this post was submitted on 10 Apr 2025
191 points (88.4% liked)
Memes
49605 readers
1647 users here now
Rules:
- Be civil and nice.
- Try not to excessively repost, as a rule of thumb, wait at least 2 months to do it if you have to.
founded 6 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
LLMs seem to be great for translations and... transcriptions?
I know Google sucks, but I will give a little credit to the newer Pixels' "Circle to Search" feature. It's nice to pull up an image and just circle search it and see where it came from, or the context behind it, etc etc.
Circle Search and the two you mention are probably the only actually useful things to come out of AI so far.
Summary stuff works really well too - I like to be able to bookmark a page and have tags/brief summary auto-gen'd
It's also useful for finding stuff in large documents or codebases. For example, I was recently trying to understand how different concepts in a paper related to each other, and LLM was able to find the relevant parts of the paper which helped me piece things together.
Most of the time these abstracts are fine. But I'm wary of it because of the cases where it doesn't. :)
yeah, but for my personal bookmarks a hallucination is not a big deal, not exactly mission-critical lol
They really are! We should use tools for what they are good at and not try to cram AI into everything. Something something hammer, nail, right?
Yes. It's normal human behaviour. There is a hype, then a bubble, then... normalization.
Multimedia, keychain, blue LEDs, curved LCD screens, LLMs,... One of them is way less useful than the others.; P