How YouTube knows what you should watch: Crash Course AI #15

Today we’re going to talk about recommender systems which form the backbone of so much of the content we see online from video recommendations on YouTube and Netflix to ads we see on Facebook, Twitter, and well, everywhere else. We’ll talk about their types of systems - content-based, social, and personalized recommendations - and take a closer look at what they're good at, but also why they often fail.

Thanks to the following patrons for their generous monthly contributions that help keep Crash Course free for everyone forever:

Eric Prestemon, Sam Buck, Mark Brouwer, Indika Siriwardena, Avi Yashchin, Timothy J Kwist, Brian Thomas Gossett, Haixiang N/A Liu, Jonathan Zbikowski, Siobhan Sabino, Zach Van Stanley, Jennifer Killen, Nathan Catchings, Brandon Westmoreland, dorsey, Kenneth F Penttinen, Trevin Beattie, Erika & Alexa Saur, Justin Zingsheim, Jessica Wode, Tom Trval, Jason Saslow, Nathan Taylor, Khaled El Shalakany, SR Foxley, Sam Ferguson, Yasenia Cruz, Eric Koslow, Tim Curwick, David Noe, Shawn Arnold, William McGraw, Andrei Krishkevich, Rachel Bright, Jirat, Ian Dundore
--

Want to find Crash Course elsewhere on the internet?

#CrashCourse #MachineLearning #ArtificialIntelligence

[Siricerasi]

I wish that when you pushed the “ive watched this” button it didn’t recommend it again five minutes later like I’ve literally watched a video and then went to the home page and there the video is even with the red line underneath to show I’ve just watched it. Grrr

[TheBlizzardfan]

Thanks from Iran, after one week of disconnectivity, I rushed to watch this new episode of AI CCourse 💖🤍💚

[innocentoctave]

No mention at all of the underlying purpose: getting you to spend more, whether directly by buying products and services, or indirectly by linking to further advertising, or simply by encouraging you to spend more time online - which increases the likelihood of being exposed to advertising, and of disclosing more information about your consumption patterns. It isn't just an abstract intellectual problem.

[eternalreality]

Title should be: “How YouTube suggests what they think you should watch”

[dvklaveren]

As the CEO of an app development company aiding the disabled community, I am very worried not just about recommendation systems giving snake oil salesmen access to the disabled community, but also that Twitter recently barred all ads targeting health.

Where before, the disabled community had to worry about being exploited, disabled entrepeneurs are now kept from reaching their own demographic.

The stupid thing is; family members can still be targeted, but they'll have far less clarity on the practicality of a solution. So a family member of someone with autism is just as likely (more likely) to be hawked on fidget toys that only their autistic family member would know if it would work for them, as actually helpful solutions like decision support systems.

It's very frustrating.

[rawsteel]

YouTube recommendations: We are the most mysterious software in the world.



Crash course: Hold my BEER.

[ericcameron]

I hate it when google shows me a bunch of ads for products I've already looked at and have purchased.

[kittentude]

YouTube gets it so wrong it has become exceptionally frustrating. The 'not interested' and 'don't recommend this channel' choices are as bogus as it gets EPIC FAIL. All that Google tech and it is simply A-1 junk.

[ishikaa]

I was trying to close the ad which was part of the video...LOL

[rchave]

Youtube only recommends videos based on the criteria that I've already watched them. I'm no programmer but I reckon I could write that algorithm myself.

[megamangos]

Hold up, I'm pretty sure all incognito browsing does is not keep/store any of your data to your local computer. Correct me if I'm wrong, but I'm pretty sure you're still tracked and monitored with or without it.

[whyuactingsopepegasusbro]

I'm getting recommended from Tinder and Mature Moms... Bruh I only watched a Tinder experiment

[phillipj]

For instance the algo got tired of me complaining about it in comments so it sent this video my way to see if it could get a little understanding. Or it's more likely because im subscribed to crash course.

[ronaldfigueroa]

Now I know, why Facebook is scaring me after I thought and shows me it

[TheTwick]

Sure, you want me to go into your “quality programming echo chamber” and watch more quality programs and buy quality goods! Ah ha! I’ve uncovered your dastardly scheme. Can’t fool me!

[erzsblasfantaven]

it knows when you are sleeping
it knows when you're awake
it knows what you like to watch
so avoid text to speech meme compilations for goodness sake

[areamusicale]

What about when you click by mistake in a video?
The saddest part is that to get rid of certain suggestions we have to dislike videos, even if the videos doesn't deserve a dislike.
For example, I have to be careful if I click videos with cats or dogs, for the rest of the month I will receive suggestions about cats and dogs .... I don't like cats and dogs!!!!

[Tfin]

Over the past week, YouTube has proven it DOESN'T know what I should watch. Just today, I watched a video that honestly could have used censoring, and was then recommended Peppa Pig and a bunch of other things aimed at 6-year-olds. Those aren't here now. Now the recommendations are for SciShow, Crash Course, and a few other things, most of which I've either already watched or decided to not watch because list shows don't say enough about any of things they list. Oh, also, a compilation video that is literally just a bunch of videos I've seen already.

[ruralpastorva]

Why did YouTube suggest this? I didn't want to watch it.

[baqirqadri]

Very informative and interestingly presented