YouTube Says Computers Helped It Pull Down Millions of Objectionable Videos Last Quarter (recode.net)

YouTube says it has successfully trained computers to flag objectionable videos. In the last quarter of 2017, the company reportedly pulled down more than six million of these videos before any users saw them. The news comes from a brief aside in Google CEO Sundar Pichai's scripted remarks during parent company Alphabet's earnings call today. "He said YouTube had pulled down more than six million videos in the last quarter of 2017 after first being flagged by its 'machine systems,' and that 75 percent of those videos 'were removed before receiving a single view,'" reports Recode.

Comments

Kinda defeats the purpose of youtube

[Anonymous Coward]

Doesn't censuring videos like that kinda of defeat the whole point of youtube? The idea that anybody could post whatever. Now Youtube is filled with commercial businesses. Wish there was a good alternative to post my cat videos on!

Re:Kinda defeats the purpose of youtube

[Anonymous Coward]

This YouTube algorithm has a nasty habit of "flagging" conservatives. For example, recently it "flagged" a discussion between Dave Rubin and Thomas Sowell.

How many of them were false positives?

[Lordpidey]

I remember not too long ago that a LOT of people were wondering why the hell their videos were banned, with something as innocuous as just a TF2 match.

AI is not good enough for detecting hate speech yet.

Re:How many of them were false positives?

[Anonymous Coward]

It's likely not that much.

It's also fairly obvious that that number is inflated. If objectional videos were really being pulled down, ALL of Alex Jones infowars content would never show up.

Rather "objectionable" , includes reaction videos and uploads/re-uploads of television footage that was live not to long ago, before ContentID gets to it. If they have successfully trained AI to be able to tell the difference between a video game and live footage, that would be an amazing breakthrough. That said, most of what is "easy" to block objectionable content is by having the auto-subtitle system pick up violent language.

Re:How many of them were false positives?

[Anonymous Coward]

"Hate speech".

I bet you would have been the type to have called the Founding Fathers hate speeches for their anti-British stance back in the day. I understand Youtube is a private company with no reason to uphold anybody's free speech, and that I don't mind... but people like you would give it away that right wholesale without a fight everywhere and help barricade any companies who do want to provide it, and then wonder why in 20 years down the road the ruling political elites have no fear not representing any of your interests.

AI is certainly already good enough at replacing your shilling with their own.

Re: How many of them were false positives?

[DNS-and-BIND]

When you silence a man, you only show that you fear what he has to say.

Re:How many of them were false positives?

[Anonymous Coward]

AI is not good enough for detecting hate speech yet.

That would a feat. As far as I'm aware, humans can't do that reliably either.

U haters, hos and bots

[harvey the nerd]

Mostly this means Youtube is a Conservative-Libertarian-TeaParty-Trump hater and an RIAA/MPAA ho and bot network.

Define 'Objectionable'

[Anonymous Coward]

YouTube Bans Firearms Demo Videos, Entering the Gun Control Debate

Re:Who Judges??

[BlueStrat]

So, if Google/YouTube had some point of view, or points of view, that they wished to either promote or demote... how do we know if they are protecting us or harming us? I do data science sorts of things as part of my job. I know that very, very minor tweaks to algorithms provide quite different results.

Well, as long as they haven't nuked my '80s retro videos, then I'm good. Ah, Leah!

It's quite possible now to place individuals and groups into internet "algorithm ghettos". Like the Nazis did to Jews in Poland by crowding them into a section of the city and erecting tall walls both to prevent escape and so the people outside the walls didn't see what went on inside, companies like Google, Facebook, and others can place individuals and groups into a digital algorithm ghetto where what they can access is controlled & filtered and what they publish/post/send can be filtered or blocked such that you'd have no way to tell nobody could actually see what you uploaded even to your own website, especially with a well-trained AI tasked with the job. If you call a friend to ask them to verify it the AI will know who you know and listen/monitor the call/message and make it visible to him so you remain unaware that the greater web doesn't know you exist.

This is an extremely dangerous time where how we react or don't right now to control these technologies and the governments and corporations that use them will have very serious and long-lasting consequences for many generations to come around the world.

Strat

Re: Who Judges??

[Reverend Green]

The answer is obvious: Big Brother Google loves us all. If Google wants to enserf us, it's *for our own good*.

Re:AI

[stephanruby]

You may have been downsized because of this, but don't believe all the hype. AI only helps partially. Youtube still needs human reviewers (even if they're unwilling to pay them).

Remember what prompted the advertiser pullouts last year, youtube was still incapable of filtering very obvious unambiguous swear and racist language from the text subject lines and the text descriptions of its hosted videos. To me, that just means that they didn't care, and/or that they were unwilling to pay for that kind of manual sifting by actual human beings.

googliness

[Reverend Green]

Capitalist stooges sure are proud of mass censorship.

Comment removed

[account_deleted]

Comment removed based on user account deletion