Press "Enter" to skip to content

The Challenge of Visualizing the Artificial Intelligence | Mauro Martino | TEDxRiga


[Applause]

ladies gentlemen thank you for inviting

me here in Riga Mario Martino I come

from Cambridge us you know and this is

sadly the wrong way to start a TED Talks

because we first to come here two weeks

ago I decided to to look around what

other people say about Chris Anderson he

just wrote a beautiful book TED Talks

and I read this book and I was not in a

fight I decide to buy all the books

available now for my son in my book shop

close to my house in Cambridge about

these topics and I do that all the time

because I really love books this is a

video that I should for you to just show

to you how many book I just bought in

the last six months so constantly I’m by

book about different topics and you can

see I working out of Big Data and Daddy

of interaction design and there’s so

many book about machine learning and

book by made by laszlo barabasi

William Britain David McLean 3 Kandinsky

Kevin Lynch Horan Tonelli

so many outer that really are part of my

life

Bruno Minari Iowa study eminent in a

protective of million so why a book are

so important I think if you look at my

collection of books in the last six

months you can see some topics emerge

like that book the reality machine

learning artificial intelligence the

book they talk about user experience and

information design graphic design and of

course this is my favorite collection or

the column regarding the field of data

visualization so I come from the field

of data visualization and if we look a

little bit deeply inside we can we can

see in every single book a story that

are very well done about other no

statistical model computational analysis

anyway inside to every single book three

something magic that we don’t really see

so well so what is this this is just an

analysis of the first page

the book of why information growth from

my friend this is a lead algo and I

think it’s very appropriate to topic so

why information girl and you can see so

many details emerge from the knives of

just this page and we can observe some

emotional analysis and we can see there

is a lot younger discussed and a lot

emotional range and it is very normal

because this is introductional page

where the outer are to really present

deeply emotional is the topic that

you’re talking about this is a way the

computer of looking the book so a big

collection my big collection of book is

this for my computer and for your

computer to see if you use the right

hacker it so a lot a lot of content that

the system highlight because are

particularly interesting and there is a

interesting way to observe with a high

of natural science of this connection

between topics and many type of entity

you can see all this connection emerge

spontaneously because the computer

recognize all this content automatic

sometimes to book a related because

there is the same topics in a different

multiple topics in common it is very

obviously to find a location

organization company and of course a

people but if you and one person can

easily identify connection between a

book to another but if you’re thinking

about the thousand thousand convention a

connection that can emerge but this

analogy you can see the spontaneously

some cluster emerges and the book will

start to be together close each other

because they have many entity in common

this is only possible when you have a

intelligent algorithm

identify different type of entity inside

without artificial intelligence without

this machine learning system without

language processing we are not really

able to ever the big picture about the

interconnection of topics that emerge

from my book collection

and I think this is a very common

problem for everybody not just for me

okay for me it’s particularly maybe I’m

too too much obsess about this behavior

to buy a book and read the paper every

time there to resolve some technical

problem but you know we are full of our

brain doesn’t work really in a way that

we dream sometimes we are some bias and

cognitive bias and we use artificial

intelligence maybe we can resolve this

bias that system is samp

like if we’re looking this guy we are

looking we are able to image Andromeda

Cassiopeia or this beautiful great

picture that I’m just a projection of

our fantasy because we’re really good in

identified pattern but too much good we

find pattern everywhere silhouette path

had been exist so we have so many type

of bias sometimes we forget that we

project on top to the other people that

are in front to us our personal emotions

so I’m feel good today and feel

optimistic and I’m able to see this a

positive characteristic around all the

people around to me so so many aspects

of our life really match with our

capability to absorb information from

outside interpreted this information but

so frequently this interpretation is

wrong let’s do an experiment together to

see if we are you’re able to not to fall

of this player and the wearing white

t-shirt and count how many times they

passes a basketball each other so let’s

see together the video count because

after we’ll ask you if you find the

right number so it’s very old experiment

but still I think is very interesting to

see so the basketball is is a okay

maybe the video is also the color not so

vivid but so how many times guess a

number 16 is the right is 15 so very

close but the point did you see

gorilla everybody so together yeah many

of you especially if you watch this

video for the first time so I’m so still

impressed about these experiments

because the first time they saw this

video I was completely unable to see the

gorilla was two times I said oh my god

there is a gorilla there but if we use a

machine to look in this video there is

no problem because we can activate some

simple system that identify is there a

woman these are men in the video and

when the usability analysis to identify

the gorilla

it doesn’t care it is not focus in the

goal so the machine is just a huge

collection of algorithm so and when the

gorillas live that’s it the gorillas

live so the point is okay maybe they go

realize very easy task but let’s see

something more complicated like a few

weeks ago I was in San Pittsburgh and

his it’s just that few minute walk on

the street and you can see how many

detail them and the computer is able to

identify in this case the algorithm to

identify people so but I have a same

question for you many people are in the

street and you can’t of course not

it’s boring we don’t want to do and many

people are addressing a blue t-shirt I

don’t know

so there are so many inform information

they tell the demerger around o our

world every day every second that we

don’t care about because you know we are

able to resolve a task we are able to

talk about a person but we cannot count

on this information especially we don’t

really care but maybe so many

interesting things emerge from this

translation of this information in a

data so this instructor information

there video at our text from our daily

work no we produce all this amount of

information and we love this information

we don’t really use it and maybe

something can emerge from use this

information in my daily work I’m

interested in a health care problem so

at the moment of what I am working what

Sun health so one of my problem is to

try to identify from a

collection of a scientific paper is

there some interesting correlation

between some gene tp53 and some disease

diabetes or gt350 3 and this is in your

plasma so in what way by using so I want

to give you a very pragmatic example and

what we are using all this amount of

data that I can extract by using these

technology so let’s see let’s see an

example of my daily work so I build in

this interface so I’m able to say that

these are disease autoimmune disease is

a related to this effort gene and if I

look inside discover this gene this is

related with these three gene that is so

before so it’s very interesting starting

point so if I look in this gene and I

try to understand the relation with is

this gene and the protein they are

related with this gene maybe I can start

to find a drug that is able to in some

way interact with this gene so let’s try

diesel interface where you can try

different type of drug available in the

market chi splitting is a type of drug

that really is related with this type

with some protein disconnect with this

gene and what is a result the result is

I can find in a direct way a connection

between a drug and a way to interact

with this gene and maybe this will be

the starting point to resolve this type

of disease

okay it’s very complicated I won’t talk

too much about this type of problem with

you but let’s find another sample they

may be more enjoyable something related

with with us what about explore and TED

Talks

no it’s like I’ll talk in a TED speech

talk about TED speech I like this type

of joke so let’s use it at talk like a

database to Splore together so

essentially if we sprawl this TED talk

we can see I probably if you’re here you

are a passionated about TED Talks so you

know the many talks are in some way

linked each other they talk about this

in the same topics or the outer is come

from the same field so this is interface

that we build where you can see a big

knowledge map all the information that

come from that the TED collection so

you can navigate a little you can see

there is some clusters and talk across

each other because they talk about the

same topics and here is a big list of

the most important topics to emerge

ranking in function to the the frequency

of this topic you know this collection

so every time you touch a topic you can

see the cluster emerge in the map but

this is a more friendly way with to

interact with this data like just ask

something very natural let’s make this

question a question like what is the

relation between money and happiness I

think is interesting topics and there is

some result so there is a collection of

for video let’s see just the first two

you spend that money on consumer goods

which you hope you’ll enjoy using but

then the money’s gone you have to work

hard to get more spend more and to

maintain the same level of happiness

it’s kind of a hedonic treadmill you

never get off and you we see it in

religions and self-help books that money

can’t buy happiness and I want to

suggest today that in fact that’s wrong

and that so in this interface we are

different component like we can avert

overview of the all the top of the

content of the speech but the center is

very cool we have a timeline where we

can explore other entity and top is that

this author was talking about during

whole speech so we can activate a new

way to jump from one talk to another by

using this dynamic topic trended

timeline and you can understand the

power of this type of technology in many

other field not just using this type of

information so in some way the dream a

few years ago from the 50 to build this

machine able to talk with us to answer

to our question is close to be done

probably we will never go to the point

where the machine will directional test

ran on so soon but we’re there and okay

we cannot be too optimistic honestly

there’s still so many technology to

build but the question answer is in the

right direction and I want to close my

speech by

deeper inside in this technology the way

to if you image a good way to really

create a connection between us and

machine is probably this type of a

question and answering a SKU something a

natural language you tell me something a

natural language so IBM is a pioneer in

this field so you remember probably a

few years ago or the job party game is a

very famous game in u.s. do you know

this game someone who you know this game

okay

few of you essentially is like a twist

as that user the traditional type of

games there is a answer you have to find

the question you can see Watson was the

winner obviously otherwise I was not

here and using this example so was a

winner and but the problem is if you

have this question so in 1898 the

Portugal celebrated the 14th anniversary

of these explorers our arrival in India

so the answer can be this so one

supporting evidence can be this because

we find inside the word the Portugal

that Portugal Portugal the India India

and in May that is May so the machine

working this way imagine possible answer

with a possible with a with a new

question but the point is this is wrong

answer is no Gary the right answer is a

Vasco de Gama why because you have to

build something more

you cannot just match this keyword you

have to use a system to match the data

the timing so that a match technology

you have to create these Jo technology

so now we know the India is kept at

beach now we know that this timing is

this timing and the Machine now is able

to give you the right answer so it’s

only mad there is no spirituality there

is no relief nothing more than just mad

so this is the end of my speech I won’t

just tell to you that to go now we are

in the Holi of semantic so we are able

to identify topics but we can we are so

far away to be an area of narrative with

the machines really able to understand

what we are writing our way to

communicate

but let’s forget that is a nightmare to

have a machine like that what is the

reality today there is a machine able to

work with us because understand our

semantics is able to support us in all

our daily work activity or hobby and

let’s use this technology because it is

the best way to really understand all

the book that I buy every six month and

all the book and paper and content that

you read every day thank you so much [Applause]

Please follow and like us: