copilot takes on a “friendly, animated presence” that resembles the opalescent Copilot Appearance Microsoft showed off last month, though in a color that makes it look more like a personified chickpea. The beige blob will float and bounce around your screen, while its mouth moves in line with its responses.
The concept is already terrible but then they had to go and make it look like this:
Ew.
Ew? That’s the cutest Entamoeba histolytica I have ever seen.
Key extract:
Here’s the part I suspect may be too tragically reactionary of me: I am worried that a reliance on ChatGPT will erode people’s ability to use their brains. I do believe that the creative imagination in particular is a muscle, and one that is rewarding to exercise. Recently I was helping a seven-year-old work on her creative writing for school. She had to describe a forest, so I asked her to close her eyes and picture one, and tell me what she could see. Oh, we didn’t need to do that, she told me. We could just ask AI to make one.
I heard secondhand about an editor asking ChatGPT for help restructuring an article. And again, call me a luddite, but I just thought: no! Some things are supposed to be difficult! It is good for the brain to have to rise to a task! I read about someone using ChatGPT to order from a restaurant menu. It is one of the small joys of life to select what food you want to eat at a restaurant. Why cede that to a machine when you don’t have to?
But that’s not even the worst of it, in my view. The worst of it is the way ChatGPT seems to be creeping into people’s personal lives. Using ChatGPT to design a workout plan, to fix a problem in some coding or to summarise a dense document, fine, fine, fine. If you like. But when I hear about people using it to write a birthday card, a best man’s speech or a breakup text, a tiny part of my soul dies. And I don’t think this is the high and mighty position of someone who’s a writer by trade. None of these pieces of writing need to be perfectly expressed or grammatically flawless. They need to come from the heart and be real.
At the root of my hatred of ChatGPT is that people’s willingness to use it in this way implies they are happy to turn meaningful interactions like these into something transactional: a task to be completed efficiently and moved on from.
Has its uses? Think we’ve passed that point of appeasement. It’s an affront to life itself, to human dignity, to real progress.
