Google to ramp up AI efforts to ID extremism on YouTube | Hacker News Reader

Google to ramp up AI efforts to ID extremism on YouTube

techcrunch.com

14 pointsjanober8 comments

Earlier related discussion: https://news.ycombinator.com/item?id=14583017

"expanding counter-radicalisation efforts by working with (other Alphabet division) Jigsaw to implement the “Redirect Method” more broadly across Europe. “This promising approach harnesses the power of targeted online advertising to reach potential Isis recruits, and redirects them towards anti-terrorist videos that can change their minds about joining. In previous deployments of this system, potential recruits have clicked through on the ads at an unusually high rate, and watched over half a million minutes of video content that debunks terrorist recruiting messages,” says Walker."

Does them performing this kind of intentional manipulation, and having such success, scare the shit out of anyone else?

How about employing real people this time, google?

Your previous efforts to police crafty humanss using AI has utterly failed. Just look at adsense and play store.

Because the massive troll armies will post far more content then a massive number of humans could police. One hour of video is uploaded to YT every second, from a statistic 4 years ago.

Using your metrics, it's just better for Google to turn the service off to free uploads.

There is something like a decade of video uploaded to YouTube every day.

Even if you employed entire countries you still wouldn't stand a chance at reviewing it all.

Automated systems are the only way it can function at all.

> Automated systems are the only way it can function at all.

Not in its current form. Google with all its might hasn't even managed to remove those "work from home for $$$" youtube comments.

Google doesn't have a good way to incorporate human intelligence (users) into its AI. Pure AI has no chance against an army of highly adaptable humans.

They're also employing more real people.

It's impossible to do at this scale, as other posters have already pointed out.

More to the point, machine learning is just building a model of what humans would do. Feed thing a million non-extremist posts, and a mission extremist posts, all of which have been manually marked as such by a human. Then ask it whether image 2000001 looks like the first million or the second million.

Reply on news.ycombinator.com