1080
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 23 May 2024
1080 points (98.2% liked)
Technology
59081 readers
2864 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
I’m confused a little by the LLM’s and datasets here.
OpenAI and Reddit have their partnership for training, so I would have assumed the pizza glue answer would have come from an OpenAI result, but Google doesn’t use OpenAI. How did Google give an answer that was clearly scraped from Reddit?
Google has a deal with reddit as well. https://www.reuters.com/technology/reddit-ai-content-licensing-deal-with-google-sources-say-2024-02-22/?utm_source=reddit.com
But I don't think it's just an issue with the dataset. It's the false promise of these LLMs having a fucking clue what a good search result is and what is not. They don't. They are just good at creating text that sounds plausible. That's not what searching for factually correct information is about though.
Thank you for the link. I didn’t realise that Google had a deal with Reddit as well, which explains why it was clearly indexed from Reddit.
I agree that AI doesn’t have a clue what an accurate response is. It’s just not sentient enough to differentiate between shitposting and fact. I also totally agree that an answer given from a search result HAS to be accurate, and we’re heading down a path of a misinformation super highway if LLMs are trained on incorrect data.
Yep. LLMs are great for bouncing ideas off, and for getting "soft answers", but no-one should ever be looking for factual answers from them.