[mvieira38]

Yes, that's what it is. Kagi as a brand is LLM-optimist, so you may be fundamentally at odds with them here... If it lessens the issue for you, the sources of each item are cited properly in every example I tried, so maybe you could treat it as a fancy link aggregator

[freediver]

> Kagi as a brand is LLM-optimist

Kagi founder here. I am personally not an LLM-optimist. The thing is that I do not think LLMs will bring us to "Star Trek" level of useful computers (which I see humans eventually getting to) due to LLM's fundamentally broken auto-regressive nature. A different approach will be needed. Slight nuance but an important one.

Kagi as a brand is building tools in service of its users, no particular affinity towards any technologies.

[pseudalopex]

You claimed reading LLM summaries will provide complete understanding. Optimistic would be a charitable description of this claim. And optimism is not limited to the most optimistic.

[sally_glance]

Another LLM-pragmatist here. I don't see why we should treat LLMs differently than any other tool in the box. Except maybe that it's currently the newest and most shiny, albeit still a bit clunky and overpriced.

[freedomben]

Fwiw, I love your approach to AI. It's been very useful to me. Quick answers especially has been amazingly accurate and I've used it hundreds of times, if not thousands, and routinely check the links it gives

[agiacalone]

Happy Kagi Ultimate user here, so thank you!

[NobodyNada]

I'm about as AI-pessimist as it gets, but Kagi's use of LLMs is the most tasteful and practical I've seen. It's always completely opt-in (e.g. "append a ? to your search query if you want an AI summary", as opposed to Google's "append a swear word to your search query if you don't want one"), it's not pushy, and it's focused on summarizing and aggregating content rather than trying to make it up.

[arrosenberg]

FYI, you can append &udm=14 to Google searches to remove AI results and a bunch of the other clutter they've added.

[b112]

I did that, and started getting flagged as a bot. Had to search elsewhere(Kagi) full time, or else suffer endless "find a bike" nonsense.

I think Google hates the loss of no/few ads or lame suggestions.

[pirates]

Google thinks the same of me and I don't even edit the URL. I can have a session working just fine one night and come back the next day, open a new tab to search for something, and get captcha'd to hell. I'm fairly sure they just mess with Firefox on purpose. I won't install Brave, Chrome, or Edge out of principle either. Safari works fine, but I don't like it.

[Jepacor]

Google will captcha me on the second or third search if I try to use the "site":" advanced keyword to narrow down search

I'm sorry I know how to use your tool?? ? Didn't you put these keywords in to be used?

[freedomben]

Google has gotten amazingly hostile toward power users. I don't even try to use it anymore. It almost feels like they actively hate people that learned how to use their tools

[paws]

Neat trick, any other params folks might want to know about?

[willguest]

I found this page that describes a variety of search parameter: https://susodigital.com/thoughts/the-mystery-of-the-google-u...

then i got the machine to write a front-end that visualises them and builds a search query for you: https://pastebin.com/HNwytYr9

enjoy

[meowface]

I consider myself a major LLM optimist in many ways, but if I'm receiving a once per day curated news aggregation feed I feel I'd want a human eye. I guess an LLM in theory might have less of the biases found in humans, but you're trading one kind of bias for another.

[rwl]

Indeed! A once per day human-curated news aggregation feed used to be called a "newspaper". You can still get them in some places, I believe.

[flexagoon]

This isn't really comparable. A newspaper is a single source. New York Times is a newspaper, CNN (a part of it) is a newspaper. Services like Kagi News, whether AI or human-curated, try to do aggregation and meta-analysis of many newspaper.

[yencabulator]

Newspapers routinely report what other newspapers said. The original(s) is the one "breaking" the story, the others are "covering" the story.

[mvieira38]

Yeah, I agree. The entire value/fact dichotomy that the announcement bases itself on is a pretty hot philosophical topic I lean against Kagi on. It's just impossible to summarize any text without imparting some sort of value judgement on it, therefore "biasing" the text

[xpe]

> It's just impossible to summarize any text without imparting some sort of value judgement on it, therefore "biasing" the text

Unfortunately, the above is nearly a cliché at this point. The phrase "value judgment" is insufficient because it occludes some important differences. To name just two that matter; there is a key difference between (1) a moral value judgment; (2) selection & summarization (often intended to improve information density for the intended audience).

For instance, imagine two non-partisan medical newsletters. Even if they have the same moral values (e.g. rooted in the Hippocratic Oath), they might have different assessments of what is more relevant for their audience. One could say both are "biased", but does doing so impart any functional information? I would rather say something like "Newsletter A is compromised of Editorial Board X with such-and-such a track record and is known for careful, long-form articles" or "Newsletter B is a one-person operation known for a prolific stream of hourly coverage." In this example, saying the newsletters differ in framing and intended audience is useful, but calling each "biased in different ways" is a throwaway comment (having low informational content in the Shannonian sense).

Personally, instead of saying "biased" I tend to ask questions like: (a) Who is their intended audience; (b) What attributes and qualities consistently shine through?; (c) How do they make money? (d) Is the publication/source transparent about their approach? (e) What is their track record about accuracy, separating commentary from factual claims, professional integrity, disclosure of conflicts of interest, level of intellectual honesty, epistemic standards, and corrections?

[xpe]

> The entire value/fact dichotomy that the announcement bases itself on

Hmmm. Here I will quote some representative sections from the announcement [1]:

>> News is broken. We all know it, but we’ve somehow accepted it as inevitable. The endless notifications. The clickbait headlines designed to trigger rather than inform, driven by relentless ad monetization. The exhausting cycle of checking multiple apps throughout the day, only to feel more anxious and less informed than when we started. This isn’t what news was supposed to be. We can do better, and create what news should have been all along: pure, essential information that respects your intelligence and time.

>> .. Kagi News operates on a simple principle: understanding the world requires hearing from the world. Every day, our system reads thousands of community curated RSS feeds from publications across different viewpoints and perspectives. We then distill this massive information into one comprehensive daily briefing, while clearly citing sources.

>> .. We strive for diversity and transparency of resources and welcome your contributions to widen perspectives. This multi-source approach helps reveal the full picture beyond any single viewpoint.

>> .. If you’re tired of news that makes you feel worse about the world while teaching you less about it, we invite you to try a different approach with Kagi News, so download it today ...

I don't see any evidence from these selections (nor the announcement as a whole) that their approach states, assumes, or requires a value/fact dichotomy. Additionally, I read various example articles to look for evidence that their information architecture group information along such a dichotomy.

Lastly, to be transparent, I'll state a claim that I find to be true: for many/most statements, it isn't that difficult nor contentious to separate out factual claims from value claims. We don't need to debate the exact percentages or get into the weeds on this unless you think it will be useful.

I will grant this -- which is a different point that what the commenter above made -- when reading various articles from a particular source, it can take effort and analysis to suss out the source's level of intellectual honesty, ulterior motives, and other questions I mention in my sibling comment.

[1]: https://blog.kagi.com/kagi-news

[b112]

Don't worry, all those news articles are of course human curated.

(I say this sarcastically and unhappily)

[leephillips]

Hard pass then. I’m a happy Kagi search subscriber, but I certainly don’t want more AI slop in my life.

I use RSS with newsboat and I get mainstream news by visiting individual sites (nytimes.com, etc.) and using the Newshound aggregator. Also, of course, HN with https://hn-ai.org/

[CaptainOfCoit]

> Also, of course, HN with https://hn-ai.org/

Ironically, this submission is at the top of that website :)

[cyberax]

You can also convert regular newspapers into RSS feeds! NYTimes and Seattle Times have official RSS feeds, and with some scripting you can also get their article contents.