Why Artificial Intelligence is more human than you think
Transcriber:
Reviewer: Maria Pericleous
Today, I want to challenge your
perspective about artificial intelligence.
You may have already heard
that A.I. is everywhere
When you scroll through content on social
media, A.I. is deciding what you see.
When you walk into some
banks and ask for a loan,
A.I. algorithms decide
whether you deserve it or not.
Some companies even use A.I.
to decide who to hire.
What you may not have heard, though,
is that you can play a part
in this revolution.
You see, even though A.I. is always taking
more and more space in our lives,
just an extremely limited amount
of people understand it,
let alone use it for their own gain.
The majority of people either fears
A.I. or passively ignores it.
And if you are part of this last group,
I want to change this.
In 10 minutes from now,
not only are you going to have
a better understanding of A.I.,
but you are going to realize that A.I.
is one of the most powerful technologies
we have available
and is just a tool in the hands of humans.
And since my assumption
is that all of you here are humans,
this means that is a tool that
you can use as well.
If I succeed,
I’m going to have inspired you
to embrace one of the most powerful
technologies we ever had available.
So are you ready?
I’m going to start with
my personal experience with A.I.,
how I decided to start working with A.I.,
because you see,
I wasn’t supposed to be here.
I’m an energy engineer, so I was supposed
to be designing solar panels
or wind turbines, that kind of stuff.
A.I. had nothing to do with it.
But then one day I received
an email that changed my life.
I had been selected to go study
and work in Silicon Valley,
absorb as much as I could
from that magical place,
and then come back to my country
to use all the knowledge.
And to give you an idea about
what Silicon Valley represents
for someone like me,
who is passionate about technology,
It’s a little bit
like the Champions League
for a football fan.
All the best minds in the world
have worked there.
From Steve Jobs
and Bill Gates back in the day,
to Elon Musk and Mark Zuckerberg today,
all of them focused in a small place
with less people than my hometown.
So I remember when I arrived there,
one of the first days I rented a car
and I started driving from Silicon Valley
up until San Francisco
and I saw a giant billboard on the highway
it said the name of a company,
Artificial Intelligence
and machine learning.
And I thought, wait a minute.
If I’m driving on the highway in Italy,
I am going to see advertising
of pasta companies,
maybe some fashion companies,
but for sure,
I am not going to see anyone
advertising A.I. on the highway.
So what is going on here?
Either these people are completely crazy,
or I’m missing something.
I needed to figure out
which one of the two was the case,
so I started doing some research
and one pivotal moment
was when I met one of the pioneers
of A.I., Peter Norvig.
He was talking about how computer
scientists were trying to build software
to automatically translate
human languages.
So this is what he told me.
Let’s suppose that you want
to build some software
that translates English to Italian.
This is what you will have
to do without A.I.
You will have to find a way
to express in rules all the nuances
of Italian and English grammar.
And that’s super hard.
Computer scientists have been trying
to do that for decades
with very poor results.
But then he told me how A.I.
completely changes the game.
This is what you will have to do with A.I.
You will take a bunch of books in English,
a bunch of books in Italian,
the direct translation of these books.
You will give them to a computer
and tell it, hey,
I am not going to explain to you
how to translate these two languages,
but these are some examples,
go figure it out.
This approach is called machine learning.
And the beauty of it is that once
computer scientists did all the hard work
of giving computers the gift of learning,
then all of us can use this technology
to teach computers what we want.
Do you want to teach
computers how to drive a car?
Cool! Get enough data
of people driving,
give it to a computer,
and it will figure out how to do it.
Do you want to teach a computer how
to diagnose a disease? Awesome!
Just get enough X-rays
of sick and healthy individuals,
give them to a computer,
and it’s going to learn how
to differentiate between the two.
And this is the first reason
that made me fall in love with A.I.
Its basic principles are pretty easy
to understand for everybody,
because this is how we learn.
I don’t know about you, but my mum
didn’t teach me how to speak Italian
by giving me a grammar book telling me:
“Hey Gianluca, go to your room, read it,
and come back once you’ve finished it”
She just spoke to me,
and my brain by itself
was able to decode all these complex rules
that make up Italian grammar.
But you may think, OK, Gianluca, cool,
this sounds pretty easy.
It’s pretty simple, right?
Why didn’t we do it 20 years ago?
It turns out that today
we are in the best moment
in history to build A.I.
And this is because to build A.I.
we need two ingredients.
The first one is data.
So all the examples that we need
to give the computer to learn. Right?
So since I started this pitch
roughly five minutes ago,
actually I had the timer there,
More than two thousand hours of video
have been uploaded to YouTube.
More than 200 million messages
have been shared on Whatsapp,
more than 1.5 million
stories on Instagram.
This is an ocean of data completely
overwhelming for any human being,
but a goldmine for computers
that can learn from all this data
and understand our world.
But we need also a second ingredient,
because you see, computers
need to be powerful
to learn from all this data.
So let’s see how we are doing today
compared to 20 years ago.
20 years ago, the most powerful
computer in the world
was owned by the U.S. Government.
They were using it to make simulations
for the nuclear power plants.
It cost forty seven million dollars.
It occupied 150 square metres, as big
as a big apartment, you know,
and it had 1.3 teraflops
of computing power.
A teraflops is just a unit of measure
of computing speed.
It makes a trillion operations per second.
Let’s see how we are doing today.
So what I brought here
is a special kind of chip.
This is called a graphical
processing unit,
and it’s particularly efficient to train
some of A.I. models called neural networks
and today you can buy
some of these models
for two thousand dollars.
And this will give you roughly
one hundred and thirty teraflops,
a hundred times more
than the most powerful computer
we had 20 years ago.
Just let that sink in for a minute.
Right now, in my hand,
I’m holding 100 times the computing power
that the U.S. government
had available 20 years ago
for the price of a used motor bike.
So this is the second reason
that made me fall in love with A.I.
It’s Democratic.
You don’t need multibillion dollar budgets
to start building A.I.,
all of you potentially
could go back home tonight,
open your laptops and start building
some powerful A.I. algorithms.
And that’s actually what I did,
when I came back
from Silicon Valley to Italy.
I was dying to learn
all this knowledge and apply it.
So I went deep down
into the rabbit hole of online courses,
and I really wanted to apply
all the knowledge.
But there was one tiny little problem.
I was unemployed, so, you know,
there was no company that would
give me a fun project to play with,
but instead of waiting for someone
to give me something to work on,
I decided to come up with my own projects.
So there was a company
in the United States
who had used A.I. to analyze
the energy consumption data
of millions of Americans
to help them save energy.
And they had been acquired
for 500 million dollars.
So they were, you know, pretty successful.
And remember at the time,
I wasn’t an expert about A.I.,
but I was an energy engineer
so I could really understand
the context of that application. Right?
So here’s what I did.
I took some free
energy data from the Internet.
I took one of the few A.I.
algorithms that I knew,
and I started to try
to replicate the project.
Then I wrote a blog post that I titled,
I Reverse Engineered a 500 million
dollar A.I. company in one week.
Here’s the full story.
Posted it on my blog
and went to sleep.
The next morning, I woke up.
Picked up my phone and it was burning.
Overnight, the article had gone viral
in the United States.
More than 100,000 people had read it
and my inbox was flooded with emails
from people that wanted to meet me,
to thank me for the inspiration,
some wanted to give me jobs.
I even found out that Jack Dorsey,
founder of Twitter, had read the article!
One of the emails
was from a guy called Alex.
Alex was the head of data science
at a Silicon Valley company.
They were also trying to use A.I.
to optimize energy consumption.
And Alex offered me a job
and I started having quite a lot
of imposter syndrome,
You know, getting my first
job in Silicon Valley.
So I told Alex I was like: ’Hey, Alex,
I appreciate the offer,
but I’m not an expert in A.I.′
His answer has been
a defining moment for my career.
Alex told me that he didn’t care
about how good I was in A.I.
What he valued was creativity.
And a burning passion to try to solve
these problems that we both cared about.
Alex told me that A.I. is just a tool
waiting for people with good ideas to use.
Since that moment,
my career, my work life
has been dedicated to trying to spread
this message as much as possible.
Just think about this: ’How many people
in the world have great ideas
but don’t know that they can use A.I.
to turn these ideas into reality?′
A.I. is a tool.
It’s no different from, say, a hammer.
If you take a hammer
by itself, it’s useless,
but give it to a carpenter
and it can be used
to build a beautiful house
for your family.
Give it to an artist
and it can be used to create
an inspiring statue out of pure marble.
All of you here, all of you
care about something,
it may be climate change,
it may be social justice,
animal welfare, whatever it is,
A.I. can be a tool that you can use
to turn your ideas
and your dreams into reality.
You see a lot of people say
A.I. is changing the world.
I don’t think that’s what’s happening.
What is happening is that people
who use A.I. are changing the world.
It’s people who are in the center of this,
so if A.I. is just a tool,
then the question becomes:
‘What are you going to build with it?’
Thank you.
(Applause)