Asklemmy

It's good for rapid output of plausibly human text that can then be sorted or assessed for adequate validity or utility. That's all.

I used to spend days rotoscoping people in videos. Generative infill for background painting and automatic rotoscoping have saved probably a year of my life at this point. Image generation relies on CLIP, which needs a language model for conditioning.

I took AI courses in college and it was fun to learn about then when it was a bunch of toy examples that showed the potential of these systems, but it was clear enough to anyone in those classes or doing that research how not ready they were for real applications because of all the known flaws with how model training worked. And then some ceos just ignored all that and started blowing up the bubble.

So my answer is the research models that could play video games kinda good. Everything after that was getting ahead of ourselves.

There's only a few use cases where I've found I prefer it to doing things the hard way.

• As a thesaurus, since it's great for going "what's that one word that sort of means all encompassing, commonly used in reference to research/studies?" and it'll end up giving me "holistic."
• As part of other software, such as how Linkwarden automatically tags bookmarks by category when I add them
• Double checking the answers I've come up with in regard to hyper-specific questions (usually about how a given piece of software can/can't be used, or how it'll interact with something else) just to make sure I'm not blatantly missing anything.

However, I try to avoid using it for anything that otherwise requires productive mental effort, because I find that I end up being a lot more informed and capable if I spend 5 minutes going through sites, learning about a topic, identifying wrong answers, and being able to put together better new queries in the first place, than I do if I ask a chatbot, even if it pulls from those same sources.

When you have a chatbot summarize or combine/condense information, you'll always lose nuance and additional context, and very frequently that context will actually be helpful to your overall understanding. There's also many cases where, for example, someone on a forum explains an issue a bit, and their profile has more related information on it that an LLM simply wouldn't go for, only summarizing from their one response on that page. This can lead me down a rabbit hole that then leads me to finding other good sources. Maybe someone mentions that a particular site is helpful for what I'm looking for, and that then becomes something I use more frequently when I do searches for things, whereas an LLM would have just ignored that comment.

You can't steal data, only illegally copy it. The original data holder still has the data, just you do too.

I find it good for music and film suggestions. You feed it a set of ( I want a suggestion like these ) and it provides a good result.

Also good at building mermaid code for diagrams, just tell it write me mermaid code for this, and drop in a descriptive paragraph, then copy paste the code into mermaid.live

That use case became very useful so there is a paid mermaid page to automate that manual process.

I'm very much against AI slop and hate how it's the most prominent use in day to day life.

With that said, I work for a small government contracting company. We are careful about what we bid on, and of course it's not a sure bet that we'll get it. There is a lot of boilerplate stuff in these proposals. When I was on the bench, my boss asked me to help find some AI tools to help with proposal writing.

Honestly? I can see it being used in cases like this. I wish there weren't so much fluff needed in these things, but that's the hand we're dealt. It's not necessarily worth hiring another proposal writer for what we do, and I certainly wouldn't use its output as-is without knowing what you're proposing, but to get some decent starting verbiage, section by section, to be adjusted after? Yeah, I can see that being useful.

We have a got at work that’s trained on all our internal docs and handbooks. It’s useful because you can ask it operational stuff like, “how do I request time off?” or “when are performance reviews?”

It’s good for stuff where you might not know who to ask or where to find the answer on internal stuff

I use it at work for stuff where it would be inefficient for me to pick up entirely new side skills to only be used rarely and sporadically.

For example, I made a spreadsheet tool to compose ordering spreadsheets in Excel for a system at work that needs them. Most of it uses basic macros that you can record with the basic macro recorder in Excel, with no special skill required, but every now and then I need to introduce functionality into it that's far more complex.

Instead of learning obscure VBA coding for something I do once every two months, I can just tell ChatGPT that I have spreadsheet A called this and spreadsheet B called that, assume that they are both open, and write me a macro that does A and then B and then C and then D between them.

It does it in five seconds, I plug the code in, test it, and then go about my day. That's its positive use case for me.

I needed the following CSS copied 51 times with a 0.05 s increment, because CSS can't for-loop and didn't want JavaScript:

#butterfly span:nth-child(1) {
  animation-delay: 0s;
}

I know I could've just for-looped it somewhere else and copy-paste the output, but I was curious if DeepSeek could do it.

I personally use generative AI for thumbnail art (Stable Diffusion models locally downloaded, with LoRA for the 1.5 models I use), and so does my producer. We disclose when we use the models, and our prompting is actually really good (just inpainting is something we don't do).

I've found lots of great uses. I find LLMs are great for grammar and spellchecking, acting as a sounding board, doing translations, writing shell scripts, digging through unfamiliar code bases, figuring out configurations for tools, finding relevant stuff in large documents, and they can be helpful for coding in languages I'm not well versed in.

I've tried ChatGPT a few times to see if it's useful for me, and it's worked surprisingly well in most cases.

I made a website that needed two modal images, one on the top and one on the bottom. I wanted them to be enlarged when they were clicked on. I found a load of guides for getting one to work, but I couldn't get both to work. A few minutes with a prompt got it working. It didn't help me to learn JavaScript, but did give me working code that I needed quickly.

I've used it to fluff up some text. I'm not very good at making things sound good in text, so it helped a lot.

The latest one I've tried is getting camera settings for a dark gig setup. I was able to give it an old photo that was under exposed but gave an accurate impression of the room, and ask for recommended settings with the same lens, a new lens, and a flash. It gave me a selection of settings with and without the flash, including settings for rear curtain sync, so when it leaves a ghost trail behind the subject. It's nothing I couldn't figure out, but would have taken a bit more trial and error in the room. I probably wouldn't have thought of the ghost trails.

Giving me a paragraph where Tom Cruise grows increasingly frustrated during his juggalo-themed commercial for shrimp fried rice.

It has been great at estimating calories in things. When possible, I compared to the actual food label, and it's usually within a reasonable margin of error. But not everything is labeled, and when you're on a diet it's better to have at least something to go off of.

• I use it for research and then verify its findings.
• It’s excellent at summarizing quickly.
• It’s great at idea creation for specific needs and outlining it.
• While it’s good at writing I enjoy that and do it myself to keep my specific tone.

That stupid song about the Windows registry. Don’t look it up, or at least don’t come crying to me after listening to it 50 times. It has no right being as catchy as it is. I fear for younger people and how God generated music will sound in a year or two.

I used to ask why classical music wasn’t performed by Metal bands. Now I can have AI make it for me and it shouldn’t even be that hard given the songs are already written. Now if I were to say “compose a new symphony in the style of Beethoven and make it sound like Iron Maiden are playing it… that would be different.

As a teacher, it's good at making assignments for students. Though I don't use it out of principal.

A couple so far. I have a local copy of Stable Diffusion. It's handy for upscaling some kinds of images. I'll also use it to flesh out my worldbuilding project with landscapes and scenes.

Less often, I'll consult ChatGPT, without logging in. For the times when a search engine doesn't cut it but a forum post would be too much. I'm usually skeptical of AI summaries, but I find it justified for boiling down poorly-written stuff that I have to read, but isn't worth my time in long form.

I actually just view it as the latest abstraction of search. Yahoo in the 90's did not give you a blurb summary of links or would do math equations for you or give answers to simple questions like what time is it or whats the capital of alaska.

I've been using it recently for generating alt text for images (my bots on Mastodon and Aunty Madge on !yoursinclair@retrolemmy.com specifically). It's pretty good at that, although does sometimes give weirdly wrong details - especially the TED Music Bot, if it gets the usual +4 startup screen it says it's +4 on key F1, instead of 3-plus-1, and tells me the wrong colours for the text and background (I think it may be getting it confused with the C64, bit the colours are right there on the image!). It's infinitely preferable to having no alt text, which would be the alternative.

The other thing it's really good at is summarising articles.

I've also used it when I've had an error in my code I can't track down, or a bracket missing that I can't figure out. It quite often gives nonsense but I've had some success. Usually a normal web search is perfectly adequate though!

Best use for me has been feeding it logs and pcaps, incorporating it into Wireshark sounds far more useful to me than summarizing a two-line email.

I find it's good as a "search engine of last resort". I would pay $2 monthly at most for one, and less if it's American.

I only really use LLMs for project ideas, naming things, or recipes, but it’s great for those. For recipes especially, trusting the gaslight machine adds an extra layer of suspense and fun to cooking or baking.

It’s also fun, but not useful, to ask Linux questions - chatgpt specifically told me to restart dbus while in an active GUI session when I asked it about a simple case of unintended behaviour. It’s probably the dumbest Linux advice I’ve heard to date, so I got a lot of enjoyment out of trying it to see how bad it would fuck things up

Despite claims to the contrary ChatGPT is very accurate with its responses when such responses involve web searches. Where it falls apart is complex multi step things like coding questions.

I make heavy use of it to skip past all the clutter of Google search results and end up with clear summaries that answer my questions. That's all I'd really use it for as anything more than that its output is highly variable in quality.