Hey fellow Lemmings,
I run AI News Summary Bot, a project that brings you News summary! The bot is once again live on our community at !news_summary@hilariouschaos.com
This is just an update as its moved instance to hilariouschaos.com
The bot is still in its early stages, and I’m excited to hear your feedback and suggestions on how to improve it. Feel free to share your thoughts and ideas. I’ve got some free time for development coming up so along with a backlog of features I would appreciate some suggestions in terms of where you would like to see that effort focused.
FAQ: Q: How do I avoid hallucinations? A: I don’t use a traditional GPT model its operating on text sections doing a semantic summarisation.
Q: What’s its accuracy? A: It has a >95% semantic accuracy.
Q: Why did it summarise 2 articles at once A: That’s a known issue with Euronews as a source in some cases I’ve tried to patch it in the past but its an intermittent issue so hard to debug.
Repository: If you’re interested in contributing or exploring the code behind the bot, you can find the repository at https://github.com/muntedcrocodile/ai_news_bot
Donations: If you’re interested in donating to allow me to spend more time developing please do: monero:8916FjDhEqXJqX9Koec9WaZ4QBQAa6sgW6XhQhXSjYWpQiWB42GsggEh73YAFGF86GU2gEE1TTRdWSspuMgpWGkiPHkgBTX
Stay informed, and let’s build this community together!
This continues to be a terrible idea. What you’ve built is an automated misinformation machine. You absolutely need to be fact checking every summary by hand, anything else is completely irresponsible no matter how good you claim your model is.
The purpose of the lemmy community is the fact checking. If you wanna practice what you preach you can go write corrections for the summaries as comments its useful data for the next processing step.
So you want it to be everyone else’s job to fact check your bot for you? If you can’t be bothered to check your own bot, why should we take it seriously either?
No, you need to fact check information before spreading it. Once you’ve posted it it’s too late, people will have read it and some of those people may believe it and spread it further. You need to either read every summary yourself (or hire a team to do it) and take responsibility for any mistakes, or abandon the project. You don’t get to blame other people for not fact checking the junk you post online.
Go find a summary that is incorrect and not blatantly obvious that it is an error.
U gonna pay me to fact check? The Monero link is right there.
BTW attitude like this makes me want to spend my time writing an ai bot to generate mass propaganda radicalising the masses on millions of puppet accounts that are indistinguishable from real people. But instead I make a tool that clearly states that it is ai and is completely foss and transparent. I like playing with ai and lemmy I write code using those 2 things for fun as a hobby hope I use that power for good. If my bot goes dark u will know why. You have been warned.
Wow that does not make your intentions seem more pure lol. “Pay me to do something bad or else I’ll do something worse mwahahaha”
I don’t have time to fact check. U said I should pay someone to fact check. I don’t have money to pay fact checkers. If u think its that important u can pay for that.
This comment turns me off of the entire concept.
It’s like announcing you’ve created a machine that can repurpose garbage and reduce waste going to the landfills, but it just spreads random garbage all over the neighborhood with the intent that the rest of us can dig through it and pull out anything that might be useful and discard the rest.
It just creates much more work for everyone else with minimal positive results.
It makes excellent data on current events for my stock trading bot.
Also its more like it goes through a pile of trash pulls out a bunch of useful stuff with 1or2% trash still in it and then sells that at a store. Don’t come to the store. Tell the store what from its collection is trash.
Uh oh. Someone doesn’t understand Lemmy’s hatred of AI!
Except for silly pictures and terrible mistakes. Keep sharing those.
What I don’t understand is people need to hate it publicly and not simply just block it and move on With their life. Some people like it, some find it useful. I don’t see why everyone is so against letting people make that decision for themselves.
What I don’t understand is people need to hate it publicly and not simply just block it
Probably the same motivation as the need to promote it publicly and not alone bye oneself, wouldn’t you agree?
95% semantic accuracy
Hey, look, a 1/20 chance of being lies! And that’s on top of whatever inaccuracies are already baked into the article it’s summarizing! What if those articles already lead with AI summaries? It’s slop all the way down!
Have u gone and read a couple of my summaries and a couple of the originals and compared? I think its pretty good.
Also the 95% is the worse case the average case is significantly better.
When I looked a couple of hours ago the first link’s summary was an enormous block of text with no formatting and the second was about 40 words summarised to 20.
That’s how summarising works. Shorter text has shorter summaries. Obviously if u summarise a postcard its gonna be shorter than summarising the hobbit.
Obviously. Point is it’s either completely unreadable or completely unnecessary. Why are you summarising a postcard?
I’m like the only person here actually giving feedback. Make it readable and don’t waste people’s time summarising two sentences.
So paragraph breaks in large articles? I disagree about the short ones its still shorter text to convey the same information don’t see how that’s a time waste?
I don’t think they do convey the same information.
“They don’t know what the f*** they’re doing.” US President Donald Trump delivered an extraordinary four-letter outburst as he criticised breaches of the Israel-Iran ceasefire. Trump aimed the bulk of his anger at Israel, saying he would try to avert further planned attacks.
vs
Trump delivered an extraordinary four-letter outburst as he criticised breaches of the Israel-Iran ceasefire. Trump aimed the bulk of his anger at Israel, saying he would try to avert further planned attacks.
All the summary has done is remove the quote. That part IS the news that’s being reported on, it matters just as much as the context.
Qatar was the “obvious choice” for Iranian retaliation against the US because it hosts the largest US military base in the Middle East while also acting as a trusted intermediary for Tehran, Al Jazeera’s Dorsa Jabbari explained after the missile attack near Doha.
vs
Qatar was the “obvious choice” for Iranian retaliation against the US. It hosts the largest US military base in the Middle East.
This one has removed the source (which is fair enough if you think that’s not useful information, but I wouldn’t agree), one of the two reasons given, and the context. Someone who only sees this summary will have no idea what it’s about or how to look into it further, which means they’re going to have to read the (one sentence) article anyway.
As for paragraphs, yes. Text that looks like this
is not useful to anybody. I’m not reading this, you haven’t read this. It’s awful.
