0:00:06.070,0:00:07.120
Hi, my name's John.

0:00:07.510,0:00:10.140
I lead the search and machine[br]learning teams at Google.

0:00:12.130,0:00:14.230
I think it's amazingly inspiring

0:00:14.230,0:00:16.214
that people all over the world

0:00:16.215,0:00:19.160
turn to search engines to[br]ask trivial questions

0:00:19.160,0:00:20.930
and incredibly important questions.

0:00:20.930,0:00:23.450
So it's a huge[br]responsibility to give them

0:00:23.450,0:00:24.864
the best answers that we can.

0:00:26.710,0:00:30.610
Hi, my name's Akshaya and [br]I work on the Bing search team.

0:00:30.910,0:00:33.190
There are many times where[br]we will start looking

0:00:33.190,0:00:35.800
into artificial intelligence[br]and machine learning,

0:00:35.830,0:00:39.010
but we have to address how are[br]the users going to use this,

0:00:39.140,0:00:42.390
because at the end of the day,[br]we want to make an impact to society.

0:00:43.780,0:00:45.400
Let's ask a simple question.

0:00:45.820,0:00:48.070
How long does it take to travel to Mars?

0:00:49.330,0:00:50.950
Where did these results come from

0:00:51.370,0:00:54.100
and why was this listed[br]before the other one?

0:00:55.700,0:00:58.150
Okay, let's dive in and[br]see how the search engine

0:00:58.150,0:00:59.860
turned your request into a result.

0:01:00.690,0:01:03.360
The first thing you need to[br]know is when you do a search,

0:01:03.430,0:01:06.480
the search engine isn't actually[br]going out to the World Wide Web

0:01:06.480,0:01:08.010
to run your search in real time.

0:01:08.140,0:01:10.610
And that's because there's[br]over a billion websites

0:01:10.610,0:01:14.140
on the internet and hundreds more are[br]being created every single minute.

0:01:14.140,0:01:16.210
So if the search engine[br]had to look through

0:01:16.240,0:01:18.690
every single site to[br]find the one you wanted,

0:01:18.690,0:01:20.120
it would just take forever.

0:01:20.500,0:01:21.940
So to make your search faster,

0:01:21.970,0:01:24.940
search engines are constantly[br]scanning the web in advance

0:01:25.420,0:01:28.560
to record the information that might[br]help with your search later.

0:01:28.930,0:01:31.270
That way, when you search[br]about travel to Mars,

0:01:31.630,0:01:33.700
the search engine[br]already has what it needs

0:01:33.700,0:01:35.728
to give you an answer in real time.

0:01:36.250,0:01:37.540
Here's how it works.

0:01:37.900,0:01:42.010
The internet is a web of pages[br]connected to each other by hyperlinks.

0:01:42.400,0:01:44.680
Search engines are[br]constantly running a program

0:01:44.680,0:01:47.380
called a Spider that cross[br]through these web pages

0:01:47.380,0:01:49.040
to collect information about them.

0:01:49.780,0:01:51.550
Each time it finds a hyperlink,

0:01:52.090,0:01:55.000
it follows it until it[br]has visited every page

0:01:55.030,0:01:57.240
it can find on the entire[br]internet.

0:01:57.335,0:01:59.170
For each page the spider visits,

0:01:59.200,0:02:02.320
it records any information[br]it might need for a search

0:02:02.500,0:02:05.650
by adding it to a special[br]database called a search index.

0:02:07.166,0:02:09.530
Now, let's go back to[br]that search from earlier

0:02:09.590,0:02:11.990
and see if we can figure[br]out how the search engine

0:02:11.990,0:02:13.333
came up with the results.

0:02:13.640,0:02:16.460
When you ask how long does[br]it take to travel to Mars,

0:02:16.640,0:02:18.860
the search engine looks[br]in each of those words

0:02:18.920,0:02:21.410
in the search index to[br]immediately get a list

0:02:21.410,0:02:24.500
of all the pages on the[br]internet containing those words.

0:02:24.890,0:02:26.870
But just looking for these search terms

0:02:26.870,0:02:28.760
could return millions of pages,

0:02:28.760,0:02:31.110
so the search engine needs[br]to be able to determine

0:02:31.110,0:02:33.120
the best matches to show you first.

0:02:33.340,0:02:36.010
This is where it gets tricky[br]because the search engine

0:02:36.010,0:02:38.040
may need to guess what[br]you're looking for.

0:02:38.930,0:02:41.360
Each search engine[br]uses its own algorithm

0:02:41.360,0:02:44.230
to rank the pages based on[br]what it thinks you want.

0:02:44.930,0:02:47.660
The search engine's ranking[br]algorithm might check

0:02:47.990,0:02:50.360
if your search term shows[br]up in the page title,

0:02:50.900,0:02:53.820
it might check if all of the[br]words show up next to each other,

0:02:54.520,0:02:57.020
or any number of other calculations

0:02:57.020,0:02:58.610
that help it better determine

0:02:58.670,0:03:01.420
which pages you'll want[br]to see and which you won't.

0:03:02.960,0:03:04.960
Google invented the most [br]famous algorithm

0:03:04.960,0:03:08.530
for choosing the most relevant results[br]for a search by taking into account

0:03:08.560,0:03:11.230
how many other Web pages[br]linked to a given page.

0:03:11.830,0:03:14.140
The idea is that if[br]lots of websites think

0:03:14.140,0:03:15.660
that a web page is interesting,

0:03:15.660,0:03:17.940
then it's probably the one [br]you're looking for.

0:03:18.190,0:03:20.020
This algorithm is called page rank,

0:03:20.590,0:03:22.330
not because it ranks web pages,

0:03:22.570,0:03:25.210
but because it was named after[br]its inventor, Larry Page,

0:03:25.480,0:03:27.333
who's one of the founders of Google.

0:03:27.940,0:03:30.520
Because a website often makes[br]money when you visit it,

0:03:30.820,0:03:32.950
spammers are constantly[br]trying to find ways

0:03:32.950,0:03:35.741
to game the search algorithm [br]so that their pages

0:03:35.742,0:03:37.931
are listed higher in the results.

0:03:38.260,0:03:40.750
Search engines regularly[br]update their algorithms

0:03:40.750,0:03:44.296
to prevent fake or untrustworthy[br]sites from reaching the top.

0:03:44.680,0:03:47.350
Ultimately, it's up to you [br]to keep an eye out

0:03:47.500,0:03:49.450
for these pages that are untrustworthy

0:03:49.690,0:03:52.990
by looking at the web address and[br]making sure it's a reliable source.

0:03:53.680,0:03:55.390
Search programs are always evolving

0:03:55.420,0:03:58.420
to improve the algorithms [br]wo they return better results,

0:03:58.540,0:04:00.460
faster results than their competitors.

0:04:01.000,0:04:03.100
Today's search engines[br]even use information

0:04:03.100,0:04:06.820
that you haven't explicitly provided[br]to help you narrow down your search.

0:04:07.150,0:04:10.120
So, for example,[br]if you did a search for dog parks,

0:04:10.240,0:04:12.190
many search engines[br]would give you results

0:04:12.190,0:04:13.840
for all the dog parks nearby,

0:04:14.080,0:04:16.260
even though you didn't[br]type in your location.

0:04:17.800,0:04:20.530
Modern search engines[br]also understand more

0:04:20.530,0:04:22.060
than just the words on a page,

0:04:22.300,0:04:24.970
but what they actually mean[br]in order to find the best one

0:04:24.970,0:04:26.750
that matches what you're looking for.

0:04:27.130,0:04:29.980
For example, if you search [br]for fast pitcher,

0:04:30.280,0:04:32.300
it will know you're[br]looking for an athlete.

0:04:32.500,0:04:34.450
But if you search for large pitcher,

0:04:34.450,0:04:36.730
it will find you options[br]for your kitchen.

0:04:38.420,0:04:41.910
To understand the words better, [br]we use something called machine learning,

0:04:41.910,0:04:43.985
a type of artificial intelligence.

0:04:43.985,0:04:46.050
It enables search[br]algorithms to search out

0:04:46.090,0:04:48.400
not just individual letters[br]or words in the page,

0:04:48.400,0:04:51.280
but understand the underlying[br]meaning of the words.

0:04:53.690,0:04:55.850
The internet is growing exponentially,

0:04:56.210,0:04:59.810
but if the teams that design[br]search engines do our jobs right,

0:05:00.080,0:05:04.090
the information you want should[br]always be just a few keystrokes away.