this post was submitted on 30 Jun 2026
94 points (100.0% liked)
chat
8627 readers
223 users here now
Chat is a text only community for casual conversation, please keep shitposting to the absolute minimum. This is intended to be a separate space from c/chapotraphouse or the daily megathread. Chat does this by being a long-form community where topics will remain from day to day unlike the megathread, and it is distinct from c/chapotraphouse in that we ask you to engage in this community in a genuine way. Please keep shitposting, bits, and irony to a minimum.
As with all communities posts need to abide by the code of conduct, additionally moderators will remove any posts or comments deemed to be inappropriate.
Thank you and happy chatting!
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Did some cursory searching and found a few polls suggesting that only around 1/4 people in the US use "AI tools" frequently. Most people just don't care. But the AI companies have infinite money and make up like 40% of the US economy so it makes sense there's a lot of marketing and reporting on said marketing.
The lefty people I talk to are into it without being full "AI-brained" are often running open source models on their own hardware, even for work. Many companies will probably see this as a viable option, or at least running cheap lightweight models for most tasks in the cloud instead of burning millions/billions of tokens on trying to replace their staff with AI. It doesn't work out and it isn't really even cost effective!
It was going to be a challenge to turn these daracenters profitable in the best case scenario where labor can substantially be replaced, and now they have nowhere to go with it.
There are no open source models. The source is all the training data, the weights are open but due to the nature of neural nets this doesn't allow studying, extension, or remixing.
It is akin to distributing the machine code. It runs the stuff sure, but you can't usefully take a bit out or say "Why is it producing text that looks like xyz?" you can only be like "yep neuron 426 has a weight of .112 and activated here. What does neuron 426 do? When did it get that weight? How does the weight change with changes to the source material? Who knows".
This matters to me. The important thing about open source isn't that you can run it for free. We had shareware. It's that you can study it, learn from it, critique it, and improve it.
Couldn't be more correct
I should probably effortpost about the history of opensource and the key characteristics of it vs shareware/freeware and how this relates to LLMs
Too many people have just blindly swallowed the corporate line that they are open source because corps say so. I think this is because most people's exposure to opensource is as free as in beer software you just passively use.
I for one would really appreciate this and wouldn't mind being tagged if/when you do.
Samesies
Literally, open source means something like linux, something that can be studied on the most fundamental level.
That's why deepseek being open source and lightweight was a big thing for homebrewed LLMs