Natural Language Processing: Crash Course AI #7

For more information go to curiositystream.com/crashcourse
So far in this series, we've mostly focused on how AI can interpret images, but one of the most common ways we interact with computers is through language - we type questions into search engines, use our smart assistants like Siri and Alexa to set alarms and check the weather, and communicate across language barriers with the help of Google Translate. Today, we're going to talk about Natural Language Processing, or NLP, show you some strategies computers can use to better understand language like distributional semantics, and then we'll introduce you to a type of neural network called a Recurrent Neural Network or RNN to build sentences.
Crash Course AI is produced in association with PBS Digital Studios
/ pbsdigitalstudios
Crash Course is on Patreon! You can support us directly by signing up at / crashcourse
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, Yasenia Cruz, Eric Koslow, Caleb Weeks, Tim Curwick, David Noe, Shawn Arnold, Andrei Krishkevich, Rachel Bright, Jirat, Ian Dundore
--
Want to find Crash Course elsewhere on the internet?
Facebook - / youtubecrashcourse
Twitter - / thecrashcourse
Tumblr - / thecrashcourse
Support Crash Course on Patreon: / crashcourse
CC Kids: / crashcoursekids
#CrashCourse #ArtificialIntelligence #MachineLearning

[@patriciamcgeorge2575]

Linguist here. Am impressed by the level of accuracy/depth. Please do a Crash Course linguistics. It's an underrated field and breaks my heart to see just how small it is in the shadow of physics or history - but you guys can definitely shed some light on it!

[@hamzadbz1]

We are on our way towards Dystopian society and you want a Crash Course on Linguistics. Why would you do that?

[@patriciamcgeorge2575]

@@hamzadbz1 What do you mean why would I want that? 1) Why do you say we're on our way to a dystopian society? Pretty big claim of you don't have backup 2) Why the hell would that have anything to do with a crash course on linguistics? The two have nothing to do with each other! Linguistics is a cool, underrated and underappreciated field. Nothing to do with dystopias

[@hamzadbz1]

@@patriciamcgeorge2575 Haha Just Messing around with you. All i want is a linguistics Crash Course now.

[@erichodge567]

The more I learn about computers, the more impressed I am with brains.

[@thisis4573]

and it all works on 20 watts

[@johnopalko5223]

Very well done and enjoyable series. I'm looking forward to the next episode. I did some postgraduate work in AI back in the 1980s and we thought we were right on the cutting edge. Seeing something like this and looking back, I realize just how primitive our tools and techniques actually were.

[@AyaanAhmed12]

Can you guys please start a Mathematics course. It would be amazing to learn maths with such amazing graphics.

[@Koldunja]

Chocolate potatoes do exist. It's a desert in Russia. There is nothing from potato in them except the shape. ;)

[@sam62831]

I went to the comments to say this!

[@mattkuhn6634]

I've been looking forward to this episode for a while - I'm a Computational Linguist, so this is all my bread and butter. In fact, I took a class taught by Ray Mooney in my undergrad, and my graduate work mostly centers around language models for home assistants, so I use a lot of RNNs. Great job with this series guys!

[@BrainsApplied]

Thanks for this clear, explanatory video! You guys are the best!

[@maelysouza3079]

Ótima aula! Adorei essa série de vídeos!

[@LearnandGrowKidsTV]

Great video! Enjoyed watching! We’re so glad we found this channel!

[@PranayZVlogs]

Excited!!!!

[@somedragontoslay2579]

That translation at 1:35 was nuts! Only the last two words were unnatural!

[@IceMetalPunk]

Hey, a surprise linguistic cameo by Physics Girl! Is it really Dianna's birthday next week?

[@a.r.5718]

Ah, it's big brain time.

[@adeshpoz1167]

I see culture here. 😄

[@000Krim]

cobblestone generator time!

[@juniormynos9457]

Damn... that was the basics?

[@exodproduction]

shoutout for Physics girl!

[@oliviar2055]

It’s super interesting 😀

[@seka1986]

Very good 👍

[@janyjj1969]

I WOULD LOVEEEE IT IF CRASH COURSE HAD AN ACCOUNTING COURSE!!❤️️.

[@janyjj1969]

@@TomMcMorrow Exactly man, im not doing so well in accounting for my IGCSE. And I do belive crash course could definetly help me, like it did with economics. My Accounting IGCSE exams start on the 30th of october. 😓💔

[@forgottenforever1036]

Lovely!

[@cartoonist7578]

WOW deep man

[@DJMarkCorneliusThaDon]

I like this host.

[@AyrtonTwigg]

I so hope there will be a lab on reinforcement learning!

[@coypus7949]

hey my dudes watch your videos a lot in our classes but it is sorta hard to suck it all in. You do give a lot of good info but are speaking realy fast. Just a little thing. Thank you :)

[@lincolnpepper816]

why do people still make this complaint? Not only is he speaking incredibly slowly, but you can slow down the video through settings -> playback speed.

[@coypus7949]

Sry not on this dude my bad

[@warpedmine9682]

When I opened this video not watching the series yet and I was not expecting my mans to be there and even using the lips instead of it being stuck like super glue

[@rehanalampanah]

hey, Your teaching style is awesome. why aren't you trying to make a complete course on AI instead crash course.

[@mglsj]

Watching this 3 years later when CHAT GPT is out which can basically do the same thing.

[@reigh7]

Can you use known identifying words like types of medication routes as statically programmed to find all the sentences pertaining to parse them out and run a loop to concatenate? Also The heading key words then search a whole reference and parse those by listing by route and add any general sentences for the heading and dose instructions sentence? I 'm pretty sure that right to get my baseline then from there have it try and use machine learning to summarize those concatenated rows one by one to reduce character counts and make it more clear understandable ie flow in sentence structures when put back together.

[@AnaleenAelwyn]

Human languages can also be visual: sign languages! 😀

[@Beanie--]

Him: I've never heard of chocolate potatoes so they probably don't exist. Chocolate covered potato chips: i'm about to end this man's whole career.

[@user-cw8vt3zc8t]

we are beating around the bush with this

[@user-cw8vt3zc8t]

not much of that is needed - just write it perhaps so that it only understands story first - references are not needed

[@DoctoreDoom]

...this fire, fu gonna be lit af b. Ai ight.. so tight *Robot explodes*

[@dullbananas9901]

Don't drink and code.

[@shhac]

I'm kinda hungry, I think I'd like some chocolate please. I'm kinda hungry, I think I'd like some chocolate thanks . I'm kinda hungry, I think I'd like some chocolate now. I'm kinda hungry, I think I'd like some chocolate later. I'm kinda hungry, I think I'd like some chocolate today . I'm kinda hungry, I think I'd like some chocolate suggestions.

[@mesplin3]

Since all are valid sentences, we would expect those to have a high probability than "chocolate swimming" for example.

[@void2509]

I'm kinda hungry, I think I'd like some bagels please. I'm kinda hungry, I think I'd like some bagels thanks . I'm kinda hungry, I think I'd like some bagels now. I'm kinda hungry, I think I'd like some bagels later. I'm kinda hungry, I think I'd like some bagels today . I'm kinda hungry, I think I'd like some bagels suggestions.

[@GeeTransit]

I'm kinda hungry, I think I'd like some chocolate bagels.

[@stevedoe1630]

Paul Somers From a language perspective, all options can work. From the perspective of predictive language generation, I would assume the weightages ware being computed for all these options...presume the model is based on a lot of things, if the speaker is polite then higher weightage for please and thanks, if the model recognized recent talks about “needing food” or “having food” then weightages would be applied as appropriate. In a digital world, there is a lot of gray area between 0 and 1 for A.I.

[@tasha9863]

Do crash courses have book out?

[@sr3yu]

Pls do a video on merchant Navy

[@fullerdb]

Three words to run through the AI: chocolate mashed potatoes.

[@sortedsortof3474]

And you start with, "Is John Green a robot?" ... ... Hahahahaahahaa

[@busyhacker8129]

Please make vedio on Deep reinforcement learning and give deep knowledge about policy networks,Value functions and tree searches.

[@KingGhidora97]

Sometimes i speak japanese fluently in spanish

[@APR4U]

The cat drove a car through the mall and collided with a skier of the same species. 🤦‍♂️ APR4U 🇺🇸

[@geoffreywinn4031]

Jabril keeps it ril.

[@jonniheinisto9825]

I also checked, and chocolate potatoes do exist. I had to know.

[@Herr_Vorragender]

If Prof Mooney is correct, then maybe text is the wrong media for neural networks. Maybe a way must be found to translate the world just as humans do, namely through pictures. Pictures are like paragraphs of sentences, but zipped. Image pattern recognition is already a thing, if only a small thing. If now a sentence can be translated into an image, then the patterns can be stored more efficient than words. Take "A man riding a horse through the woods". I imagine a man my age, then a horse, the man on the horse, then woods and finally a direction which alters the legs of the horse. I imagine laying these pictures on top of each other. And it doesn't even matter if my image of woods differ from any body else. Every picture element can very much be translated into a vector. And these vectors can be set into a distance calculation to lets say rain forest, dark, light or any other kind of forest. There would not need to be a vector for the meaning of a sentence. The vectors for the elements would suffice. There is a flip side to all this though. It opens up the the guarantee that AI could very much be equally as filtered, plastic, biased and faulty as the human mind. Also AI then might just be equally as creative like the human mind. At least the AI attention span would exceed humans by worlds.

[@dejabu24]

images are much more complex to store and process than text , it depends on the application imo , besides that there are things that you can't represented by images

[@masternobody1896]

Petition to make forrest robot again

[@mentality_speech]

4th ;)

[@shis10]

👍😃

[@BradleyKoch]

Man, are computers incapable of learning what a run-on sentence is? GIGO

[@Diamond_Hanz]

Take a shot Everytime he says 'or'

[@smooofer]

take a shot everytime he says "the"

[@paulgraebnet7896]

First

[@splatboynick5149]

im pretty hungry so im gonna eat some chocolate milk

[@KiraFriede]

Now take a language in which articles change with gender and grammatical case of the word and you will have a mess.

[@bobthepuppet5834]

Hi

[@lincolnpepper816]

hi

[@cagethemouse]

poggers

[@spud1958]

ili "mačka" = cat :)

[@shanshanlin]

Make one about the cOroN4v1rUs

[@sanjaysantiago7414]

3rd First time I got someone to like my post...thx

[@TeflonDon451]

He looks like Joel Embiid..

[@SansAppellation]

Natural English Processing or NLP for short Huh?

[@prisonss]

You missed out trump language !

[@ridwansetiadi8393]

At the very moment I heard him saying that chocolate potatoes didn't exist, I was pretty sure commenters would find where to find chocolate potatoes. And, according to some comments, chocolate potatoes exist. 🤣

[@doha17461]

WHY DIDNT I KNOW THERE WAS A CRASH COURSE AI?????????? OMG!!!!!!!!!11

[@robossthinking1056]

So Billie Eilish is a cat ,

[@reigh7]

I want to see about getting Dianna a house near the LIGO in my home town and live peacefully with her. My cousin wants me home too and even though my Uncle wants to sell my Grandma's place that her dad built she was talking about getting me in there away from here back to home for her birthday if my birthday present of this 10 year programming project worked out I wouldn't even need it we could build a new one on the property. That's how my cousin was talking despite me being a on disability. Wait too forward huh ok slow it down take it easy breath ligo girl who likes ligo beautiful intelligent but be realistic screw it I'm manic let me go on a date with her.

[@notthetsar438]

feel bad for your fridge mate.... voel sleg vir jou fridge maat…. ndjeheni keq për bashkëshortin tuaj të gjallë…. վատ զգալ ձեր գորշ զուգընկերոջ համար…. чувствате се зле за вашата хладилна половинка .... чувствую себя плохо из-за твоего приятеля sentirsi male per il tuo amico compagno .... Sentirse mal por su amigo frige ... mår dåligt för din frige kompis ...

[@masterbaiter5533]

I say reinvent the wheel. Cant understand any of it, must destroy for safety. Have you though of slowing down, life is confusing already. Wait for everyone to catch up. There is nothing out there.

[@MrSoggycat]

Chinese isnt a language. Mandarin you mean?

[@ICouldntThinkOk]

I’d call it a language. Arabic, for example, has multiple types of itself, both spoken and written differently.

[@MrSoggycat]

@@ICouldntThinkOk nah. Tell me what is the "chinese language" then

[@ICouldntThinkOk]

Jason Jason The “Chinese language” is a general term for the different types of Chinese.

[@MrSoggycat]

@@ICouldntThinkOk I am chinese =/= I speak chinese. I speak mandarin, hokkien, teochew, hakka, cantonese, etc etc. IE there is no chinese LANGUAGE per se.

[@lincolnpepper816]

@@MrSoggycat but people understand what you mean when you say you speak Chinese.

[@benji4114]

I'm getting distracted with all the head movements

[@maxmusterman3371]

Why does it feel like this channel is being owned and used by big companies? Training their workers and fixing them onto their products as well. This is concerning to me. Where are the days of the chom choms? Sad that such important subjects are sold ergo lose independance

[@IraFinn]

Was disappointed y’all left out sign languages. They’re natural languages just like spoken languages. “Languages are spoken, written, or signed". Inclusivity is just a single word away.

[@thepengu1no227]

lol first dislike just cause I would have liked it, but first 10th comment, too! Great job with the vids CC, i'm happy to be learning and entertained at the same time!

[@ICouldntThinkOk]

But you wanted to like it? Why? Why would you dislike it? If you like it it’s going to recommend more like this...

[@dullbananas9901]

chee chea shusss chung chang chung this is natural chinese language Edit: 把我的筷子給我。

[@thepengu1no227]

racist, but funny

[@tommycastro9424]

Harman Grewal but what if he/her/they/it/them/xie/cis/zy\(you get the idea) is chinese

[@dullbananas9901]

@Truth Only use google translate

[@dullbananas9901]

@Truth Only no. my wheelchair logo is just a weird design i made for some reason. subscribe to my channel

[@dullbananas9901]

@Truth Only you might like the one with sid the science kid

[@asekuvena]

The simplest possible language is a monogrammar ☝ language 📜. The grammar rule ➊ of a sentence existing is all the grammar needed. 😆 Directly translating a monogrammar ☝ language 📜 into the non-monogrammar ❌☝ languages 📜📜📜 humans 👥👥👥👥 have results in gibberish 😵‍💫. Thus, in order to translate a monogrammar ☝ language 📜, you need to know a monogrammar ☝ language 📜. ----------------------------------------------------------------------------------------------------------------------------- Anyway, the point is, it is useful for AI 🤖, as it is modular, and easy to start learning. It can also be used as a reference for understanding common human languages. 😆

[@bobthepuppet5834]

Hi

[@lincolnpepper816]

hi

[@bobthepuppet5834]

Hi

[@lincolnpepper816]

hi