there are not any mass appeal viable use cases for LLMs and anyone trying to convince you otherwise has a vested interested in lying to you. i'll cede to there being potential fringe cases in certain sectors, but on the whole not only is the tech "not there" but there isn't a There for it to get to
Reposted from
Hank Green
There are a lot of critiques of LLMs that I agree with but "they suck and aren't useful" doesn't really hold water.
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
Comments
like its going to help us with data management, coding, and shitposts. that's basically it.
nothing you can get anyone to pay more than $20/mo for
using it to do the low-level tedious work so humans can focus on integration is cool.
or stuff like "hey [llm] can you tell me why my code is broken"
they just say it will do everything bc they are con men
Everyone in any field even remotely embracing LLMs will tell you it’s good for boilerplate easy stuff, but still requires expertise to double check.
And it’s going to get worse as it trains on data it generated.
I work in software, and doing the boilerplate like setting classes for database models or video game items familiarizes myself with the code I’m going to work in.
Which if you are using them for slop tasks you would be better off changing systems so you don’t have to do slop tasks
1: sounds like a skill issue
2: not worth the environmental cost
3: probably bad code lmao
If our coders rely on AI online infustructure will likely fall apart.
**have to grade an essay, not hear an are I essay
Good Fuck I cannot type in a touchscreen
Brilliant.
You can't, for example, recognise the use of Alt text on images, and not acknowledge that,
LLMs ain't that
People always pitch that it can "write unit tests", I think largely due to not understanding the value in tests.
This becomes a problem years down the line when maintenence becomes dependent on the LLM to write around this design.