Technology
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- 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, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
I wish they had broke it out by AI. The article states:
"Gemini performed worst with significant issues in 76% of responses, more than double the other assistants, largely due to its poor sourcing performance."
But I don't see that anywhere in the linked PDF of the "full results".
This sort of study should also be re-done from time to time to track AI version numbers.
It doesn't really matter, "AI" is being asked to do a task it was never meant to do. It isn't good at it, and it will never be good at it.
Using an LLM to return accurate information is like using a shoe to hammer a nail.
Except that a shoe is vaguely hammer ish. More like pounding a screw in with your forehead.
We've all done it?
Nope, my soles are too soft.
Wow, way to completely ignore the content of the comment you're replying to. Clearly, some are better than others... so, how do the others perform? It's worth knowing before we make assertions.
The excerpt they quoted said:
"Gemini performed worst with significant issues in 76% of responses, more than double the other assistants, largely due to its poor sourcing performance."
So that implies that "the other assistants" performed more than twice as well, so presumably that means encountering serious issues less than 38% of the time (still not great, but better). But they said "more than double the other assistants", does that mean double the rate of one of the others or double the average of the others? If it's an average it would mean that some models probably performed better, while others performed worse.
This was the point, what was reported was insufficient information.
Yes, you are a techbro. You suck because your ideas doesn't take into consideration actual real life. Fuck you.
Wow, that's just incredibly dismissive and rude. And in response to a completely reasonable comment!
Look, forget the whole AI discussion, I don't care. Here's the thing, I really like Lemmy. I really like this community and I want to continue using it as a way to have discussions with people about interesting topics. What I don't want to see is people yelling insults and swearing at any user they disagree with.
Frankly, that behavior is unwelcome. That's reddit behavior, you can go there if that's what you want to do.
And also which version of the models. Gemini 2.5 Flash is a completely different experience to 2.5 Pro.
There’s a few replies talking about humans misrepresenting the news. This is true, but part of the problem here is that most people understand the concept of bias - even if only to the extent of “my people neutral, your people biased”. But this is less true for LLMs. There’s research which shows that because LLMs present information authoritatively that not only do people tend to trust them, but they’re actually less likely to check the sources that the LLM provides than they would be with other forms of being presented with information.
And it’s not just news. I’ve seen people seriously argue that fringe pseudo-science is correct because they fed a very leading prompt into a chatbot and got exactly the answer they were looking for.
I hear a lot of people say "let's ask chatGPT" like the AI is god and know everthing 🙏, that's a big problem to be honest
I wonder if people trust ChatGPT more or less than an international celebrity who is also their best friend.
I dont understand the use people make of AI. I know a lot of of professionnal composer who are like "That's awesome, AI does the music for me now!" and I'm like, cool, now you only have the boring part of the job to do since the fun part was made by AI. Creating the music is litteraly the only fun part, I hate everything around it.
It's a word predictor. It is good at simple text processing. Think local code refactoring, changing the style or structure of a small text piece, or summarizing small text pieces into even smaller text pieces. It is ok at synthesizing new text that has similar structure to the training corpus. Think generating repetitive boilerplate or copywriting. It is very bad at recalling or checking facts, logic, mathematics, and everything else that people seem to be using it for nowadays.
The AI creating music is not an LLM
Ah, sorry, missed the context
Makes sense. I have used AI for software development tasks such as manipulating SQL queries and XML files (tedious things) and am always disappointed with how AI will misinterpret some things. But it’s obvious with those when the requests fail. But for things like “the news” where there is no QA team to point out the defect, it will be much harder to notice. And when AI starts (or continues) to use AI generated posts as sources, it will get much worse.
Replace CEOs by AI
I've had someone else's AI summarize some content I created elsewhere, and it got it incredibly wrong to the point of changing the entire meaning of my original content.
Will they change their disclaimer now, from "can be wrong" to "is often wrong"? /s
Parrot is wrong almost half of the time. Who knew?
Do you realize what you just said????!!!
Wow! They have reached parrot intelligence!
Next they might teach it to butterfly! You know, like you're off the ground and going somewhere in open air, but they just keep building shit right where you're flying.... And lamps!
From there, who knows?!
Precision, nuance, and up to the moment contextual understanding are all missing from the “intelligence.”
Like the average American with an 8th grade reading comprehension.
Which is what they used for the training data.
So it's about on par with humans, then.
And then I wonder how frequently humans misinterpret the mistranslated news.
Humans do it often, but they don't have billions of dollars funding their responses.
Worse: One third of adult actually believe the shit the AI produces.
wrinkle: AI used for this study
"misrepresent" is a vague term. Actual graph from the study
The main issue is usual.. sources. AI is bad at sources without a proper pipeline. They note that Gemini is the worst at 72%.
Note, they're not testing models with their own pipeline. They're testing other people's products. This is more indicative of the product design than the actual models
This graph clearly shows that AI is also shockingly bad at factual accuracy and at telling a news story in such a way that someone who didn't already know about it to understand the issues and context. I think you're misrepresenting this graph as being bad about sources, but here's a better summary of the point you seem to be making:
AI's summaries don't match their source data.
So actually, the headline is pretty accurate in calling it misrepresentation.
Yet the LLM seems to be what everyone is pushing, because it will supposedly get better. Haven't we reached the limits of this model and shouldn't other types of engines be tried?
shouldn’t other types of engines be tried?
Sure, but the tricky bit is to be more specific than that.
Well, you know...
"Waves vaguely"
The info-sphere today is already a highly delusional place and news can be often contradictory, even from day to day especially by outlets like BCC who is more focused on setting global narratives, not being a reporter of facts as best understood at the moment. No wonder AI would be confused, most readers are confused when navigating every statement made by experts or anonymous officials on every subject. Seems like this study really measured an AI models ability to vomit out the same text in different words and avoiding using any outside context be it accurate or hallucination.
BCC
Could be better, but still a huge step up from the hate rhetoric magats get spoon fed 24/7 from Fox and friends.
buT AI iS hERe tO StAY
So less of a percentage than the readers and mass media