How do you deal with AI summaries distorting news sources by hallucinating quotes or having factual errors about articles?
Like Apple AI headlines: https://www.bbc.com/news/articles/cq5ggew08eyo
this post was submitted on 11 Mar 2025
-5 points (35.3% liked)
Community Promo
1936 readers
111 users here now
Promote your favourite communities here, or ask about a community you are looking for.
This includes groups for Lemmy, Kbin, Telegram, Matrix, Discord, Reddit, etc.
๐ฃ๏ธ Sharing a community:
Please make a descriptive title describing the purpose of the community, and link the community in the post URL or body fields.
When linking to your community, use this universal format: !communitypromo@lemmy.ca
๐ Searching for a community:
Include details about what you are looking for
Rules:
- Reposts: Don't post about the same community more than once a month. However, you can share new details as needed (ex. upcoming event, updated content)
- Instance Rules: Follow instance rules when posting and commenting. See sidebar here
Rule breaking posts will be removed at moderator discretion (ex. scams, dangerous links, duplicate posts)
This community is also on Matrix and Discord, but those spaces are not managed by the moderation team here.
founded 2 years ago
MODERATORS
I use a finetuned T5 summarisation model that is relatively accurate. It has some minor issues with occasional miss assigning quotes but it doesn't hallucinate like a traditional GPT style model does. It is 60% identical to that of a human summary and >95% accurate in terms of meaning. It is more accurate than traditional nonai based summarisstion tools (I'm not sure how it compares to a human) but I belive it is as accurate and nonbias as possible.
Its biggest flaw is actually the traditional nonai web scraper which sometimes pulls the wrong content. Its all foss so if u wanna go make a pull to improve it that would be greatly appreciated.
EDIT: I've been experimenting with having a tradition GPT LLM look over the summary and original to catch these errors but have had little to no success without using large models which I cannot run on my local hardware (I unfortunately can't afford to pay for inference at the scale my bot runs).
Thanks for the explanation. I think if you combined that with a method to retract or edit summaries based on human reports, you can probably fill in the remaining 5%. I am unsure how feasible that would be though. Good luck with the community!
Yeah I'm not sure how that can be achieved in a way where I single report can catch errors without letting every single user mess with it. I could perhaps expose the section breakdown to users and allow users to regenerate specific sections but that would require a lot more complex interaction. But thanks for the suggestion tho I'll look into it.