upvote
> Or, you could sign up for Opencode for $10 and just be productive.

You forgot the step before where you spend months waiting for security to vet it, legal to sign off and procurement to approve it.

Or you could use hardware your team has lying around. Everyone isn’t working on cloud-hosted CRUD APIs.

reply
You might be anti-ai in the sense you aren’t comfortable with all your data being shipped back and forth to a third party.
reply
Im anti AI in the sense of VC backed global warming far right accelerationists.

Local AI is almost perfect. But its like all democracy: its history is marred with lots of crap.

reply
All of your objections have already been addressed by the previous comments.

The original comment states that the person in question already had a suitable graphics card to hand, so it did not require a monetary investment.

GP clearly states that "someone who is into that that stuff" was guiding the process, so it did not require a significant time investment.

> I'm particularly calling out the hypocrisy of the original comment. Being Anti-AI, and then spending hours on setting up a less than frontier AI model.

I see no hypocrisy in the original comment.

You've also assumed the skeptic in question doubts the capabilities of AI. That may be the case (like you, I have no idea), but they may also have privacy concerns, in which case a local model is the appropriate choice.

There are plenty of reasons to be skeptical of AI.

reply
install LM Studio, download the automatically selected quant based on your hardware, start a conversation with the automatic context size. 10 minutes at best and zero effort
reply
Maybe not '0 investment and 0 commitment', but incredibly little depending on what you have laying around. It takes less than 5 minutes to download say LM Studio and an Open Model and as long as you have the hardware to support it, you start moving along. If you are on AMD in some ways it's even 'easier', you can download Lemonade and it will tell you exactly what will fit and best options based on what you are trying to do.

For me at least the local AI stuff, powered with solar has been pretty great. Would that scale to a large business? Goodness no, but for my tinkering and learning, it works great.

reply
> You at least need a capable machine, so that's not 0 monetary investment

It is 0 monetary investment if I already have said machine lying around doing nothing.

Which is exactly the story OP talked about.

reply
But most people don't have an RTX 5090 lying around, so the story doesn't apply to them, right?
reply
You don't need a 5090 to run local AI. A whole lot of people out there are doing it with Macs. Unified ram is the biggest thing.
reply
Correct. If the premise doesn't hold, it's ex falso quodlibet for anyone.
reply
I'm running Qwen 27B no problem with an AMD 9070XT + 24gb DDR5 ram. Does basic web search for me (tool call with tavily, costs nothing I get 1000 searches a month) and is great for creative writing (primarily breaking writer's block). Until the recent surge in ram costs, that wouldn't be hard to do. I built the computer for ~$1600 a year ago.
reply
I am getting ~13-15 tps with my 9070XT for the 27B (~35tps for the 35B-A3B), but I think for me the main bottleneck is the 64gb of DDR4 3600 memory. What kinda speeds are you getting with what speed of DDR5?
reply
Back in “the day” nerds just bought the hardware to fuck with. Some of us still do. Claiming that compute is the barrier to entry just means you’re not a nerd. That’s ok.
reply
deleted
reply
> You need to setup a harness that works against your model - You might need to setup additional websearch tools, image tools, etc since harnesses like pi don't come with the model

Pi has a nice guide on it (https://pi.dev/docs/latest/llama-cpp) and it is really not that hard.

How is that hypocrisy? Self hosting is somehow anti AI? Its not anti AI. Its literally using AI!

…and honestly, at a higher technical level than slapping your wallet against a token provider and running prompts in a hosted sandbox you can't even see the prompts in.

reply
deleted
reply
Look man I am incredibly skeptical of how LLM’s have been rolled out and all the promises people make (it’s so much snake oil and pipedreams), but I also found it very trivial to hop on LM studio and start tinkering with models. If you’ve already got a decent midtier computer on hand, which I imagine a lot of us already do, then it’s really not hard to get started and get immediate results.
reply