AI: Toy for the rich or tool for equality? | DW Business

Summary

The transcript discusses AI fairness and access, highlighting perspectives from a Ugandan lawyer seeking to leverage AI to provide legal services to underserved communities, an AI ethics leader from a major tech company on regulating AI responsibility, and an artist using AI art to raise awareness about biodiversity and climate change.

Takeaways

• AI's benefits should be equally accessible globally, not just limited to wealthier nations.
• Regulating company/individual behavior around AI is key - AI itself simply enables what we program it to do.
• Transparent and auditable AI models can mitigate issues like biases and fake outputs.
• Sustainability and purpose should be central when developing AI models and applications.
• AI art can play a meaningful role in education, research and prompting urgent climate action.

Q & A

What perspective does Gerald the Ugandan lawyer provide on AI fairness?

He argues the benefits of new tech like AI should be as available in poorer nations as wealthier ones - fairness is ensuring access regardless of geography.

How does the AI ethics leader suggest governments approach regulating AI systems?

She argues AI itself does not require isolated regulation - rather we need to govern company/individual behavior surrounding AI to uphold existing laws and ethical norms.

How can blockchain technology help address issues with AI models?

Blockchain enables tracking all data in/out of models. If biases/errors emerge, version control lets you audit what happened and revert to a previous unbiased state.

What is the purpose of the AI art museum project announced at the forum?

To create an open-source AI model for preserving and understanding nature's intelligence, helping education and research around AI's connection to the urgent climate crisis.

Outlines

00:00

Perspectives on AI's Impact and Regulation

This paragraph covers perspectives from a lawyer, Gerald Abila, on how to ensure AI benefits people equally across geographic regions. He advocates for guard rails to prevent perpetuating inequality and believes AI can help increase access to justice. He shares how his organization, Barefoot Law, has used technology to help nearly 1 million people in Africa and aims to help 50 million more by 2030 using their AI lawyer Winnie.

05:00

Ensuring Accuracy and Transparency in AI

This paragraph discusses the problem of inaccuracy and lack of transparency in AI models, with 20% producing fakes. Raluca Manolache of blockchain company Calabs explains their solution to audit AI models and track their data inputs and outputs. This transparency allows going back to previous versions if errors emerge. Manolache acknowledges blockchain has been used by bad actors but argues its accountability makes AI more trustworthy long-term.

10:02

Using AI Responsibly to Understand Nature

This paragraph covers artist Refik Anadol's AI installation to safeguard biodiversity. He aims to connect humans, technology, and nature by using AI to remember and preserve nature's intelligence. He shares how the project minimizes its energy use through Google Cloud and educates people on AI's purpose beyond pixels through an AI art museum called Dataland.

Keywords

💡AI regulation

The concept of regulating artificial intelligence technology and its applications is discussed extensively in the video. Experts emphasize the need for appropriate regulation to steer AI in a direction that benefits humanity and avoids risks. For example, Ivana Bartoletti talks about the European AI Act which takes a risk-based approach to regulating AI based on its potential harms.

💡AI transparency

The video highlights the problem of AI being a 'black box' and the need for more transparency in how AI systems work. Raluca Ada Popa discusses using blockchain technology to enable traceability of data flowing in and out of AI models. This could help identify issues like biases.

💡AI accountability

Closely tied to transparency, several speakers underscore the need for accountability in the development and use of AI. Ivana Bartoletti notes that existing laws already apply to AI, and regulation is needed to hold companies and individuals accountable for harmful uses of AI.

💡AI auditing

Auditing AI systems is presented as a way to monitor their functioning and trace problems. Raluca Ada Popa explains how her company allows users to audit what happened in an AI model and rollback changes, like 'version control' in software.

💡AI harms

Many examples are given of potential harms from irresponsible use of AI, including biases, loss of privacy, spreading misinformation etc. Speakers emphasize avoiding harm as a key aim of AI regulation and governance.

💡AI opportunities

At the same time, speakers highlight the many benefits AI can offer humanity, if steered positively. Examples include expanding access to justice, education, preserving nature etc. But these require inclusive development of AI.

💡Inclusive AI

Making development and use of AI inclusive is a common theme. Gerald Abila stresses that AI benefits should be 'as prevalent in the global South as the global North'. Inclusive AI is positioned as an alternative to AI worsening inequality.

💡Nature-inspired AI

Artist Refik Anadol talks about using AI to study and preserve nature, which he calls the 'most intelligent thing we have'. His art uses natural imagery and data to create AI models aimed at fostering appreciation for nature.

💡Responsible AI

The speakers converge on the idea of responsible or ethical use of AI. This includes self-regulation by tech companies as well as formal regulation. Responsible AI considers social impact alongside technological capabilities.

💡AI for social good

In several examples, AI is presented as a technology that can help tackle social problems like access to justice, education, inequality etc. But this requires ethical frameworks and data/compute resources aimed at social good applications.

Highlights

AI benefits should be available globally, not just in wealthy nations.

Barefoot Law has used technology to provide legal services to almost 1 million people in Africa.

Barefoot Law aims to reach 50 million people across Africa by 2030 using AI.

Existing laws like privacy and human rights already apply to AI systems.

The EU AI Act focuses on regulating company and government behavior around AI.

Businesses need to innovate with AI responsibly without harming society.

AI is becoming more prominent at the World Economic Forum.

Some companies are promoting ethical and responsible AI.

Calabs uses blockchain to audit AI models and improve transparency.

Blockchain can help track what data goes into AI models.

Artist Rafik Anadol uses AI art to showcase nature's intelligence.

Anadol's project uses renewable energy and limited data for sustainability.

The project aims to educate and preserve biodiversity data.

Anadol is creating an open source AI model focused on nature.

Experts warn we are nearing an irreversible point of climate damage.

Transcripts

can we trust Ai and its creators I'm Ben

fulan I'm at the world economic Forum in

Switzerland to find out if artificial

intelligence is set to become a toy for

the rich or a tool for equality take a

look at this everchanging AI based

artwork what's it good for though I'll

get to its creator in a moment and talk

to him first let's talk about the

fairness of AI with a loyer Gerald ABA

is from Uganda Rich nations are driving

Technologies like AI at the moment where

does that lead lowincome countries I

guess uh when it comes to new

technologies we should ensure as

Humanity that the benefits be us all

regardless of geographical location the

benefits of new technologies should be

as prevalent in the global South as they

are in the global North and this is what

I consider fairness how how do we do

that I mean it's a a question of

investment isn't it I think it goes

beyond investment it's a question of

will what do we intend to achieve as

Humanity professor profess Claus in his

opening remarks today spoke about

building guard rails when it comes to

the new technologies that we're facing

and I think these guard rails for us

should be guard rails that prevent

against perpetuating inequalities that

we've already experienced as a society

as Humanity while at the same time

amplifying the benefits to make sure we

all achieve uh the potential development

that these new technologies give us how

many people have you managed to help

through leveraging technology in your

work as a lawyer and how many more

people do you hope to help in the 10

years Barefoot law has been operating

we've uh reached close to a million

people and we hope to reach 50 million

people across Africa by 2030 is AI going

to change that yes we've uh been very

interested in new technology from the

founding of barefoot law and we've

invested a lot of time and resources in

developing a new technology that we call

Winnie which is an AI lawyer that

improves on our efficiency and we hope

with the breakthroughs that we're

experiencing now we're able to

amplify these benefits Gerald thank you

very much great talking to you thank you

it's been a

pleasure our lawyer says it's all about

access to Justice especially in remote

areas AI will boost that by answering

simple questions

quickly for more on regulating

artificial intelligence Ivana bti is

Chief AI governance officer at the tech

giant whpr this is a runaway technology

and Regulators are playing catchup will

they keep up with this technology it's a

fantastic question and I think uh the

answer is in the first place AI is

already regulated in the sense that AI

does not exist in isolation privacy laws

human rights legislation consumer

legislation non-discrimination laws they

already apply to artificial int

intelligence so what is missing what's

missing is regulating the behavior of

companies of individuals around

artificial intelligence this is for

example what the European AI Act is

about which is a risk based legislation

and says if an AI is likely to cause

risks to people then it has to be

regulated in a stronger way now the real

issue here is first of all to avoid

thinking that AI is an excuse to really

breach existing regulation and second is

to really build on this alignment that

we have right now that we're seeing here

in Davis between government private

sector saying yes we need some

regulation in place but the regulation

is really about regulating the behavior

it's not regulating AI is really about

regulating the behavior of companies and

business and governments around

artificial intelligence what could be

the impact though on businesses could

this put the breakes on well I think um

businesses need to innov

and it's really really important it

means that the companies need to say

okay how do I innovate and transform

with they ey but how do I do that in a

way that is not detrimental to society

it's not detrimental to consumers in a

way that enhances the Dignity of human

beings and in a way that does not put

people at risk this is really really

important we have to govern change we

are not going to have change just to be

done to us and it's B businesses are

ready to take this responsibilities I

want to see the conversations really

moving forward on this not just in

business in government but also with

civil society well AI was just a

buzzword here in Davos last year this

year take a walk down the prominade and

it's in almost every shop window where

the most influential companies in the

world have set up some are selling good

AI or responsible AI what about accurate

AI

Ral manoha is CEO of calabs a blockchain

company that claims to have found a

solution to ai's accuracy problems and

tendency to produce fakes which is at

what a 20% uh rate according to experts

I believe yeah that's about right I mean

I don't know what exactly the percentage

is but you know even if you look at like

recent media stories and copyright

information being found within AI models

there's clearly been a problem about AI

being a bit of a black box and what

we're trying to do is hey how do we take

something that's opaque and make it

transparent so how are you making it

better uh the way to do it is to use

what blockchain has always been great

about which is being zero trust and an

excellent governance layer so what we're

able to do is enable companies to

completely track what data is going in

and out of their models and if a

hallucination or bias does occur with

one click you can do what you can do

with every other software which is you

have Version Control so if you're like

hey my model was working 4 days ago but

suddenly it's saying crazy things you

click a button it goes back to what it

looked like four days ago and you can

audit exactly what happened in the

background so like a legion system um so

when people hear blockchain though many

also know that criminals use it that

it's used uh by states that may not be

Democratic yeah of course I mean look

technolog is Like a Knife right you can

use it to like chop wood or stab someone

I mean the internet in its early days

was just porn and gambling I mean every

technology like especially in its early

stages tends to attract a lot of Bad

actors but if you actually think about

it blockchain is one of the most

evocative Technologies out there I think

the big Enterprise leap and using it to

solve our biggest problems like auditing

AI hasn't happened yet but it's going to

happen I mean the internet started like

in the' 60s or 70s with arpa right and

it was only till like the 2000

when it really took off I think now

we're seeing like a convergence of

Technologies and like with any

technology you know there's Bad actors

but they get weeded out and I I think

that's a good thing what happens though

when you get the technology so good and

it falls into the wrong

hands this is this is why transparency

and auditability are really really

important right like going back to my

previous analogy like even with the

internet right we do a lot of banking

Finance commercial activity on the

internet there's a bunch of hackers out

there but as the internet has developed

further and further it's more auditable

it's more trustworthy and that's the

same thing that's going to happen I mean

all we are trying to do is make sure

that hey instead of it being a hard

Landing let it be a bit of a soft

Landing artist rafik anadol is here to

show me his AI installation which is all

about safeguarding biodiversity tell us

how yes so we are here in World economy

forum announc ing our next ambitious

project large nature model and in the

current debates in AI research and AI

practicing is is sometimes missing the

connection with humans us and technology

and the nature I do believe nature is

the most intelligent thing we have and I

think it's time to use this technology

to remember to understand and preserve

um so we are here showcasing how to

preserve data of nature in collaboration

with ethical resource of data like

cultural archives of simonian initial

History Museum and Beyond and also use

um sustainable Computing to create a

meaningful and personal AI model because

I mean this is a lot of data it needs a

lot of computing power and alone this

installation uses a lot of energy I'm

sure yes so this installation is um

required a lot of thinking we only use

50 million data sets to understand the

the promise then we decid to extend our

data set but I'm happy to say that we

made a software that also exactly tells

us how much compute power we are using

for each steps and we can use renewable

energy thanks to Google Cloud custom

software developments and we also create

lots of new tools to understand having

an awareness um especially at a time

that experts here are saying we're

reaching the point of no return

absolutely it's just literally a very

important time and and by the way one of

the reason we are announcing our Museum

of AI Arts called datal land is

literally to take attention and educate

and create a research around the medium

of AI and its Arts I do believe we need

to take attention going Beyond just

shiny pixels or technology but deep dive

into the purpose of AI it's also this is

a virtual Museum isn't it yes but are

you documenting what's disappearing from

the planet or are you creating some sort

of fantasy world here no it's a

completely functional research it is

using science-based data it's not about

hallucinating information it is about

not about human intellect it's about

natural intelligence so we are trying to

Capt

what we have what we lost and try to

explain this in a one AI model open

source free for Humanity so it will help

for Education research or a museum

activity for art making and so it's the

hope that will make that impact refi

canol thank you very much great talking

to you inspiring stuff thank you so

[Music]

much