If there is a job that requires generating bullsh*t, then definitely it is at risk. Otherwise, no.
A lot of job are at risk. Most of them no because of full automation, but because less people will be required because of productivity increases "thanks to" AI. And many more job will become boring.
The problem is this productivity gain will hurt a lot of people, and there are probably/possibly only a small minority of people that will fully benefit from it
The human with the art history background at least has some credentials, whereby the output can be trusted, to some degree and explained if necessary.
This is, imo, such an obviously near term capability that its not worth bringing up as a limitation.
That model doesn't exist.
We have models that can read text and then probabilistically create a summary that's the most likely to match the original.
But none of these models "extract meaning" in a way that can be used for fact checking purposes in the way you describe.
What you're describing is basically a world model, and while there's a few folks claiming these LLMs are showing the glimmers of building world models, it's mostly speculation.
I asked 3.5 to do it for you.
> The paragraph that explains that models cannot be used for fact checking is:
> "But none of these models "extract meaning" in a way that can be used for fact checking purposes in the way you describe."
It’s not hard to imagine a very simple tool to try and find citations for claims. Or perhaps, derive claims from a source with citations.
If i know for sure it is 100% GPT-X generated bs, do you think I'll read it?
I acknowledged in my first post that a particular quality standard had already been met by AI. I also concede that this quality standard could be sufficient for many businesses. God help us all if it's 90% of businesses.
Even for stuff like movie making, a director or producer would hire an illustrator or team of illustrators and send off a script to get it storyboarded and probably have to wait at least a few weeks. Now they can just go to a prompt engineer and have it done in a single afternoon. There’ll be similar developments in VFX, if you have a look at the latest Unreal Engine 5.2 demo they do motion capture from a woman’s face from a mobile phone camera in a matter of minutes. The actress in the demo says that kind of thing used to take a month to get the same result. When you’ve got a bunch of competing streaming services and studios trying to churn out as much content as possible to keep subscribers, it’s going to be difficult for them to justify the slowness and cost of keeping those human workers compared to the AI.
I believe GPT-X will generate tons of ... something ...
How you would force people to READ IT ?
If you know your incoming info is 90% BS, not a guess, not a thought, just knowledge that this is all auto-generated by some multiCPU entity - why would you read it? What are going to gain or to learn?