A cool new technology that can do marvelous things has been released to the public for free. It has never been done before to make such a sophisticated and expensive technology available to everyone for free. Promising the democratization of skills that take years to learn to give to everyone equally (everyone that can pay for access to a computer somewhere that is).
Loads of people that cannot afford to go to good schools and universities to learn computer science or 3D graphics design,you name it are lured in by this promise.
They even get hooked to the power of being able to do things they could not even imagine before.
And then the axe falls:
specialists tell them it does not work as advertised
they tell them you’re being lied to by billionaires
they discover it costs a lot of money and was only free because the companies making it were borrowing loads of money using the same lies they told you about this technology
Would you believe that? After all you were able to do with this tech?
I understand. But rip off the band-aid, because it is true and the specialists trying to warn you for years are tired too.
Maybe I prefer get suggestions at the design stage of one functional modules(i am not sure wheather that is a professional word). For example, if my game needs a simple, intuitive but efficient backpack management system, i will search forum’s help. This is a kind of work from 0 to 1 which LLMs lacks.
That was sarcasm on my part, which I know doesn’t come through well in writing. Just keep in mind, that those of us who are frustrated by people like you using LLMs are the ones who supply the LLMs with the information to answer your questions. If people stop asking us, or if we get annoyed enough to stop answering, then that well of information dries up and LLMs cannot answer questions about new things.
LLMs are literally destroying the sources of information that power them. And that has long term consequences that it will be too late to pull back from by the time they are fully felt.
It’s a multi-headed beast, because even specialists get bedazzled and Nobel prize winning AI professors are either bedazzled enough or somehow willing to lie for pay to keep selling this lie.
It’s a form of psychosis that occurs when all the sceptic voices are drowned out.
Is there an English equivalent for the saying: “al is de leugen nog zo snel, de waarheid achterhaalt hem wel”?
“The lie may be very fast, the truth will catch up eventually.”
The longer it takes and the more are convinced, the bigger the pain too, so it seems.
A valid use if a language model is to help you structure your own ideas. But find one on the interwebs that does not need an internet connection to work, so you are sure it is not burning up the planet.
Or just use this forum of course. And take your time to learn stuff well.
There’s a variant which is “a lie will get around the world before the truth has even got its shoes on” and similar, probably conveying the same meaning
I ultimately disagree that it is useful to organize your ideas, at least in any real way that’s better than just writing them down or structuring it, and that part is without the data theft and it hallucinating things for you, or deleting all of your work (or injecting bigotry in it, etc.)
Never used it myself so can’t speak to it specifically, but on the basic concept that you need it in that way
Also, it’s always important to remember one thing when considering the output and what biases might be in it:
It was trained on the internet, meaning the volume of content here affects the output. The input it got includes things like Reddit and 4Chan, which I think anyone can imagine affects the output
It ultimately reflects humanity, in our best and worst, and it is very much not free of our collective biases
But unlike picking a person to collaborate with on a project, or doing your own writing, you can’t control what bigotry it “chooses” to regurgitate today, and it is very unreliable, for some topics it might be very progressive, for others extremely regressive, simply based on what the dataset looks like for that particular topic
Yeah I was imagining a local instance trained on a corpus you control.. but that defeats the purpose, I guess, because a normal person (not CS schooled) will not be able to create such a thing in the first place.
It got a really good point. If an LLM fails to properly understand something and that misinfo gets distributed across many places, then another LLM end up training on that incorrect information. Over time, this could create a loop of misinfo!
We’re cooked then!
Sorry, I am late in the party xD 135 is a lot to read xD
Even without the “AI” suggestions at the top of Google the internet is slowly but surely becoming slop, with generated content both polluting the available sources, as well as making more and more text functional poison for the models as training a model on the output of a model has a lot of inherent issues
It’s even getting to the point where reputable scientists are citing hallucinated articles in their theses and papers, and that gets worse and worse as the databases for citations are themselves sourced from existing data and contain more and more slop
You even have lawyers citing non-existent legal cases
People in the 1940s:
We used to search the whole library for a tiny piece of info. I spent the entire day going through books just to find out what language Waiting for Godot was originally written in…(It’s French )
People in the 2060s:
It was hard to find an answer on the internet because it was full of SLOP!
It just makes sense, the Godot maintainers made headlines back in February after whining about the slop being fed to the project. Why continue to burden themselves and the community?
Reading this just made my day. I have a strong anti AI sentiment and as a professional my policy is not to use it (unless forced to) as I have experienced first hand how much I progress and learn by writing my own code.
I know I shouldn’t, but whenever I scroll into social media I get overflown by pro LLM posts. Most of them are just hype slop, but sometimes I read someone I respect encouraging their use and get really discouraged because my experience in several work environments has been awful. Not only do they lead to awful slop, most of my colleagues are delighted by it even though it makes things some much harder for us. And don’t get me started on bosses and managers and so on.
So far game dev circles and godot in particular were my comfort zone, mainly because the “artistic” mindset tends to be against AI of any sort, but it seems that is also changing
This is a kind of work from 0 to 1 which LLMs lacks.
)