UN Security Council meets for first time on AI risks - BBC News

 

way we live but the rise of chat GPT and

the other fast advancing systems has

been accompanied these past months by a

sharp increase in anxiety it's the

ability of AI systems to teach

themselves or to grow in ways we don't

fully understand yet that has the

researchers predicting a large-scale

catastrophe

that makes it the business of the United

Nations security Council and today for

the first time they began the dialogue

about its potential impact on Global

Peace and security the UK which holds

the rotating presidency of the council

so the challenge is to mitigate the risk

AI presents through coordinated action

while making sure we all benefit from

the tremendous things it could deliver

let's speak to our North America

correspondent nomia Iqbal who is

watching that session so what has the

foreign secretary James cleverly been

saying Nami

yeah so Christian this was a big deal

for the UK for the foreign secretary

James cleverley the UK really wants to

show that it can compete with the global

Giants in what is the most hyped up area

of tech that it can be a leader

geographically and intellectually when

it comes to artificial intelligence so

you know you mentioned there they are

holding the rotation of the presidency

this month and so they've wasted no time

putting on this uh the security council

meeting and I spoke to the foreign

secretary Mr cleverley uh just before

the meeting started and I I started off

by asking him what was the point of this

meeting what was he hoping to achieve

with it

well AI is having and will have an

amplifying effect an accelerating effect

and all the things that we currently use

technology for

and that can be used for good for

medical research for a research on

climate change on analyzing big data

sets but there are also potential malign

uses of AI so what we are saying today

the U.N security Council is that we need

to work internationally to understand

the risks to look to mitigate those

risks and also to put some structures

and regulations in place that can only

be done internationally that's why we

discussed it give me an example of what

you think is malign

well the use of AI for example to

develop weapons and just as AI can be

used to develop drugs that could perhaps

solve some of the disease challenges of

the world it could potentially be used

to create bio weapons so that's the kind

of thing that we need to do harvest the

positives and protect ourselves

internationally from are there any

countries that you're particularly

concerned about in terms of their

development of AI well I think one of

the things we have to understand

is that it it wouldn't just be a state

actors that could potentially use AI for

a negative or malign purposes so it is

about thinking about how we develop it

what we release publicly what is a more

tightly held now I don't pretend to have

all the answers no one does but working

together we can analyze some of the

challenges and look carefully at what we

do as I say to protect ourselves

internationally whilst also harvesting I

mean you've got countries like China

Avenue that are behind the US and ahead

of the UK does that concern you well we

have one of the people that we brief in

the UN Security Council today is a

Chinese technologist we also have a

British technologist there

it is in everybody's interest

everybody's interest every nation and

every person's interest to get this

right and better that we get it right

together than in silos or isolated from

each other that's why the U.N security

Council I think is a really good

starting point the UK is hosting an AI

safety Summit later on this year and we

will play our part working with the

International Community to harvest the

benefits whilst protecting ourselves

so that was the UK foreign secretary

James cleverley chairing that meeting

and Christian it was one of those

meetings where there were was a lot of

talking there were lots of statements no

concrete goals were achieved at the end

of it but the UN Secretary General and

Tony gutres was in that meeting and he

did say that he his vision is for the UN

to create this sort of a sort of

governing body to to govern uh

artificial intelligence and the way it

does when it comes within the way it has

bodies to govern the use of nuclear

weapons uh you know Aviation energy and

meatballs the challenges of climate

change as well

um he taught James cleverley there about

this Chinese technologist that was in

the room giving evidence but what does

the Chinese Camp think about this

generally

there was some tension there because uh

China is uh you know leading sort of

second just behind America in terms of

advancement of artificial intelligence

and a spokesperson for China said uh

that if there are going to be these U.N

bodies or rules that it should reflect

the views of developing countries and it

shouldn't be uh these uh the Western

Nations that uh that decide how

artificial intelligence uh you know is

is governed that that other country

should be able to establish their own

regulations there is some tension

between China and America of course

there always is there are some reports

that the the US is looking to limit the

flow of powerful artificial intelligence

chips and there was also

um you know the US and the meeting they

didn't they didn't directly sort of uh

address China's concerns but they did

make a a dig at China accusing China of

using technology to uh to monitor ethics

thank you for that uh joining me from

San Francisco California is Anthony

aguiri he is the executive director of

the future of Life Institute uh he's one

of these people that knows everything

there is to know about this issue um

let's talk about what was proposed today

the UN says it would like to form a

governing body what would that look like

in your opinion what would it be tasked

with doing

well I think there are a lot of models

that people have proposed for bodies

like this this is still very early days

I think the crucial thing is that we we

have a body that can address some of the

sort of urgent and and profound risks

that this these Advanced AI systems are

starting to pose

um these are Broad and extreme and I

think we're really at a new stage as we

were at the beginning of the nuclear era

where we have to change the way that we

do things internationally and have new

bodies that can uh that can have the

sort of safety of humanity as a whole uh

first

I asked chat GPT today what risks it

presents to humanity and it came up with

a list fairly exhaustive list actually

let me put them on screen for you so it

says there are cyber security

vulnerabilities there's the

weaponization of AI there's

misinformation and disinformation data

privacy and mass surveillance bias and

discrimination which would be the unfair

targeting of specific populations the

unemployment it could create which of

course leads to social unrest and

finally the AI arms race that nomia

talked about that is a lot to get your

head round and it is developing so

quickly do you subscribe to the theory

that perhaps we're all ready too late to

put a governing body around all those

issues

I don't think we're too late but I think

we have to act quickly and I do think uh

you know I was one of the initiators

behind an effort to pause AI we had a

large open letter I think we do need to

slow down and take a little break from

the Breakneck and competitive speed of

development so that we can allow the uh

governance and the regulation and the

creation of new institutions to catch up

so I think we need to act fast and slow

down uh the race you know to get these

super powerful AI systems

um just reflect for me if you would on

the view of some within the Chinese camp

that um if there is a governing body it

needs to reflect the the wishes and the

the demands of developing nations you're

sitting there in San Francisco and it is

true that in in the social media space

it's it's America that is commanded it's

America that has led the way to the

exclusion of other countries around the

world sometimes

well I think there are there are lots of

different Arenas that AI is going to

play in uh they're gonna it's gonna be

incorporated in many parts of our

economy and different countries are

going to have different ways that they

want to regulate and sort of govern how

AI plays out in their countries and in

their societies I I think there are

issues that transcend individual sort of

countries and and governments uh which

are the ones that you know threaten

Humanity as a whole so this you know in

the same way that the U.S and the Soviet

Union came together during the Cold War

to make agreements about nuclear weapons

I think countries are going to have to

come together at this very high level

and make agreements about the most

powerful AI systems and how we're going

to keep them safe and under control but

at the moment China is is lagging behind

America and the development of

artificial intelligence you can see a

scenario though where they they might

catch up and they might see this as the

short cut to not only

um an economic Advantage but also to

China you know becoming the most

powerful Nation on the earth if they

were to go their own way and not

subscribe to a rules-based system would

that undermine the rest of the club

well I think one of the reasons we we

need to get together and coordinate

internationally is the alternative is

this sort of pernicious race that

somehow by racing to these ever more

powerful systems delegating more to them

uh giving more capability and more

decision making to systems where we

don't really understand how they work

that we're somehow going to win and I

think this is just getting into a race

like that is a race that that people are

not going to win you know the human race

is not going to win the AI is going to

win that race uh at all of our expense

just on on security and defense which

was part of the discussion today there

are concerns that AI could be used to

set false targets or to put satellites

off targeting and the we might in the

future have to think of very different

ways to how we approach defense is it

incumbent then on NATO and the Western

allies and maybe we've learned some

lessons during the war in Ukraine to

find their own solutions to some of

these problems

um well I think you know countries are

going to have different their own

decisions and how to incorporate Ai and

various things parts of what they do uh

one of the things we've been concerned

about is incorporation of AI into

command and control systems this is

there's a big push for this uh and and

one of the things we really want to be

careful not to do is over delegate and

definitely not you know this seems sort

of obvious but definitely not

incorporate AI into nuclear command and

control systems uh this just seems like

a terrible idea uh our organization

actually put out a short film today kind

of illustrating the risk of over

delegating AI including a nuclear

command and control I think we could all

agree this is a a bad idea and that we

should keep certain high-stakes

decisions really in human hands so I'm

hoping that some of the low-hanging

fruit for international agreements can

be things like we're not going to

incorporate Ai and nuclear weapons we

are going to keep human hands on certain

things like uh the decision to take

lives and and to uh you know make large

decisions and command and control I

think there's low hanging fruit where we

can all agree just in in sort of Base

levels of sanity and how we use AR

that's where I would start

Anthony again very interesting

non-proliferation of AI within nuclear a

discussion no doubt for the future thank

you very much indeed around the world

and here in the UK you're watching BBC

News