Live data from Hacker News

Kagi News

blog.kagi.com

401–410 of 454 posts

Re: Kagi News

#401

Earlier quoted context omitted.

I'm not the person you asked, but it's useful to know if the summary was generated using a method prone to inaccuracy.

That's all methods, though. Have you seen humans?

Someone needs to coin the fallacy that, when anyone criticises LLMs, the speaker retorts with "but how humans are any better?"

I've seen it so many times it definitely needs a name. As an entity of human intelligence, I am offended by these silly thought-terminating arguments.

Re: Kagi News

#402
post #24

> One daily update: We publish once per day around noon UTC, creating a natural endpoint to news consumption. This is a deliberate design choice that turns news from an endless habit into a contained ritual. I might not agree with all decisions Kagi makes, but this is gold. Endless scrolling is a big indicator that you're a consumer not a customer.

> Endless scrolling is a big indicator that you're a consumer not a customer. Someone recently highlighted the shift from social networks to social media in a way I'd never thought about: >> The shift from social networks to social media was subtle, and insidious. Social networks, systems where you talk to your friends, are okay (probably). Social media, where you consume content selected by an algorithm, is not. (im…

CDO = collateralized debt obligation.

Please expand obscure acronyms, not everyone lives in your niche.

Re: Kagi News

#403
post #401

Earlier quoted context omitted.

That's all methods, though. Have you seen humans?

Someone needs to coin the fallacy that, when anyone criticises LLMs, the speaker retorts with "but how humans are any better?" I've seen it so many times it definitely needs a name. As an entity of human intelligence, I am offended by these silly thought-terminating arguments.

It's called "the perfect world fallacy".

Re: Kagi News

#404

I like Kagi and want them to succeed. But currently (according to LinkedIn) theres 26 employees. They are building search, LLM assistant wrappers, a browser and now news. Please don't overextend the same way Proton is currently doing.

…They are also building email

I'm waiting until they announce their own smartphone OS before accepting they are following in the footsteps of Google.

I mean, it keeps bothering me that their search engine logo is a "g". Anything to position themselves as close to google.

Re: Kagi News

#405

Earlier quoted context omitted.

I'm not sure I agree that humans are more accurate at summarizing, but I don't have data, so I'll take your word for it.

I'd point to Wikipedia. You can say the content is "wrong". But the links go to the right place.

In my experience with Claude research, the links ~always go to the right place.

Re: Kagi News

#406

> This is pulling the content of the RSS feeds of several news sites into the context window of an LLM and then asking it to summarize news items into articles and fill in the blanks? This is awful. It's cutting out any money going to the news agencies that go out there and write news. If they didn't exist, Kagi wouldn't work.

Given the buzzword, social media, and advertising infested garbage that ~80% of news sites have turned into...it's hard to sympathize with them.

The content isn't any better, so a news aggregator simply becomes an aggregator of buzzword, advertising infested slop. Garbage in, garbage out.

Re: Kagi News

#407

Looks great for general content. Love that it offers translation. But the Sports section is bad. The game finished 10 hours ago and it's still showing a match preview.

It's generated once a day. Which I love.

Re: Kagi News

#408
post #141

This is honestly very disappointing. Not using LLMs, but the complete lack of transparency about their usage. You can already see in the repository issues related to hallucinations[^1]. This is _fine_, but not if you seem to obscure the fact that these can be very, very wrong. This seems to only be mentioned in the very brief loading screen and at the bottom of the about page[^2]. Also, apparently many of the "core R…

Ive stopped using all kagi stuff because of the lack in transparency and my official requests for transparency remained unanswered. I don't trust any American corporation that refuses to be transparent. US law allows for way to much dangerous stuff compared to the EU. That is also why the US lost its democracy recently.

Re: Kagi News

#409

Earlier quoted context omitted.

Just for concrete confirmation that LLM(s) are being used, there's an open issue on the GitHub repository, on hallucinations with made up information, where a Kagi employee specifically mentions "an LLM hallucination problem": https://github.com/kagisearch/kite-public/issues/97 There's also a line at the bottom of the about page at https://kite.kagi.com/about that says "Summaries may contain errors. Please verify imp…

Man am I tired of this stuff.

[dead]

Re: Kagi News

#410
Is there anythong in the "backend" pre-ai, data scraping etc that is open and decentralized that people could use?

It feels to me like the bigger problem is more about assembling time series of "news" not "news today".

Like if you wanted "show me all stories about crime X from the BBC since 1980" or whatever but then you want to do this across many sources.

This is the missing piece for most new analytics. I think there are legal blockers to getting this done and why I mention decentralization.

Post reply on HN