I really want to use AI like llama, ChatGTP, midjourney etc. for something productive. But over the last year the only thing I found use for it was to propose places to go as a family on our Hokaido Japan journey. There were great proposals for places to go.
But perhaps you guys have some great use cases for AI in your life?
I find a ton of uses for quick Python scripts hammered out with Bing Chat to get random stuff done.
It's also super useful when brainstorming and fleshing out stuff for the tabletop roleplaying games I run. Just bounce ideas off it, have it write monologues, etc.
Almost nothing. I sometimes use it to rephrase a question or answer. I refuse to become dependent on AI or contribute to it more than I already unwittingly have.
I’m using Claude (subbed) to help me do qualitative coding and summarizing within a very niche academic framework. I was encouraged to try it by an LLM researcher and frankly I’m happy with the results. I am using it as a tool to assist my work, not replace it, and I’m trying to balance the bias and insights of the tool with my own position as a researcher.
On that note, if anyone has any insights or suggestions to improve prompts, tools, or check myself while I tinker, please, tell me.
I’m a bit disappointed to the practical uses, but I still get some value out of AI. I sometimes use chatgpt to tweak existing SQL scripts at work, and as a trouble shooting assistant. Also I use this tool ultimate vocal remover to extract stems from songs, mainly to make myself instrumentals to practice singing over. Those are really only things I do regularly, despite trying different self hosted AI tools. Most are cool but not very useful.
So far, there have been two interesting uses I’ve seen for chat gpt.
One is I’ve used it to help me write regular expressions in the very rare time I need to for my job.
The other is kind of cool but also kind of troubling in a way. But I’ve come across a couple of therapy style chat bots that are essentially just reading off a list of “here’s what to do for XYZ”
I’ve tested them a bit, and I’ve found I’m 1) concerned who gets access to the information shared. 2) If/when these kinds of bots will be used to manipulate people in a negative way. 3)The possibility of a bot replying in a bad way that could make an issue worse for someone
Overall, I like the idea of them. I find it’s hard to process information if it’s coming directly from myself, or accept compassion from myself. So funny enough, these chat bots actually work really well in that respect.
In some cases, I’ve had better discussions than I have had with actual therapists, which is funny but also sad.
So while there’s some troubling possibilities, I think there’s a lot of positives that I’ve seen from my time with it.
Nothing but have it write stories (not shared or used for anything but just for fun). That, and come up with names for things since I struggle with that.
I use it to see the answers to problems on my physics homework when I can’t figure it or myself. It works far better than forums, which are mostly all paywalled these days.
If you are using ChatGPT for academic purposes, start your prompt with “pretend you are an expert professor on {subject} helping me understand {topic}”
LLMs can be great for explaining things that have concrete solutions, like physics and math problems, when they have a separate “computations” AI bolted onto it, like ChatGPT does. Usually, you can check the answer in the back of the book anyway, so it’s very easy to catch fact hallucinations.
I wouldn’t worry about source hallucinations with this either. I don’t think it would even come up?
General purpose LLMs are starting to replace everyday queries I used to posit to Google. Perplexity can be quite good for this.
Copilot as enhanced autocomplete when writing code. A particularly good use-case is writing tests: with a few test cases already written, a sufficiently good test name will most often generate a well-written test case.
LLMs for lazy generation of SQL queries can sometimes be quite nice.
Writing assistance for things I find myself struggling to get written by myself. A writing unblocking tool, if you will.
It’s reducing the effort and time I have to put into some things, and I appreciate that. It’s far from perfect, but it doesn’t have to be perfect to be useful.
Asking extremely niche scientific questions: I don’t depend on these answers but in the answer is usually the specific terminology I can then search and find the answers I was looking for. I have learned a lot about the properties of metals and alloys this way and what the planet could look like with different compositions.
Re-phrasing things: At work when I’m drained and out of patience I can tell that what I’m writing in my emails is not really appropriate, so I have GPT re-phrase it. GPT’s version is typically unusable of course but it kicks my brain in the direction of re-phrasing my email myself.
Brainstorming: The program has endless patience for my random story-related questions and gives me instant stupid or cliche answers. This is great for me because part of my creative process since I was a kid has been seeing in media something that was less than satisfying and my brain flying into all the ways I could have done it better. I ask the program for its opinion on my story question, say “no idiot, instead:” and what comes after is the idea I was looking for from my own mind. Sometimes by total chance it has a good suggestion, and I can work with that too.
Fun uses which are less common:
Comedy use: I once had it generating tweets from Karl Marx about smoking weed every day. The program mixed marxist philosophy and language with contemporary party music to endlessly amusing results. Having historical figures with plenty of reference material from their writings opining on various silly things is very funny to me, especially when the program makes obvious mistakes.
Language Manipulation: If some philosophical text which was written to be deliberately impenetrable is getting too annoying to read, the program is decent at translating. If I plug in a block of text written by Immanual Kant and have the program re-write it in the style of Mark Twain, the material instantly becomes significantly easier to understand. Re-writing it in the style of stereotypical gen-z is hilarious.
I use it quite a bit. I don’t trust big companies who commercialize AI so I run my AIs locally on my old retired gaming desktop that I’ve turned into a homelab/media server.
I use Kobold.AI to self host an LLM like ChatGPT (Dolphin-Mistral7b if you are curious). I mainly use it for low effort knowledge searches for stuff that is easier typed out long and descriptive (since google struggles with this). Since it’s AI I have to be careful about what I search since I’ve seen it make stuff up but for the majority of what I use it for (Programming syntax, Linux troubleshooting, general questions) it’s pretty good.
I also have Stable Diffusion running as well using the ICBINP model (which is pretty decent for photorealistic images). I use this AI to generate social media display pictures and porn :) it’s fun because it’s a surprise at what you’re going to get but sometimes it generates absolute horrors. Anatomical horrors. Those are genuinely horrific. Other times it’s really good.
How do you set up stable diffusion to run locally? I have been trying out llama.cpp for text and was looking for a similarly easy tool to try image generation.
I’ve found it helpful at work for things like preparing agendas for meetings, or creating an outline of a presentation or document I need to write.
I’ve also found it helpful when I’m trying to Google something where I need to be pretty specific and then I can’t find exactly what I mean by searching.
Add comment