1y ago

Discussion on Concerns over Auto tl;dr bot

Hello everyone ,

There have been concerns raised lately over issues with the Auto tl;dr bot which creates summaries of news articles from several known sites, however only really ABC news is applicable here. Relevant threads:

Truck stowaway 'lucky to be alive' after travelling nearly 400km on metal racks beneath B-double (no problem with summary - just discussion of concerns)
Small modular nuclear reactor that was hailed by Coalition as future cancelled due to rising costs (Australian Politics)

There are also many other occurrences (I haven't been keeping track), if there are some you would like appended to this list comment with a link below.

Most concerns are that the bot misses important information and/or gives a misleading summary. I'd like to see where people sit on the issue and how we could potentially deal with it. There are a few options I can think of:

Remove the bot (through a ban)
Get @dalekerrigan@aussie.zone to comment a disclaimer underneath all of its comments
Get @dalekerrigan@aussie.zone automatically delete all comments by the bot which have been reported (may open door for abuse)
Do nothing

I don't hate the bot - it can be useful, and I like the concept, however, just like us it gets things wrong.

Anyway feedback is welcome, if you have an opinion on this please comment below so I can judge where we all stand on this and try to make the right decision

15 comments

If it can't put an accurate summary, and it can't put the text verbatim, I'd rather it just buggered off I guess. Not much point reading half the story and missing context and critical paragraphs.
- Agreed.. full article. We can skim read and choose what time we save.

Ban it.
It's basically a misinformation machine even if it doesn't have an agenda.
Disclaimer is pointless.

I’ve stopped reading them, found the summary to heavily change a story as context was completely removed, especially when political

It has about a 60% usefulness ratio in my opinion but I'd suggest option 2 an auto comment disclaimer that it often leaves relevant stuff out AND to downvote it when the summary isnt useful.
The latter because a) it's a signal to later readers that the summary is misleading and b) if the maintainer is monitoring (prob not) that's a clue as to which summaries need to be looked at
- Yeh, agree with approach, seems sensible
- This is equivalent to option 4 IMO.
  No one will heed the disclaimer. I mean the disclaimer basically means the whole thing is pointless - you need to read the article I'd you want to be sure you're not missing something very important.
  Down votes are meaningless. No one will look at the down votes and conclude that in that specific case the summary is unreliable.

ban. A culture of reading the source must be encouraged

If the bot is taking clicks away from the ABC; that's not a good thing.

I like it.
The summaries it provides fall into one of the categories for me:
Ignore because I've already read the article
The summary is interesting enough that I'll go read the article
Ignore, wouldn't have read the article anyway
Overall it's more articles read. Maybe a disclaimer or have it respond to votes or comments of good bot/bad not
¯⁠\⁠_⁠(⁠ツ⁠)⁠_⁠/⁠¯

I like it more than hate it and if it's wrong I just close its comment. Tbh if there's no pay wall or some kind of soft wall it should be encouraged for people to read the article rather than assume a bot or OP has represented the news issue in an unbiased way.

I'm more concerned the community doesn't have a YT Piped redirect bot.

Keep it and let people decide by themswlves if they wanna read it or the source article. Just add a autocomment disclaimer. Hell its opensource just make ur own ibstance that has the disclaimer. As long as it doesnt have an agenda its errors will balance eachother out on average.

I like the bot. You just need a common sense approach, and realise it's trimmed a lot of content (which it tells you), and if it feels off, just click the link if the topic interests you.

I've whinged about it.
It's almost random selection of what to include or exclude can heavily slant the summary.
Just like humans. Just like AI. Just like any bot. Maybe I'm a bot. Maybe you're a bot.
We better get used to it, and be able to critically analyse a post rather than concern ourselves with 'who' wrote it.

15 comments