• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

Tech Thread

CH1 said:

View: https://x.com/heynavtoor/status/2030010676157239600?s=46&t=ZFoeJpONFeL00lgZvCrr2Q
Click to expand...


Yep, AI skeptics have been talking about this for a few years:

1772907545529.png

The benchmarks and contests have all been designed by the model makers are a marketing tool, not by data scientists to properly assess the usefulness of the core technology.

Best case scenario with this stuff really does appear to be highly optimized, less powerful models that work as productivity boosters for people who are already skilled in the area they're using it for. There's just no way for a non coder to be able to know hallucinated code from good code, someone with no artistic or editing skills to be able to fix bad image or video output, etc. So yeah, there's probably space in there for Adobe to keep rolling refined AI tools into my workflow, Suno to help music producers find the right drum kick a bit faster, writing programs to help scientists write papers faster, etc. But there is no AI autonomous world coming in any sort of functional way. Which doesn't mean our Epstein class won't insist on breaking everything so that they can achieve their ROI on it, just that it will be fucking broken and not useful.
 
Last edited:
I keep revisiting general LLMs to help me to do "stuff" and I end up more and more disappointed. They are, for me essentially a Google search distiller, sifting and amalgamating knowledge efficiently. But if ask it to help design an in class exercise for my college students for example it just has honestly horse shit ideas. I know I'm only using free versions for the most part but I've not really noticed improvement over the past 8 months at all.
 
I really feel that the general impact of AI is and may always just be letting idiots bullshit more convincingly.

And, of course, less people bothering to check that bullshit.
 
zeke said:
I really feel that the general impact of AI is and may always just be letting idiots bullshit more convincingly.

And, of course, less people bothering to check that bullshit.
Click to expand...
I feel that way about LLMs to some degree. BUT the ability of basically anyone to make software for highly specialised tasks could have a huge impact. Also I think AI systems built for solving complex problems and simulations could be game changing.
 
I’m excited for its larger scale application like medical research and drug development
 
Aberdeen said:
I keep revisiting general LLMs to help me to do "stuff" and I end up more and more disappointed. They are, for me essentially a Google search distiller, sifting and amalgamating knowledge efficiently. But if ask it to help design an in class exercise for my college students for example it just has honestly horse shit ideas. I know I'm only using free versions for the most part but I've not really noticed improvement over the past 8 months at all.
Click to expand...

Long winded way of saying “I’m the Treliving of prompting”
 
You must log in or register to reply here.
Back
Top