Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The new Google News: AI meets human intelligence

blog.google

Post reply on HN
Print view View original

51–60 of 66 posts

Re: The new Google News: AI meets human intelligence

#51

Post by gowld »

fareesh wrote: post #12
croon wrote: post #10

Earlier quoted context omitted.

Have any examples of things they've gotten wrong?

I find that they include their interpretation of a statement when it suits a particular point of view, whereas other times they take things very literally, without a measured reading. One example is this case: https://www.snopes.com/fact-check/clinton-compliant-citizenr... There's also the fact that the team behind the website does not have a stellar reputation: https://www.forbes.com/sites/kalevleetaru/2016/12/22/th…

I don't understand what you are trying to show.

Snopes debunked Infowars' absurd mischaracterization of the Ivey email (which was however poorly written in confusing pretentious wording), and produced the evidence in full to justify its position.

Regarding your second link, I read it and don't see what you are driving at. It says that Snopes has some politically engaged staff members, and doesn't do individual bylines, and someone got divorced.

Re: The new Google News: AI meets human intelligence

#52

Post by gowld »

bbitmaster wrote: post #46
croon wrote: post #10

Earlier quoted context omitted.

Have any examples of things they've gotten wrong?

Snopes is very quick to claim things as "False" if it goes against their left leaning lens of the world. As an example arguments used by pro-gun advocates are marked as false. Here's a specific example: They falsify a claim that doctors kill more people than guns https://www.snopes.com/fact-check/doctors-kill-more-people-t... Yet this article linked on hacker news article states that medical error is the third bigges…

IIRC, that "250K people killed by medical error" includes every time a medical caregiver failed to prevent an already sick or injured individual from dying and didn't follow ideal procedures in retrospect. One of the examples was "the doctor didn't tell a patient with heart disease that jogging would be dangerous, and then the guy had a heart attack while jogging".

That's like saying that "hospitals cause most cancer deaths" because people who have cancer are in the care of professionals before they expire.

People who die from guns aren't being euthanized from already impending death.

Re: The new Google News: AI meets human intelligence

#53

Post by gowld »

nickysielicki wrote: post #31
danso wrote: post #23

Earlier quoted context omitted.

I don't understand your objection. The Politifact assessment of "pants on fire" is not for the public official, it's for the conspiracy websites that are peddling the public official's mistake to intentionally mislead people. Read the end of the article ("Our ruling"): > A website pushing conspiracy theories claimed that the March For Our Lives permit was actually prepared months in advance. > We contacted the police…

Well for starters, they haven't ignored the correction. https://fellowshipoftheminds.com/2018/04/05/parklathe-curiou... It's a valid question: large events like this march require considerable coordination among public services and permits are necessary to make sure that police and EMS and public transport can react, and navigating the bureaucratic machine for getting a permit is, by official D.C. guidelines, estimat…

They didn't ignore the correction, they responded to it with a flurry of tangential bullshittery. They also never updated the original article with the correction: https://fellowshipoftheminds.com/2018/03/30/application-for-...

Even in the followup/correction post, they state both:

"In the March 28th email, Officer Earhardt states:

“In reference to your inquiry concerning the March For Our Lives Demonstration, here in the District of Columbia on March 24, 2018. MPD received a permit application several months prior to the actual event, and there was several months of planning for this large event.”"

and then also says: "we have no reason to believe Officer Scott Earhardt." [regarding his correction]

So they are taking two contradicting quotes from one person, and saying one of them is true and the other is false.

What are you trying to argue here?

Re: The new Google News: AI meets human intelligence

#54

Post by ehsankia »

metasaval wrote: post #27
3steve wrote: post #9

I don't want personalized news, I use Google News for the exact opposite reason. I want to see semi-random headlines from a variety of source to increase my exposure to stories I might not have seen otherwise.

1. I haven't used Google news until today, but I believe that's what the "Headlines" tab at the bottom bar is for. It seems to just show you the major headlines from everywhere, no personalization or AI involved. 2. Every news story has a "Full Coverage" button on the bottom-right that pulls up that same story from a variety of stories, so you can get different perspectives. I agree that injecting AI and personalizat…

Personalisation isn't inherently bad. For example, I have zero interests in sports, and I don't think I'm putting myself in a "bubble" for not wanting any sport related articles in my feed. I also may have niche interests, such as puzzle games, which a generic feed would never surface.

Where AI gets dangerous is focusing on either a single source, or providing one side of the story, but that's exactly what "Full coverage" attempts to solve as you mention.

Re: The new Google News: AI meets human intelligence

#55

Post by ehsankia »

combatentropy wrote: post #42
kevincox wrote: post #6

The thing that gets me about Google news is that the personalization is incredibly coarse. I can see a "17 apps that take great selfies" article and the only way to express disinterest is by removing Android from my interest list. I might dislike an article because the publisher is low quality or I'm not interesting in one aspect of the content but it just picks one "reson" to justify showing the article.

From the article: > We’ve also built easy-to-use and easy-to-access controls > so you can decide if you want to see more or less of a topic or publisher.

You didn't really answer the problem? The person above was complaining that "17 apps to take a better selfie" is under the "Android" topic. Your only other choice is to block the entire publisher, which may be good if all that publisher posts is similar bad articles, but it's hard for you to know.

A far better way would be to get upvote/downvote per article, and then have the AI figure what your interests are, rather than base it purely on clicks, which tends to favor clickbait.

Re: The new Google News: AI meets human intelligence

#57

Post by croon »

thankthunk wrote: post #43
adrianmonk wrote: post #29

Earlier quoted context omitted.

> Now, google news is just mostly NYTimes and Washingtonpost stories. What? This is certainly not my experience. I just went to news.google.com in my desktop browser, and I clicked on the World section. On the Iran sanctions/deal story, there are several sources grouped together in one card. The top source is indeed Washington post but the same card also contains coverage from Mehr News Agency (which it highlights as…

> What? This is certainly not my experience. Your comment describes the experience to the tee. > The top source is indeed Washington post but the same card also contains coverage from Mehr News Agency (which it highlights as "From Iran"), Where are you from? In the NY area, all I see is US related. A couple of washingtonpost, a couple of nytimes, rolling stone, cnn, etc. Even in the expanded "view full coverage", I d…

When even Reuters and Associated Press gets branded as echo chambers, a lot of people have ostensibly let fully go of objective reality.

Re: The new Google News: AI meets human intelligence

#58

Post by croon »

remyrylan wrote: post #38
cevn wrote: post #7

Earlier quoted context omitted.

The agenda is the garbage article you just posted. I've never seen such a severe case of ad hominem.

The garbage article I posted at least sheds a little bit of light on the fact that snopes is not the bastion of integrity that for some reason it's being portrayed as via both Google and FB. Also congratulations if that's literally the most severe case of ad hominem you've ever seen. Also worth noting: I linked that article because it was the first source I found via searching that highlighted all the shady stuff tha…

> The garbage article I posted at least sheds a little bit of light on the fact that snopes is not the bastion of integrity that for some reason it's being portrayed as via both Google and FB.

The thing about that is that the "article" doesn't mention a single case of where Snopes was wrong (I only read the first few pages, so they might've buried something, which would be strategically unwise). Which means they want a character assassination because they can't argue the on the issues alone.

The contents might be true, or not, but it doesn't affect the argument.

Re: The new Google News: AI meets human intelligence

#59

Post by 16bytes »

gowld wrote: post #50
crazygringo wrote: post #32

Earlier quoted context omitted.

All I want is an option to pay $10/mo for news to all sources, and let each publisher receive the proportion of that that my clicks determine.

That's less than the price of any single news source. Why would they offer that price to you?

Because the vast majority of news consumers online are ad-supported non-paying customers.

I'm not going to pay $20/mo each to WaPo, NYTimes and WSJ. But I would pay $20/mo for "entry" level subscriptions to all of them, which I'm pretty sure is more than the Ad revenue that I would otherwise bring in.

$10/mo may not be the best price point, but IMHO having cheaper pay-as-go options would drive in a lot of unrealized revenue.

Re: The new Google News: AI meets human intelligence

#60

Post by touristtam »

thankthunk wrote: post #15
3steve wrote: post #9

I don't want personalized news, I use Google News for the exact opposite reason. I want to see semi-random headlines from a variety of source to increase my exposure to stories I might not have seen otherwise.

> I don't want personalized news, I use Google News for the exact opposite reason. That's what I used to use google news for too. Then a few years ago, google "personalized" and "localize"d news to your area and it became pointless. Google news was especially great for international events in the early 2010s because you'd get international sources listed back then. So you could see the difference in coverage between…

This is exactly how I feel but I am located in the UK and sources can be really biased on how and what they report. I usually end up consulting several foreign news outlet including what our national media would qualify as not so reputable.
Post reply on HN