There’s a massive difference between AI being used to help the user, and AI being used as a method to spy on users, collect data, monetize from, and weaponize.
I’m happy with using local AI tools, if needed. For example, using local AI contextual search on my self-hosted IMMICH photos is awesome.
But I absolutely do not need or want AI features that have to connect somewhere. Because that just means I’m being data harvested and profiled for someone else to profit from.
right now AI is mostly used for spying, and stealing data, thats why all the tech bros are pushing it. For spying in general, something like thiels palintir is doing for evil purposes, and probably musks AI too.
Agreed. I’ve also been very impressed with Perplexica (linked to a self-hosted LLM on Ollama). It ties into SearXNG and will perform web searches, dive into the results, and summarize what it finds. Not just the pages themselves, but the specific information on those pages that addresses your original questions, including references which link back to the pages that were used to generate the summary. It’s easy to identify hallucinations when it links to the specific page where it got the information from (though I have yet to experience any hallunications with Perplexica yet).
I think Lemmy’s userbase is a bit predisposed to that. Unfortunately, that sentiment isn’t common enough, and while most people don’t want to be monetized if asked, with the convenience the reaction is a collective shrug.
But another thing we are predisposed to is dev bugs, and I think the average person won’t like how unreliable many such features are.
There’s a massive difference between AI being used to help the user, and AI being used as a method to spy on users, collect data, monetize from, and weaponize.
I’m happy with using local AI tools, if needed. For example, using local AI contextual search on my self-hosted IMMICH photos is awesome.
But I absolutely do not need or want AI features that have to connect somewhere. Because that just means I’m being data harvested and profiled for someone else to profit from.
right now AI is mostly used for spying, and stealing data, thats why all the tech bros are pushing it. For spying in general, something like thiels palintir is doing for evil purposes, and probably musks AI too.
Agreed. I’ve also been very impressed with Perplexica (linked to a self-hosted LLM on Ollama). It ties into SearXNG and will perform web searches, dive into the results, and summarize what it finds. Not just the pages themselves, but the specific information on those pages that addresses your original questions, including references which link back to the pages that were used to generate the summary. It’s easy to identify hallucinations when it links to the specific page where it got the information from (though I have yet to experience any hallunications with Perplexica yet).
I think Lemmy’s userbase is a bit predisposed to that. Unfortunately, that sentiment isn’t common enough, and while most people don’t want to be monetized if asked, with the convenience the reaction is a collective shrug.
But another thing we are predisposed to is dev bugs, and I think the average person won’t like how unreliable many such features are.
Most people do want to be monetised as long as they don’t have to pay money for anything.