🚀 VivaTech 2024 : Keynote - AI for Good

Summary

このスクリプトでは、テクノロジー界において大きな成功を収めた一方で、責任を負うべき問題にも直面していると語り、AIの善用が社会の難しい問題解決への鍵になる可能性を探求します。AmazonのCTOであるWerner Vogelsが、過去3,000年間の哲学者から現代の革新技術までを振り返り、AIが持つ可能性とその社会的影響について語ります。また、テクノロジーを通じて世界最难問題を解決する企業の事例を紹介し、AIが持続可能な社会を実現するための重要な役割を演じるべきだと主張しています。

Takeaways

• 🌟 テクノロジーの成功と規模は大きな責任につながります。テクノロジーを良い目的に応用することが重要です。
• 🔮 ジェネラティブAIは文化に敏感になるよう進化し、健康技術、女性向けの技術が発展すると予想されています。
• 🛠️ AIコーディング支援は開発者を助けるために存在し、より迅速に作業を進めるのに役立ちます。
• 🎓 教育はテクノロジーの進化に合わせて進化する必要があります。
• 🤖 古代ギリシャ哲学者から人工知能の思想は長い歴史を持っており、現代の自動化技術につながっています。
• 🧠 脳は思考と哲学を制御すると考えられていましたが、コンピュータの登場と共にそれらの機能を模倣するAIの研究が始まりました。
• 🌐 テクノロジーは持続可能性を確保しながら新しい技術を提供する必要があります。新しいタイプのチップやプログラミングインターフェースがそれに貢献しています。
• 🌱 AIは今現在も多くの問題を解決するための有効な手段として機能しており、ジョン・マッカーシーの言葉からインスピレーションを得ています。
• 🌍 世界的な課題に対するテクノロジーの応用が求められており、特に若い企業が積極的に取り組んでいます。
• 🐟 食糧不足や医療アクセスの確保など、持続可能な開発目標(SDGs)に関連する問題に対するテクノロジーの解決策が重要です。
• 🚑 テクノロジーは医療分野でも重要な役割を果たしており、遠隔地での医療支援や医療機器の開発に貢献しています。
• 📈 データはAIを通じて良い意思決定を行うために不可欠であり、オープンデータの提供はテクノロジーが社会貢献をするための鍵です。
• 🛡️ AIは倫理的にも良いデータと共に働くべきであり、テクノロジーを通じて社会の最も困難な問題を解決する力を持っていることを示しています。

Q & A

スピーカーは誰を紹介していますか？

スピーカーはAmazonのCTOであるWerner Vogelsを紹介しています。

スピーカーが言及した「AI for good」とは何を指しますか？

「AI for good」とは、人工知能を用いて社会の難しい問題を解決しようとする取り組みを指しています。

スピーカーが述べたジェネラティブAIの進化のステップは何段階ありますか？

スピーカーはジェネラティブAIがマラソンの最初の3ステップにしかいないと述べており、これからの影響を示唆しています。

スピーカーが提唱する「AI for now」とは何を意味しますか？

「AI for now」とは、現在利用可能なAI技術を活用し、世界最难関の問題を解決するアプローチを意味しています。

スピーカーが紹介したユニセフが取り組んでいる問題とは何ですか？

スピーカーはユニセフが農業における小さな農家の問題に取り組んでいると紹介しており、彼らが銀行からローンを得られない問題や収穫量のデータ管理システムを構築していると説明しています。

スピーカーが挙げたAIを活用した農業における成功事例とは何ですか？

スピーカーは、国際米種質研究所がAIを用いて米の品種を管理し、バックログを排除する成功事例を挙げています。

スピーカーが紹介した健康ケア分野におけるAIの活用事例には何がありますか？

スピーカーは、ワクチンを離島などに自動ドローンで届ける取り組みや、医療現場でのAIを活用した予測モデル、新生児の脳損傷を早期に検出する技術などがあります。

スピーカーが強調したデータの重要性とはどのようなものですか？

スピーカーは、AIが良好な分析を行うためには良好なデータが必要ですと強調しており、データはAIが機能するための重要な資源であると述べています。

スピーカーが紹介した「Tech to the rescue」はどのようなプログラムですか？

「Tech to the rescue」はテクノロジーを用いて社会貢献を目指すプログラムであり、スピーカーはこのプログラムに関連してCTOフェローシップを発表しています。

スピーカーが引用したJohn McCarthyの言葉とは何ですか？

スピーカーはJohn McCarthyの言葉を引用し、「それが機能するようになると、それをAIと呼ばなくなる」という意味を持ち、AI技術が成熟すると一般には認識されなくなると述べています。

Outlines

00:00

🌟 テクノロジーの責任と将来予測

スピーカーはテクノロジーの進歩とその責任について語り、2024年の予測を紹介します。テクノロジーは成功を収めてきた反面、広範な責任を負うべきだと指摘。AIの応用について語り、特にgenerative AIの進歩について触れます。また、教育の進化やテクノロジーの持つ可能性についても言及しています。

05:02

🤖 AIの歴史と現代の技術動向

AIの歴史を振り返り、古代ギリシアの哲学者から現代のAI技術までを概説。過去数十年間のAIの進歩と、特に最近のTransformersやファクターデータベースなどの技術革新について解説。AI技術が世界を変える可能性についても触れています。

10:05

🌱 テクノロジーを活用した持続可能な開発

スピーカーはテクノロジーを使い、持続可能な開発目標(SDGs)を達成するビジネスの例を紹介。特に、農業におけるテクノロジーの活用と、小規模農家への金融サービスの提供について語ります。

15:07

🍚 食糧生産のテクノロジー

世界食糧問題にテクノロジーがどのように貢献できるかについて説明。特に、米の研究と栽培の最適化、農薬の使用の最適化、そして養殖業におけるデータ分析の活用について詳述しています。

20:08

🏥 テクノロジーを活用した医療へのアクセスの拡大

医療へのアクセスを拡大するテクノロジーの活用について話します。ワクチンの自動ドローン配送、医療費用の削減、新生児の脳損傷の早期発見など、テクノロジーが医療分野で果たす役割について解説しています。

25:10

📊 データの重要性とテクノロジーの役割

データの重要性とテクノロジーがそのデータを活用する上で果たす役割について語ります。データの共有と開放性、そしてAIがデータから価値を引き出す方法について解説しています。

30:12

🛡️ AIによる社会問題への取り組み

AI技術を用いて社会問題に取り組む方法について話します。特に、子どもの性虐待や人権侵害に対処するテクノロジーの活用について詳述。AIが社会貢献に向けて重要な役割を果たすことについて強調しています。

🌐 テクノロジーを活用した社会貢献への道

テクノロジーを活用して社会貢献を目指す道について話します。AI for Change MakersプログラムやCTO Fellowshipの紹介を行い、テクノロジーが持続可能な社会貢献に向けて重要な役割を果たすことを示しています。

Keywords

💡テクノロジーの責任

テクノロジーの責任とは、テクノロジーを利用して社会の様々な問題を解決する責務を指します。ビデオでは、テクノロジーが成功を収めてきた一方で、持続可能性や社会問題への対応など、テクノロジーが担うべき広範な責任について議論されています。特に、人口増加による食糧不足や医療へのアクセスの不平等といった問題にテクノロジーを適用する可能性が示されています。

💡生成的AI（Generative AI）

生成的AIとは、新しいコンテンツを創造的に生成することができる人工知能のサブ分野です。ビデオでは、生成的AIが文化的意識を持つようになると予測されており、その進化が今後のテクノロジーの進歩に重要な影響を与えるとされています。また、AIコーディングアシスタンスが開発者の仕事の効率化に寄与するとも触れられています。

💡文化的な意識

文化的な意識とは、個々の文化背景や価値観を理解し、それに基づいて適切な行動を起こす能力です。ビデオでは、新しい大きな言語モデルが文化的意識を持つようになると予想され、それがAIの文化的な適応性や多様性に寄与すると示唆されています。

💡テクノロジーの持続可能性

テクノロジーの持続可能性とは、環境保護と経済発展のバランスを保ちながら、長期的な視点でテクノロジーを開発・利用する考え方です。ビデオでは、新しいタイプのチップやプログラミングインターフェースが持続可能なテクノロジー創造にどのように貢献するのかが語られています。

💡AIの教育への影響

AIの教育への影響とは、人工知能が教育システムに及ぼす変化を指します。ビデオでは、教育がテクノロジーの進化のペースに合わせて進化する必要があると強調されています。これは、教育者がテクノロジーを活用して新しい教育方法を創造する可能性を示唆しています。

💡社会影響ビジネス（Social Impact Business）

社会影響ビジネスとは、企業活動を通じて社会問題を解決し、益利を上げることの両立を目指すビジネスモデルです。ビデオでは、農家に対するアイデンティティの提供やデータの活用によって、農家が持続可能な生活を送れるようにする企業例が紹介されています。

💡人工知能の歴史

人工知能の歴史とは、人工知能技術の発展とその背後にある哲学的な思想の変遷を指します。ビデオでは、古代ギリシャの哲学者から現代のAI技術まで、人工知能の発展に寄与した人々とアイデアが振り返られています。

💡自然言語処理（NLP）

自然言語処理とは、人工知能が人間の言語を理解し、生成する能力を指します。ビデオでは、自然言語処理がテクノロジーを活用して社会問題を解決する上で重要な技術となっていると強調されています。

💡データの民主化

データの民主化とは、データにアクセスする権利を広く持たせ、情報の平等な利用を促進することを指します。ビデオでは、公開されたデータセットがテクノロジーを活用して社会貢献にどのように役立つかが語られており、その重要性が強調されています。

💡人工知能による医療への貢献

人工知能による医療への貢献とは、AI技術が医療分野での問題解決にどのように役立つかを指します。ビデオでは、遠隔地帯へのワクチンの配送や、医療問題の早期発見・介入など、AIが医療に与える影響が紹介されています。

💡テクノロジーの持続可能性

テクノロジーの持続可能性とは、環境保護と経済発展のバランスを保ちながら、長期的な視点でテクノロジーを開発・利用する考え方です。ビデオでは、新しいタイプのチップやプログラミングインターフェースが持続可能なテクノロジー創造にどのように貢献するのかが語られています。

Highlights

Tech Guru's role in Amazon and predictions for 2024 including culturally aware generative AI, advanced health technologies, AI coding assistance, and evolving education.

The importance of technologists taking responsibility for solving the world's hardest problems with technology.

John McCarthy's quote on AI and the evolution of generative AI's cultural awareness.

Aristotle and Plato's early discussions on automation and humanoid robots 3,000 years ago.

The shift from symbolic AI to emulating human senses with technology as a path to success in AI.

Amazon's fulfillment centers utilizing 175 autonomous robots that enhance operational efficiency.

The transformative impact of underlying technologies like Transformers and factor databases on generative AI.

The early stages of generative AI's impact compared to the marathon of technological evolution.

The necessity for sustainable technology development, including new chips and programming interfaces.

AI's current and potential applications in solving global challenges, such as population growth and food security.

The UN's sustainability development goals as a framework for segmenting technological efforts for social good.

Startups like Hara providing identity and data to smallholder farmers for economic sustainability.

The International Rice Research Institute's use of AI to manage rice diversity and reduce backlog in seed storage.

Precision AI's application in agriculture for efficient and environmentally friendly crop management.

Aquabyte's use of computer vision to monitor fish health and contribute to global food sustainability.

The importance of democratizing data access for tackling global issues like healthcare and natural disasters.

Foreign's innovative use of AI in combating child sex trafficking and abuse through image and data analysis.

The symbiosis between good AI and good data, emphasizing the need for quality data for effective AI solutions.

The launch of the now go build CTO Fellowship to support innovators in technology for good.

McCarthy's quote on AI as a reminder to focus on current technologies' practical applications to solve today's challenges.

Transcripts

now you may not want to run away so fast

because we have

brought a very very inspiring Tech Guru

he plays such an important role in

Amazon so our next speaker is a science

research an expert in computer science

and even a tech startup founder today

he's one of the key Innovation figures

behind Amazon as I was saying and he's

well known for his predictions what did

he say for

2024 well new large language models will

make generative AI culturally aware fem

Tech will take off with Advanced Health

Technologies for women that's something

I'm certainly looking for and waiting

for AI coding assistance will help

developers work faster by taking care of

he heavy lifting and fourth ucation will

evolve to match the pace of evolving

Tech we missed him last year but he was

finally able to come this year please

allow me to

welcome verer Bogel CTO of

[Applause]

[Music]

[Applause]

Amazon good afternoon Paris um um

I think if we look back at the past two

three decades we as technologist and as

u in digital technology but also in any

type of innovation we've had great

success massive successful companies

millionaires billionaires but I do

believe that with that success and scale

comes a really broad responsibility at

the same time if you look at the issues

that we're facing in the world today we

as technologists have a responsibility

to try to solve a number of the hardest

problems that we see in this world and

yes of course you you think applying

technology for good uh these days of

course you immediately start to think

about how can we apply AI for good yeah

and there's a huge amount of excitement

around generative Ai and things like

that but I would like to hold there and

actually go back to a quote of one of

the famous founders of artificial

intelligence John McCarthy who said as

soon as it works we don't call it AI

anymore there's a real big body of work

out there that is actually AI but that

we don't call AI anymore because we only

talk about AI in terms of looking

forward kind of things that we could be

doing and more importantly I think it is

also to realize that there is a massive

body of work that works really really

well we just don't give it that new

stamp of AI that AI that we've seen come

to life in the past say year and a half

yeah and it's quite successful but I

think to really understand this we need

to go back in time actually quite far in

time 3,000 years ago Aristotle and Plato

were already discussing about the use of

intelligence to solve particular tasks

by

Automation and actually even to the

point that play in the Republic actually

described a number of humanoid robots

that were performing household tasks

remember this is 3,000 years ago that

they were already sort of envisioning

what we actually have been coming to

life in the past 10 20

years now most of those philosophies and

if you actually go back to the past

2,000 years those have who have been

thinking about this artificial

intelligence

they all still were thinking about that

everything should be controlled by the

brain yeah and that actually any

thinking and any philosophy about that

is actually driven by the brain now

brain was actually just they thought it

was actually manipulating symbols so if

the first arrival of computers who were

also manipulating symbols we really got

to the point immediately to start to

think about can these machines actually

think yeah alen touring probably our our

Century's most famous uh computer

science

philosopher really had this first

question of can machines think and

actually the famous touring test is made

by Ellen T can you distinguish answers

from machine from those are given by a

human actually the word artificial

intelligence at this moment doesn't

exist yet it is for the first time

coined in 1956 in a workshop in dmth

with McCarthy there Minsky and quite a

few others but still they were thinking

that artificial intelligence should be

something driven from the top down

basically thinking about sort of really

how does the brain

control the rest of the body that didn't

go anywhere yet I think eventally

symbolic AI as we would call it results

in automated reasoning and many of the

other things that are really successful

areas now but that was not what drove

the success of AI in the earlier days

what drove the success was really

thinking different by the way one of the

things that we were building in those

days and yes that is me I had more hair

in those days I build a few of those in

the 80s using Prolo but what really

started to make a major change is the

people that were actually working on

robotics they were saying you know what

why don't we do it top down from the

brain down but why don't we pick

individual senses of humans and try to

emulate those in technology yeah Vision

sound speech all of those and you really

see that we have made great strides

there because that actually was a path

to

success this is a fulfillment center of

amazon.com we have about 175 of those

around the world and in each of those

you may see about 30,000 robots running

around and we can only do this because

they can be autonomous and they can

understand different senses not maybe

the robots are not necessarily speech

enabled but they definitely have liar

and radar and other senses that we as

humans even don't

have now what we've seen over time is

that we got all these different steps in

technologies that slowly advance and

more and more of the underlaying

Technologies Drive the kind of things

the solutions that we can build with it

and of course the last one the most

recent one is that of Transformers

together Factor databases which Drive

the generative AI world at this moment

now everybody's talking about large

language models and I will skip that for

the most part because I think many other

people will be talking about that

although it is actually getting a lot

closer to Plato's dream and the

assistance that it can give in efficieny

and areas like that is

unparalleled I do believe generative AI

is just at the first three steps of a

marathon in terms of impact on the

solutions that we're building it's very

very early days and I also think that is

not the end of the implications of AI

we'll build newer underlaying

Technologies which makes other advances

again possible which ones those are

we'll need to see in the coming three to

four years because that actually the

cycle in which we see underlaying

Technologies improving together with the

hardware that runs in lockstep and

especially the hardware part is

extremely important because if you are

concerned about technology for good you

also need to make sure that your

technology is created in the most

sustainable way possible yeah new types

of chips new types of programming

interfaces will allow us to be

sustainable and at the same time

actually deliver new

technologies so I want to go to the area

of what I would call AI for now ai for

now is a massive body of work it

actually all works which like John mcari

we then say you know what we don't call

it AI anymore maybe maybe the infers of

uh mti's quote is that we call it AI if

it doesn't really work yet yeah but if

you think about all that work that has

been done you know can we actually use

this technology to solve some of the

world's hardest

problems now so what are those I mean

the present is really important let's

take all the technologies that we have

that we've created and that make many of

our businesses successful but at the

same time we need to take the

responsibility to use this technology to

solve some of the world's hardest

problems and I wasn't really aware of

that until you know quite some time ago

when I realized that you know some of

the numbers that we're going to see in

this world the growth of our population

is disastrous for the current Ser

circumstances you think about that in

the coming 20 to 30 years we'll see the

population grow by another 2 billion

people that's an increase of 25% in the

population how are we going to feed them

I'm going to make sure that they're

economically sustainable problems that

even at this moment we already have so

what can we do in terms of tech

technology and I had been thinking about

that for quite a while how to organize

that and I think the UN sustainability

development goals are a right way to

sort of segment the kind of work that

has been going

on and so there's these are the goals

set by the UN but let's pick out a few

of them and look at how young businesses

especially are trying to solve these

problems I'd be very fortunate indeed in

the past 10 years 10 15 years to travel

the world and meet many startups around

the world one of the things that I

started to see is outside let's say the

main master world that many splots were

not looking at becoming the next unicorn

they were really looking at solving some

of the world's hardest problems and very

fortunate at some moment to take a TV

crew along to uh create this TV series

called now go build that actually

highlights these companies that solve

some of the world's hardest problems

that you can find it on Prime video and

on aws's website and and on YouTube but

you know if you're interested in

companies that are really solving hard

problems this is a really great TV show

but it also motivated me to look at sort

of how can we use technology for

good so a really great example is this

actually one of my first episodes that I

did was with a company called Hara out

of

Jakarta and if you think about Indonesia

or most of Southeast Asia where there

are many small hold of farmers

none of them have an identity that means

if you don't have an identity you can't

go to a bank for a loan you have to go

to the Village loone shark who charges

60% by the way yeah to get some money to

buy your seeds for your next cop it

means that half of your crop is already

sold gone out of your hands before you

even start growing

it so they built this system that gives

um Farmers an identity measure their lot

of land now measure the yield of that

land and then keep that data together to

actually make that available to

organizations that are interested in

this data the first organization of

course that is interested in this are

the banks and why is that is because

these these Farmers they're not looking

for a million dollar loans they're

looking for 10 to $100

loans and each of those actually have a

100% repay rate so by just make doing

this by giving this farmers in identity

by associating data with them about sort

of the yield of that plot of land all of

these financial institutions are eager

to support these Farmers something they

would never had before so now they can

actually have a sustainable living why

did this company actually started

tackling this problem because they were

children of small holder Farmers

themselves they understood the problems

that their parents that their

communities are going through and

decided to actually take action and

build a business out of that now this is

maybe what we call a social impact

businesses actually not just you're

trying to do good but have a good living

and a good business at the same time and

this is possible because it's not the

farmers that actually are paying har it

is the BS and the government

institutions that want to have access to

the data that are actually making this

business very

successful so if you think about one of

the biggest problems that we'll have

with two billion more people on this

planet is access to

food and if you look at sort of the most

important staple of food is out of rice

well over 50% of the planet is dependent

on Rice as the main food uh

Source there's an organization in the

Philippines called the international

rice Research Institute they are

absolutely brilliant if you want to know

more about sort of food research and

things like that absolutely look up what

these guys are doing they have the

largest collection of rice DNA in their

freezers 200,000 strands of different

types of rice by the way they're also

backing it up in the north in Norway in

very cold areas now their biggest

challenge is actually taking in all of

these rice and then uh sorting them

looking at which R seats are actually

useful or not and that is actually such

a high level that there's a huge backlog

in actually getting these seeds into

storage this is one of the simple

problems that you can easily solve with

with uh object recognition just use a

small video camera train it and see

which wise uh seats can actually be

useful or not the longer these seats

actually kept out of storage the bigger

the problem is for them because they

have to discard the

R so very simple system little camera

determining which R seats are are are

good and immediately by installing this

system which is not a magic system but

uses Rock Solid proven tech technology

they immediately eliminated all of the

backlog and they do a lot of research

there many different types you have to

imagine that most of the farmers that

they're targeting actually don't have

cell phones they can't read or write

often they do can speak into a phone

though so they developed a system for

which farmers can actually call into

describe their patch of land and where

it is and then the system will give him

advice about how much fertilizer to buy

and when to apply it because most of

these Farmers have grown up with not

much let's say scientific knowledge

about how much fertilizer to use they

just buy a lot and apply it to their

land the same goes for

pesticides and so here the key is that

you're able to actually use voice as an

access to digital systems remember

there's a very large part of this world

where reading and writing is not common

and as such if you want to give those

people access to digital Technologies

voice is critical in all of

that if you think about indeed about

planting you know most of uh not only

fertilizer but most of the biochemicals

used to actually protect the plant are

often just being sprayed yeah Precision

AI is one of these companies that

actually uses drone with very high

quality imagery to pick out individual

wheat plants in rice plantations so that

they can take them out and do

individualized Precision management of

these crops instead of just spreading it

over everything the runoff often both of

wellas um uh fertilizer as well as these

biochemicals often run off in the rivers

and actually create massive pollution

and creation enormous amount of algae

for example in these Rivers

now that's rice yeah so what about

protein we kind of all like protein and

it's an important part of our

diet

right and so definitely if the world is

growing how can we make sure that 25%

more people have access to

protein the most efficient way to do

that is to actually create fish because

one kilo of fish feed results in one

kilo of fish

if you want to do it with cattle you

have to give it seven kilos of feet

before you create one kilo of

protein and not only that cattle farming

has had a disastrous impact on many of

our surrounding

areas so there's a company uh in Norway

where I met them first called

aquabyte and they make they have these

massive pens in the fs in Norway in each

of those pens there is about 200,000

salmon in there

and they get in there when they're

really small and what they do they use

computer vision and object detection to

track the individual fish their growth

but more importantly their health

because one fish with lice in that pen

will infect you other 200,000 and you'll

have to destroy them all and so they

built this system this Vision detection

system together with things like sine

levels and other iot levels to actually

keep this fish healthy and so this might

be one of the approaches that are being

taken to see whether we can at scale

grow sufficient protein to solve sort of

the upcoming disaster of not being able

to feed everyone in this

world they have created these massive

data stores already about how to

actually sort of track fish and make

that available

worldwide so if you look at some of the

other

um areas in the sustainability goals

it's of course healthare how can we

ensure health care for 25% larger

population if we cannot even do it

today we as technologists are the ones

that actually should stand up and

actually find solutions to solve the

healthcare problems in this world or

access to

healthcare now if you uh think about at

this moment at 202 one I think is a

report by the World Health Organization

who estimates that about 2 billion real

populated two billion people living in

rural areas do not have access to health

care at all things that we consider to

be just a fundamental right to hex to

it uh suero is one of these companies

that are trying to solve how to get

vaccines into rural areas these drones

they're not piloted they're completely

automated drones are one of these

examples of oldfashioned a just working

really well a drone will easily have

about a 100 Sensers on it yeah lighter

radar Vision you need to be able to

detect obstacles in flight because

remember these these drones need to be

autonomous there is no there's nobody

sitting at the the joystick to get this

to a place where you want and then when

you arrive somewhere you know you don't

want to hit the dog that are there nor

do you want suddenly this fishing line

that actually was was out there and want

to run into that so a lot of autonomous

work using AI is being done to make sure

that we can deliver vaccines into areas

that have never actually reached any

seen any healthc care professional at

all and if you look at that I think it

was about a few years ago that I went to

Brazil and in Brazil still you know

there's about 200 million citizens and

they consider that about 150 million of

them are medically homeless Dr Kila uses

AI Technologies to actually first of all

provide very lowcost Health Care by

looking at where the places in the world

they can buy generic medicine for the

lowest cost but also make use of AI over

patient records to start to predict what

kind of health problems people may

exhibit and try to intervene at a much

earlier phase

there

CICS isra in an Ireland is very

interesting company it actually turns

out that about was it one in every 200

newborns have some degree of brain

injury and these brain injuries are

often not detected until months or years

after birth yet with a very simple test

they can give a that we now give babies

a test whether they can hear well

whether they can see well by just

placing a little cap on their head with

in

EEG and then actually can detect these

brain uh these brain damages very early

and immediately start to try and

actually tackle those particular

problems yeah so they make use of EG

data now why is that so important why

can't we just do it regularly it turns

out EEG data of babies is radically

different than that from us as uh as

adults now in quite a few examples I

gave you computer vision plays an

important role and why because it makes

nicer slides but there's a whole range

of other Tech technologies that work

really well and are the foundation of

how we actually applying AI for good

where it's natural language processing

speech translation all of these work

well know these are Technologies but you

don't even think about anymore look at

for detection a company like like Amazon

we literally sit on billions of orders

from the past we know which ones were

forland and so we can build a model out

of that a new order comes in it gives us

the score what the likelihood is that

this is also a forent order we don't

kill the orderan actually the order then

goes off to a human

investigator remember that AI predicts

but we make make decisions it is still

not the case that these Technologies

decide for us we are the ones that are

in charge and we are the ones that are

deciding where these tools just help us

get to that

decision now in all of this know we

cannot have good insights driven by AI

if we do not have good data and data is

crucial in all of this now the problem

is often that data is a privileged asset

companies holding them close to

themselves they're not opening it up

although there's a lot of efforts going

on to create more and more open data

sets having access to public data is

crucial for those companies that want to

do good in this

world a good example there is one of the

other now go built episod that I did uh

which was in the Philippines now the

Philippines sit in what is called

typhoon alley they have natural

disasters several time each year but the

significant portion of the country is

what we call unmapped and the question

is even do you exist if your street is

not mapped and why is it not mapped by

the way is because those companies that

create the maps that we use on a daily

basis are only interested in those areas

that they economically viable 80% of the

Philippines do not fall into that

category

so the human open Street bed Foundation

actually tries to tackle that they have

people on the ground going through these

unmapped areas and mapping them by hand

and also indicating especially with back

to earthquakes which buildings are

actually stable and mapping that out the

Red Cross in the Philippines lives by

this data set because that's the only

data set they can get their hands on to

actually Reach people in need during

these disasters

so I honestly believe one of the biggest

challenges we have at this moment is to

make sure that we can

democratize access to this data for good

and many organizations already doing

that AWS the cloud computing arm of

Amazon uh for example has a large

collection of open data uh one of them

is uh it digital Africa this contains

imagery over the years from three large

satellite grips um the one from the US

from NASA and USGS uh there's a European

one from the project Copernicus and Jack

the Japanese uh satellite company also

makes these digital imagery available so

you can look for the deterioration of

mangr you can look for illegal roads

being built because then thaty at the

end of the road illegal mining is

happening so making sure sure that we

have this data available for everyone is

actually crucial in improving access to

data now why is this data so important

why is it different from the past I

think one of the major changes that we

saw let's say in the '90s we already

knew what kind of questions we wanted to

ask and that drove what kind of data we

collected SE queries but with cloud

computing making data storage so cheap

suddenly you could keep all your data

around and then it becomes sort of a

massive heap of data where you looking

for the pot of gold that may be in there

now what is the technology that you use

to find the needle in the Hast stack

that's a

magnet yeah in this particular case to

find the needle in a digital Hast stack

you use AI or machine learning as a

technology that is

crucial now I want to leave you with a

last example which I think is most

telling there's an organization called

foreign is a nonprofit organization that

has as a goal to stay ahead of SE child

sex trafficking and other child sexual

abuse yeah they use image recognition

they use data mining they use

collaborative filtering all these

different techniques to prevent and

detect child sexual abuse

they build a system called Spotlight in

which they have imagery of missing

children from all around the world they

compare it each day for example in the

US against the 100,000 new ads for

escorts that are being released each

day and

Spotlight has been quite successful they

already found 18,000 victims of child

sexual trafficking and rescued and more

than 6,000 were actually very young

children they build a different system

called saver because I assume none of us

here would like to have child sexual

abuse material on our

servers if you have any way for your

customers to upload imagery you may want

to use saver to find those images that

they have a massive database of

uh uh of uh sort of hashes of this

imagery and they can detect this for you

many of these organizations have

actually used humans to do this believe

me if you need to look at these images

just for a day you will need therapy at

the end of the day if we can actually

scale this up by using technology and

actually really find this imagery and

actually nip it when we can that's

crucial they also actually make you

safer is also a tool that can actually

look at communication patterns between

individuals and look for patterns of

grooming these Technologies are crucial

if we want to solve some of the hardest

problems in this

world now good AI needs good data

there's a symbiosis between the

two and that means that good data

absolutely needs good AI to be able to

make sense of it but more importantly

good work needs good people it is us as

technologists with the right mindset to

want to solve some of the world's

hardest problems because we've been

successful in actually solving spam

filters let's make sure that we can

solve a number of problems that are much

harder than that talking about good

people uh this organization called Tech

to the rescue actually launched a

program called AI for Chang makers it's

a global ACC accelerator that focuses on

different areas of tech for good and I

think um they are running this year five

different cohorts in different

areas jumping on that I've been so

impressed by this program that today I'm

announcing a CTO Fellowship the nowo

build CTO Fellowship because I believe

that many of these innovators in this

space can actually make use of sort of

the experiences that I and my team have

as uh as driving technology for so many

years that we can actually help them

with that with this CTO fellowship and

really technology plays a crucial role

in technology for good and to find Mark

more to the fellowship by the way and

for the AI for change makers look at the

uh capture the QR code there so I like

to leave you again with uh McCarthy's

quote Yeah remember that some of the

current technology

is clearly that of what I would call a

dancing bear we really think this we

really are amazed that this bear can

dance we're not really looking at

whether the bear dances well or not yet

for that we need some longer time so as

soon as it works nobody calls the AI

anymore there's a massive body of work

out there that you can use to solve

today's hardest problems now I'll leave

you with this yeah now go build thank

you very much