#usecases of ai

[3-loch-prinzessin]

glad to know that all my horny posting on here and from my old blog has been used to train AI...

[aiwikiweb]

Transform your videos with 10LevelUp! Create short, engaging clips for social media automatically with AI. From highlight selection to auto captions, grow your channel on autopilot.

Learn more : [ aiwikiweb.com/10levelup ]

[gurupanguji]

+ AI tools are constantly "6 months" from really getting awesome • Birchtree

AI use cases aren't here yet and I predict it's not even 6 months away. #ai #llm #tech

A challenge these A.I. tools have is that we’re coming down from the novelty high we had at the launch of ChatGPT a year and a half ago and now they need to deliver on actual use cases that make people’s lives better. As someone who uses Claude almost everyday, I think there’s clearly something there, but the big promises from these companies feel like they’re constantly falling short of those…

[wall-maria-around-ba-sing-se]

I know some people have very fair concerns about people calling others "gooners" all willy nilly but until there's a better word for "transvestigor-adjacent 20-30something year old man who thinks it's a conspiracy whenever he looks at very normal or even extremely beautiful women and gets mad because she does not look like an anime child/blender animation model/ai porn" that also makes them upset to be called, this word will still have a legitimate usecase

[shrine-of-the-exoletus]

hm. interview opportunity with an ai-based surveillance state company, which is a naturally scary thing, but when i check out their website their usecases are all like "we can tell if a car parks in front of the bus stop and give them a fine. we can tell if a car drives in the bus lane and give them a fine. we can tell if someone parks in the bike lane and give them a fine." so maybe this wouldn't be a horrifying company to sell my soul at? hard to judge

[ashe-delta]

I have a feeling that people might not read a long ass text post, so to just clarify: I don't like using AI, but I don't think AI is a demon for existing. The people using it are just assholes who don't understand it's actual usecases, and its going to get a lot of people hurt if people don't understand this (both the people calling it a demon, and the AI's usecases).

And for the record, while I'm here, a lot of people in the scene, outside of blind tech bros, don't actually think AI is all that smart. For one thing, the way you have to twist prompts to actually tell you what you want is pretty silly, and something I learned in the class. It's also infamously bad at math. Actual engineers know its downsides, but engineers aren't the ones raving about it, are they?

[rainydayscore]

oh artists have such brainrot now they literally forgot that AI has uses and purposes aside from image generation, they can't even concieve of alternate usecases anymore

[zorciarkrildrush]

ChatGPT and Copilot and to some extent Perplexity are really good as shortcuts for programming. You still need to know programming to catch mistakes or inefficient code, but over all they're fantastic when I already know how to program some thing or other, and because I know all the steps and this isn't artistic expression I can have the machine do the steps for me, with some mild editing at most, and save time. Even in the worst case, where it completely fails to understand what I want, I can still copy useful bits.

This is the definition of using a tool - anybody thinking you can now replace programmers with it is dumb, because they don't properly understand how limited the tool is, but anybody who thinks it has no value also doesn't understand (or plain doesn't have use for) the tool.

For this usecase there's not even an ethical question about the training data - it's github. It's public repos on github that were already open source and available and stackoverflow posts you would just google for beforehand. Programming was always, to some degree, just googling to see if somebody already made the thingymabob you need right this second in your own project and did not provide or expect attribution for it.

I'm all for shitting on shitty "ai", and I'm all for shitting on tech bros who think they can fire professionals and replace them with it wholesale, but there's a good chunk of people who have actual usecases for it and it's okay if you don't. It's also fine and good to be mad at the shitty things "ai" companies do, but pretending it has no value whatsoever and everyone is being stupid is useless. It's not going to magically disappear - the good tools will become either better or stay good enough for some cases, irrelevant ones will go bust, there will be more and more open source ones with less overhead so you can just run them on your machine without a dyson sphere around the sun.

[aiwikiweb]

Elevate math practice with AI Math Coach! Generate personalized worksheets, mirror classroom exercises, and track progress to help students master math effortlessly.

[lensdeer]

I've been begrudgingly using ChatGPT for work for about a week now. My boss (to whom I have repeatedly explained how LLMs work and why they don't fit our usecase too well, but I digress) is getting inceasingly annoyed I don't use it and write my own shit instead. Figured I could get rid of some heat by using it for ONE project, like, I can say I "tried" it and can go back to actually doing things properly after this is done.

So I decided to use it as "spicy regex": I gave it multiple examples of how to convert structured plaintext data from a public website into a JSON; nothing I couldn't do with a sufficiently long regex string. This is legitimately the perfect use case for a LLM! Yet it's fucking failing at least once for EVERY SINGLE INPUT!

Repeatedly forgetting instructions, hallucinating random bullshit, giving answers that have nothing to do with my request, giving me incomplete outputs and ignoring direct instructions to complete them just to character-for-character repeat the same mistake it just did...

The moral panic and misinformed environmentalist fearmongering have obviously been useless at taking this thing off the pedestal tech oligarchs paid to get it put on, but, like,,, It's honestly MISERABLE to use??? You don't need a moral panic, like, just make someone use this thing for something they're good at and watch them slowly realize how fucking awful it is. How the fuck do AI bros cope.

At this point I've wasted enough time repeating instructions on how to do shit correctly that it would've been MUCH faster to write a script to grab all the information and format it without any user input. But nope. Gotta finish the project with this useless fucking toy or eventually get fired for "not catching up to new technologies". Lovely.

[aitoolswhitehattoolbox]

Python Node Developer Lead

Responsibilities 1. Good Experience on Python scripting. 2. Experience on any web framework like Tornado/fastapi. 3. Experience implementing AI/ML and Generative AI usecases, RAG implementations. 4. Experience working on Embedding a… Apply Now

[ai-news]

#AI #ML #Automation

[razorsign]

Empower your sales team with RazorSign’s AI-powered CLM solution. Streamline processes, reduce errors, and take your sales strategy to the next level.

𝗘𝘅𝗽𝗹𝗼𝗿𝗲 𝗠𝗼𝗿𝗲! https://www.razorsign.com/useCase/sales

[chappydev]

I don't really think they're like, as useful as people say, but there are genuine usecases I feel -- just not for the massive, public facing, plagiarism machine garbage fire ones. I don't work in enterprise, I work in game dev, so this goes off of what I have been told, but -- take a company like Oracle, for instance. Massive databases, massive codebases. People I know who work there have told me that their internally trained LLM is really good at parsing plain language questions about, say, where a function is, where a bit oif data is, etc., and outputing a legible answer. Yes, search machines can do this too, but if you've worked on massive datasets -- well, conventional search methods tend to perform rather poorly.

From people I know at Microsoft, there's an internal-use version of co-pilot weighted to favor internal MS answers that still will hallucinate, but it is also really good at explaining and parsing out code that has even the slightest of documentation, and can be good at reimplementing functions, or knowing where to call them, etc. I don't necessarily think this use of LLMs is great, but it *allegedly* works and I'm inclined to trust programmers on this subject (who are largely AI critical, at least wrt chatGPT and Midjourney etc), over "tech bros" who haven't programmed in years and are just execs.

I will say one thing that is consistent, and that I have actually witnessed myself; most working on enterprise code seem to indicate that LLMs are really good at writing boilerplate code (which isn't hard per se, bu t extremely tedious), and also really good at writing SQL queries. Which, that last one is fair. No one wants to write SQL queries.

To be clear, this isn't a defense of the "genAI" fad by any means. chatGPT is unreliable at best, and straight up making shit up at worst. Midjourney is stealing art and producing nonsense. Voice labs are undermining the rights of voice actors. But, as a programmer at least, I find the idea of how LLMs work to be quite interesting. They really are very advanced versions of old text parsers like you'd see in old games like ZORK, but instead of being tied to a prewritten lexicon, they can actually "understand" concepts.

I use "understand" in heavy quotes, but rather than being hardcoded to relate words to commands, they can connect input written in plain english (or other languages, but I'm sure it might struggle with some sufficiently different from english given that CompSci, even tech produced out of the west, is very english-centric) to concepts within a dataset and then tell you about the concepts it found in a way that's easy to parse and understand. The reason LLMs got hijacked by like, chatbots and such, is because the answers are so human-readable that, if you squint and turn your head, it almost looks like a human is talking to you.

I think that is conceptually rather interesting tech! Ofc, non LLM Machine Learning algos are also super useful and interesting - which is why I fight back against the use of the term AI. genAI is a little bit more accurate, but I like calling things what they are. AI is such an umbrella that includes things like machine learning algos that have existed for decades, and while I don't think MOST people are against those, I see people who see like, a machine learning tool from before the LLM craze (or someone using a different machine learning tool) and getting pushback as if they are doing genAI. To be clear, thats the fault of the marketing around LLMs and the tech bros pushing them, not the general public -- they were poorly educated, but on purpose by said PR lies.

Now, LLMs I think are way more limited in scope than tech CEOs want you to believe. They aren't the future of public internet searches (just look at google), or art creation, or serious research by any means. But, they're pretty good at searching large datasets (as long as there's no contradictory info), writing boilerplate functions, and SQL queries.

Honestly, if all they did was SQL queries, that'd be enough for me to be interested fuck that shit. (a little hyperbolic/sarcastic on that last part to be clear).

ur future nurse is using chapgpt to glide thru school u better take care of urself

[aiwikiweb]

Enhance your music with AI Mastering! Automatically improve audio quality, balance loudness, and achieve professional-level mastering with ease. Take your tracks to the next level