Ask HN: How have you applied AI in your life, work or hobbies?
I'm looking for concrete use-cases where people are applying AI models and tools in real-life scenarios.
For example, at work we found a benefit to use an AI-based image removal tool for poor-quality product photos sent in by our customers.
There's a lot of hype out there, and lots of demos. I'd love to know where you're really getting use out of it!
[+] [-] MH15|3 years ago|reply
There's a quote, apparently by Donald Rumsfeld (TIL) that reads “There are known knowns, things we know that we know; and there are known unknowns, things that we know we don't know. But there are also unknown unknowns, things we do not know we don't know.” I find the known knowns and known unknowns are well served by traditional web search, but ChatGPT and similar tech can help me explore the unknown unknowns.
[+] [-] cableshaft|3 years ago|reply
You have to evaluate it to see if it fits your goals, and work out the details and prototype and playtest it, but sometimes just seeing those ideas can get you unstuck or enhance the design or spark another idea which works even better. And some of those ideas I know I would never think of, and probably never get them from a playtester, so I'm thankful for them.
Also used it recently when my wife was feeling stuck in the motivations for certain people in her book, I gave it a paragraph overview and asked it to come up with motivations and it provided quite a few, including about six or seven we had previously tossed around and discussed. She was skeptical beforehand but was more impressed afterwards, in particular because it provided a bit of validation that the ideas we were tossing around weren't bad ideas.
Just this morning I asked it for what tools within Azure cloud tech to use for accomplishing a programming task, and the information it provided me seemed a bit more useful than me asking other experienced users in our organization as well as a few videos and articles I read (because it was able to provide an idea more directly applicable to our specific use case). Even provided some sample code.
[+] [-] opyate|3 years ago|reply
However, as I'm not a rich indie dev, I have to do client work too. Currently working in insuretech, where I'm focused on document extraction (layout detection, OCR, labelling + retraining models, etc).
0. https://juanuys.com/blog/2023/03/07/thoughts-on-ai-and-games...
[+] [-] drakonka|3 years ago|reply
I first got some ideas for how this could be done by straight up asking for different methods character A could use to get out of B's grip. I then got it to write some ways that B might explain this, or instruct A. It went so far as to have B offer encouragement at strategic parts and such.
The final version ended up being heavily edited of course, but it helped greatly to visualize the scene and the physical actions.
I also used it to help me brainstorm my way out of some plot knots I got myself into. Like "Imagine such-and-such situation. What would be some possible reasons why so-and-so would choose NOT to perform this particular action?"
[+] [-] hitori|3 years ago|reply
[+] [-] Awelton|3 years ago|reply
[+] [-] nmfisher|3 years ago|reply
The sentences still need human vetting as it will occasionally get things wrong, but it’s a massive productivity multiplier. I’m mostly happy that it’s available via API so I can stop endlessly copying and pasting.
If you consider speech recognition/synthesis “AI”, then I try heavily on that too. I did. dabble with using Stable Diffusion to generate images to assist learning. That shows promise but it’s not a priority at the moment so I left it.
[0] https://polyvox.app
[+] [-] dstala|3 years ago|reply
GitHub CoPilot : I come from embedded engineering background, but now working in web tech space. CoPilot helps as a very good assistant when I am working with TypeScript. Essentially lesser number of lookups now to Google & StackOverflow
Content creation : I take help of ChatGPT to give me some pointers when I am about to write on a topic / blog post. For tech articles, it's a good help. Helps correct grammar, change content to professional tone.
[+] [-] imhoguy|3 years ago|reply
For now it is just "Explain it nicer:" plus seed with some topic context, and then "In German", "In French"...
I think soon I can probably do an entire customer support work over email in any foreign language which I may not even know much but with an aid of LLM these bariers disappear easily.
[+] [-] netsharc|3 years ago|reply
Edit: come to think of it, actually not (because the operator in the Chinese room has no understanding), but is helping you ChatGPT understand the questions (I presume?).
[+] [-] danielmakestech|3 years ago|reply
[+] [-] devstein|3 years ago|reply
It's been a ton of fun. I'm basically using AI automation to fight AI automation :)
You can follow along at: https://sharedrecruiting.co/
[+] [-] alvivanco|3 years ago|reply
[+] [-] ezedv|3 years ago|reply
[+] [-] ActorNightly|3 years ago|reply
I have found quite a few errors in generate response though, but 90% of the time it gives me what I need.
Its also pretty good for generating boilerplate code.
[+] [-] hodovaniuk_m|3 years ago|reply
[+] [-] replwoacause|3 years ago|reply
[+] [-] aussiegreenie|3 years ago|reply
If I employed a real typist in a cheaper country, it would be faster but ChatpGPT is avaiable when I wnt to work.
[+] [-] henly|3 years ago|reply