What do you use AI for?
Ok, so the best thing that I have been using my local AI for lately is acting as Socrates and helping me delve into my thinking
I'm still pondering my next response, but this is great, I have a bunch of other ones on virtue and reasoning, but you get the gist. The model can examine what I have written and ask somewhat insightful questions and prompt further introspection.
I'm building a trip itinerary for upcoming travel. It's helpful but very iterative. Gemma QAT has weird memory
Practicing my drawing, then it brings the photo more to life
My girlfriend and I have been doing this thing where we take 30 second turns creating a dumb drawing and seeing what it pops out after 2 or so turns.
Comes up woth some cool results she goes crazy for cause it's dumb art we made together
I love this answer! What a wonderful context for a generative model to bring people together and add positivity to daily life. Thank you very much for sharing.
Yeah totally agree. That’s a novel use I’ve not considered.
I use it to write Ansible.
You see, Ansible is horrible and toxic to brain cells as it is, and I'm really not interesting in retaining anything related to it. So, by using Ai to write most of the code I'm ordered to write in ansible, I don't burn out brain cells better suited to holding knowledge about something that isn't a sad throwback to what we were doing in 2003.
Ai protects me from RedHat's slop 'automation'.
Oh no.. Ansible is on my "to learn" list to help with various deployments. Is it that bad? Do you have you'd recommend instead?
A fine use of the tool. You could almost consider it PPE.
Jesus, are you me?
The most useful thing my LLM has done is help me with hobbyist computer coding projects and to ask advanced stem questions. I try to use my llm to parse code that im unfamiliar with and to understand how the functions translate to actual things happening. I give it an example of functioning code and ask it to adapt the logic a certain way to see how it goes about it. I have to parse a large very old legacy codebase written in many parts by different people of different skill so just being able to understand what block does what is a big win some days. Even if its solutions aren't copy/paste ready I usually learn quite a lot just seeing what insights it can gleam from the problem. Actually I prefer when I have to clean it up because it feels like I still did something to refine and sculpt the logic in a way the llm cant.
I don't want to be a stereotypical 'vibe coder' who copies and paste without being able to bug fix or understand the code their putting in. So I ask plenty of questions and read through its reasoning for thousands of words to understand the thought processes that lead to functioning changes. I try my best to understand the code and clean it up. It is nice to have a second brain help with initial boiler plating and piecing together general flow of logic.
I treat it like a teacher and an editor. But its got limits like any tool and needs a sweet spot of context, example, and circumstance for it to work out okay,
I do similar, I'm not a professional, I use the LLM as a tool to enhance my abilities
The whole vibe coder thing is weird...it is like a hobby woodworker who has powerful tools but doesn't really know what they are doing, yes you can knock together a functional table, but it doesn't look that nice.
Only thing I use it for is creating portraits of my D&D characters to use in virtual tabletops
I want to do this, it is next on my list when I get some time to myself. I have a game that I am playing and it would be cool to generate a character picture.
I am terrible at drawing so it is not really something I can do manually.
Right now, I am using a model to write agentic tool calling functions in Doom Emacs. Specifically, I am having it open a new buffer and navigate Emacs documentation to cite sources. I also pulled up a Python script I wrote that converts any length text to speech and working to integrate it. I'm mostly developing the key/values needed to convert things like "JSON" in text to become "Jay Son" for the speech model, or like "GPT" to "Jee Pee Tee". All of this is done while only running models locally. I also like messing around with diffusion stuff too.
What're you using for local diffusion?
Not the person you asked but I used kobold.cpp to generate images with SD models. it works as a okay introduction to image gen. Their wikihas everything you need to get it working
Mostly ComfyUI with CyberPony for any NSFW stuff. I also use Flux SD1.x and SD3 sometimes.
I have it role-play for my amusement, bounce ideas and regularly ask it to give me 5 ideas about something, if I'm not creative myself. I also use it for translation. Either copy-paste an entire text or help me phrase things.
I've also fooled around with image generation, music, tried to program a website, had it rephrase my emails... But all of that wasn't super useful to me. At lest not for my everyday tasks.
Other AI things I use (occasionally) are: Text-to-Speech, Speech-to-Text and text recognition.
Entertainment, laughing at small models trying to sound human is the only thing I've ever really been interested in with LLMs
HELLO.
I AM DR SBAITSO.
Replied to the wrong comment
I used a local model based llm thing to separate a drum part out of a song that I was working on learning with some friends. That was a lot of fun, and didn't feel like I was ripping creators off.
I also recently tried out gemini's research mode thing to research a subject I already knew a good bit about as a test. It got some decent answers and showed its sources, which was great for double checking its work, but it still had that oddly verbose and repetitive quality to it. I asked for an audio summary and it spat out an absolutely horrifying uncanny valley fake podcast on the topic. I still kinda shudder when I think about it.
I use it to translate text on websites when it can't be translated by the browser, sometimes to translate ebooks, to identify bird calls, and to make user pics to use online.
Writing code in languages I'm not familiar with yet. Reading code is much easier than learning all the syntax details, so I can still verify that it's correct. I've even written a poc in lang 1 and used LLM to rewrite it in lang 2.
I use QwQ-32B with 128k context to parse web queries when, for example, I need to heat my basement.
Wait, but what if the user doesn't have a basement?
So this was fascinating and I've since installed gemma3 on my laptop. It's giving me gardening and BattleTech tips so far, with varying degrees of usefulness 😅 still, very cool and I look forward to playing with it more.
Google and Python replacement.
I use it to create music: https://youtube.com/@codedtone
I also use it to generate the thumbnails and album art for said music.
In addition, I use ChatGPT to brainstorm ideas for various campaigns I'm currently DMing including ideas for encounters, characters, locations etc.
I use it for driving directions
/s
That's funny. When I first starred using an LLM, I threw a list of cities at it and asked for the shortest trip though all of them ( traveling salesman problem) and it was so bad I immediately knew not to trust answers from then on
For once I'm glad of the /s; I can legitimately see people using AI for driving directions