AI: What is the future of artificial intelligence? - BBC News

Summary

In this discussion, experts Evan Burfield and Gary Marcus debate the need for global governance of AI to prevent job loss, misinformation, and potential threats posed by AI. They argue for a coordinated approach to AI regulation and research, warning against waiting to act until problems arise. The conversation highlights the urgent need for policymakers to understand and prepare for AI's societal impacts, particularly in the upcoming 2024 U.S. election cycle.

Takeaways

• 🤖 There is a growing concern about the rapid advancement of AI and its potential societal impacts.
• 🚫 Elon Musk and other influential figures are calling for a temporary halt in AI development to better understand its implications.
• 🌐 Professor Gary Marcus advocates for global governance of AI, similar to the International Atomic Energy Authority, to prevent fragmented and ineffective policies.
• 🧠 Evan Burfield highlights the difficulty of enforcing a moratorium on AI development due to competitive pressures and varying levels of compliance.
• 📈 The potential of AI to significantly disrupt society and democracy is a major concern, with the 2024 political cycle being a critical timeframe for impact.
• 🛑 The idea of a moratorium is not only about regulation but also about preparing society for AI's wide-ranging effects.
• 📚 Lessons from the internet and social media's impact suggest that proactive measures are needed to avoid repeating past mistakes with AI.
• 🔒 There is a pressing need for education of policymakers to understand AI technologies and their implications for governance and society.
• 🌐 The creation of a global AI governance body is proposed to foster international cooperation and to manage the rapid pace of AI advancements.
• 🔮 The future of AI is not just about technological advancement but also about the ethical and societal frameworks that guide its development and use.

Q & A

What is the main concern regarding the advancement of AI as discussed in the transcript?

The main concern is the potential loss of control over AI, the spread of propaganda and fake news, and the possibility of machines outsmarting humans, which could lead to significant societal and ethical challenges.

Who are some of the key figures calling for a moratorium on AI development?

Key figures calling for a moratorium include Elon Musk, along with thousands of entrepreneurs, academics, and scientists.

What does Professor Gary Marcus suggest as a solution for global AI governance?

Professor Gary Marcus suggests a global governance model similar to the International Atomic Energy Authority, where the world collaborates to set rules and develop tools to mitigate AI threats.

What are the potential negative impacts of AI on society according to the discussion?

Potential negative impacts include job displacement, the spread of misinformation, cybercrime, and the risk of AI models being used to create propaganda.

What does Evan Burfield think about the feasibility of a moratorium on AI development?

Evan Burfield believes that a moratorium is hard to enforce and might give a false sense of security. He suggests focusing on preparing society for AI implications rather than trying to halt progress.

What does the discussion indicate about the current state of AI understanding among policymakers?

The discussion indicates that many policymakers lack a deep understanding of AI and its implications, which is a significant concern as AI continues to advance rapidly.

What historical lesson from the internet and social media does Professor Marcus highlight?

Professor Marcus highlights the lesson of not waiting too long to act on new technologies, referencing the issues of privacy, polarization, and misinformation that arose from the internet and social media.

What does Evan Burfield foresee as the impact of AI on the 2024 political cycle in the U.S.?

Evan Burfield foresees AI having a profound impact on the 2024 political cycle in the U.S., including the potential for AI to influence elections through misinformation and propaganda.

What is the 'tsunami'比喻 mentioned by Evan Burfield referring to?

The 'tsunami'比喻 refers to the overwhelming and transformative impact that AI advancements are expected to have on society, economy, and various aspects of life.

What does Professor Marcus recommend for dealing with the rapid advancements in AI?

Professor Marcus recommends immediate action and the establishment of a central oversight or global organization to coordinate responses to AI advancements and mitigate potential threats.

What is the Auto gbt mentioned in the transcript, and why is it concerning?

Auto gbt refers to AI systems training other AI systems, which is concerning because it represents a rapid and potentially uncontrolled advancement in AI capabilities, raising questions about the speed at which AI is evolving and the potential lack of oversight.

Outlines

00:00

🤖 AI Ethics and Global Governance

The paragraph discusses the ethical implications of AI advancements and the necessity for global governance. It mentions concerns about job automation, AI-generated propaganda, and the potential for AI to surpass human intelligence. The conversation highlights the call for a moratorium on AI development until further understanding is achieved. Elon Musk and other influential figures are noted as supporters of this pause. The discussion introduces two experts: Evan Burfield, a tech investor, and Gary Marcus, a professor at New York University, who advocate for coordinated global governance of AI, drawing parallels to international organizations like the International Atomic Energy Authority.

05:00

🌪 Preparing for the AI Tsunami

This section of the script focuses on the inevitability of AI's impact on society and the economy. Evan Burfield, based in Austin, Texas, observes that startups are actively integrating AI into their operations. He discusses both the dystopian risks and the potential for AI to enhance various sectors, including medicine and government services. The conversation touches on the challenges of regulating AI and the importance of adapting social and market policies to harness AI's benefits while mitigating its risks. There's a consensus that policymakers must be better educated about AI to prepare for its transformative effects, which could significantly impact the 2024 political cycle in the U.S.

10:01

🏛 Policymakers and AI Understanding

The paragraph emphasizes the gap in understanding AI among policymakers. It points out that while there's a growing awareness of AI's potential, there's a lack of proactive measures to prepare for its implications. The discussion suggests that policymakers are not considering the broader implications of AI, including its potential to be supercharged by other technologies like quantum computing. There's a call for the establishment of institutions that focus on AI's long-term impact and the need for a global organization to coordinate AI governance, similar to existing international entities.

15:03

🗳️ AI and Democracy: Challenges Ahead

This final paragraph of the script delves into the potential for AI to disrupt democracy through misinformation and propaganda. It raises concerns about AI's role in influencing election outcomes, drawing parallels to past instances of foreign interference. The conversation underscores the urgency for policymakers to understand AI to effectively address these challenges. It also highlights the need for education and proactive measures to harness AI's potential for positive societal impact while safeguarding against its risks.

Keywords

💡Artificial Intelligence

Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the video, AI is a central theme with discussions revolving around its development, governance, and potential risks. For example, the script mentions 'the development and training of artificial intelligence' and the need for 'Global governance for AI'.

💡Moratorium

A moratorium is a temporary ban or suspension of a particular activity. In the context of the video, a moratorium on AI development is proposed to better understand the implications and potential dangers before progressing further. Elon Musk is mentioned as one of the proponents of halting AI development for at least six months.

💡Global governance

Global governance refers to the management of public affairs at the global level, involving multiple countries and organizations. The script discusses the need for a global approach to AI governance to prevent a fragmented and potentially ineffective regulatory landscape. An example from the script is the call for a system 'modeled on something like the international atomic energy Authority'.

💡Misinformation

Misinformation is false or inaccurate information, often spread unintentionally. In the video, the panelists express concern about AI being used to flood the internet with propaganda and fake news, which can lead to misinformation. This is tied to the broader conversation about the need for guardrails in AI to prevent such misuse.

💡Cyber warfare

Cyber warfare involves the use of technology and the internet to disrupt or attack another nation's critical infrastructure. The script mentions the need for new tools to combat threats like cyber warfare, illustrating the broader security concerns associated with AI and its potential misuse.

💡Competitive Advantage

Competitive advantage refers to the unique attributes that allow a company or nation to outperform its rivals. The script discusses concerns that a moratorium on AI development could lead to a loss of competitive advantage, as companies and countries might not want to halt their progress in this field.

💡Quantum Computing

Quantum computing is a type of computation that uses quantum bits to perform operations on data. In the video, it is mentioned as a technology that could supercharge AI, providing machines with human-like emotions. This highlights the potential for future advancements in AI to be driven by breakthroughs in other fields.

💡Policymakers

Policymakers are individuals or groups that make and enforce policies, often within a government context. The script discusses the challenges policymakers face in understanding and regulating AI, emphasizing the need for education and proactive measures to prepare for the implications of AI.

💡Election Security

Election security involves measures taken to protect the integrity of electoral processes from tampering or interference. The video highlights concerns that AI could be used to spread misinformation and propaganda, potentially affecting election outcomes, making election security a critical issue in the context of AI development.

💡Terminator Scenario

The 'Terminator Scenario' refers to a hypothetical situation where AI becomes self-aware and poses a threat to humanity, as depicted in the Terminator movie franchise. The script uses this term to illustrate the potential risks of developing AI without proper safeguards, although some panelists argue that we are far from reaching such a scenario.

💡Generative Models

Generative models are a type of machine learning model that can generate new data samples similar to the training data. In the video, generative models are discussed in the context of startups applying AI to various problems, highlighting the rapid advancements and applications of AI in creating new content.

Highlights

Discussion on the potential risks of AI and automation, including job displacement and the spread of propaganda.

Call for a moratorium on AI development until we better understand its implications.

Elon Musk and thousands of others signed an open letter advocating for a pause in AI training.

The need for global governance of AI to prevent a patchwork of regulations.

Proposal for an international AI authority similar to the International Atomic Energy Authority.

Concerns about the difficulty of enforcing a moratorium on AI development.

The importance of preparing society for the implications of AI.

Discussion on the potential for AI to cause a 'tsunami' of upheaval in the next five years.

The current state of AI is already profound and will change how we live, work, and engage with each other.

The impact of AI on democracies and the 2024 political cycle.

The challenge of educating policymakers about AI and its implications.

The potential for AI to be supercharged by other technologies like Quantum Computing.

The UK government's decision not to have a dedicated AI regulator and the implications of this decision.

The need for a central oversight body for AI in the United States and potentially a G7 meeting of AI ministers.

The rapid pace of AI development and the challenges it poses to regulation and governance.

The potential for AI to interfere in democracies and elections, and the need for education and preparation.

Transcripts

should we automate away all the jobs

including the fulfilling ones should we

allow AI machines to flood the internet

with propaganda and fake news should we

develop non-human Minds smarter than our

own machines that might one day

outnumber us or outsmart us do we risk

losing control now you might think that

sounds like some futuristic script from

a Terminator movie but last month some

of the most well-known figures who are

involved in the development and training

of artificial intelligence call for a

moratorium until we better understand

where we're going an open letter was

signed by thousands of entrepreneurs

academics and scientists including Elon

Musk who wants the training of

intelligence haltered for at least six

months we're going to dig deep into this

over the next 20 minutes or so in the

company of two people who know a thing

or two about it joining me is the tech

investor even burfield the Evan burfield

the author of regulatory hacking A

playbook for startups he's in Texas I'm

Professor Gary Marcus is in Vancouver

he's the professor emeritus at New York

University and author of rebooting AI

Professor let me start with you

um clearly with such advances that we're

seeing we have to set some guard rails

who do you think should be in charge of

that

uh were we both professors I think that

we need Global governance for AI I think

that we have a lot of patchworks right

now almost balkanized

um the worst case from the company's

perspective and the world's perspective

is if there's 193 jurisdictions each

deciding their own rules requiring their

own training of these models

um each run by governments that don't

have much specific expertise in AI so

what I called for an economist editorial

earlier this weekend in a TED talk

earlier this week was to have a global

system modeled on something like the

international atomic energy Authority

where the world comes together and says

we have a new threat here but it's

really a new set of threats and we need

to work together on this so I think the

number one thing is it should be Global

and the number two thing is it can't be

just

um policy but it also has to be a

research side because we need to invent

new tools like we had to invent for

fighting spam and cyber warfare and so

forth there's so many different threats

as you mentioned around misinformation

cyber crime and so forth so we need to

have a kind of standing organization

This Global and well-financed to try to

build tools to mitigate those threats so

Evan burfield there are many many people

who who just want to press the pause

button until we work out some of these

things but I can already see and I've

heard uh the reasons why that probably

isn't possible and and that is because

not everybody will stop and people are

worried about losing competitive

Advantage so how do we best go about

this

yeah I think that's exactly right

um you know the the challenge with a

moratorium is that it's incredibly hard

to enforce

um the responsible actors would be more

likely to follow it the irresponsible

actors wouldn't but that's actually not

so much my concern with the moratorium

there's absolutely questions we need to

be asking about the governance of AI

about what industry can do what

government can do uh I think the letter

did spark a conversation Schumer is

working on a new AI bill here in the U.S

rumors McCarthy's working on a

republican version

but what I think is actually much more

important is to start to have the

conversations about how we prepare our

society our economy uh our political

system democracy itself for all of the

implications of AI that are coming one

way or another and I suspect we'll see a

year from now we'll go this was less

impactful than we thought five years

from now it will be an absolute tsunami

of upheaval and we have this window

right now where we can have this

conversation and we can get creative and

I think we've got to use it and a

moratorium gives us this false sense of

security that we have control and can

stop it versus figuring out how we ride

this tsunami and try to direct it in a

much better Direction Professor Marcus

did we

did we learn anything from the the last

technological Advance the advance of the

internet of social media are there

lessons from that which we let's face it

we didn't do very well that are

applicable here

I think the number one lesson is you

don't want to close the Barn Door after

the horse has left I think you know

we're very late in in figuring out what

to do about social media I think we

probably handled privacy alone in the

wrong way

um we wound up with so much polarization

and hostility we wound up with

misinformation

um I think we waited too long to act I

think the number one lesson is we should

get on it right now and I agree with the

other panelists that the moratorium You

could argue about the merits whether it

was the right thing or the wrong thing

it was absolutely the right thing to

raise this and get it on everybody's

agenda this is not something we want six

months from now something we need now so

Evan burfield when you talk about a

tsunami in five years time what does

what does that look like

uh look I'm down here in Austin Texas uh

at Capitol Factory startup accelerate

I've spent the whole day uh you know

meeting with startups and there's not a

startup right now out there that is not

applying these AI generative models

these large language models to every

interesting problem of the Sun and

there's all of the the scary dystopian

possibilities that you uh led into this

segment with but there's also uh

incredible advances in how to make work

more fulfilling and more impactful how

to apply uh tremendous personalization

to Medicine based on our genetics our

environment the particular issues we're

having how do you make government more

responsive and feel more like a

concierge to Citizens all of that is

also being worked on

um and I think figuring out how we put

the guard rails in place around some of

the scarier things which isn't just

about regulating Aya it's about changing

our social policy changing our Market

policies themselves

so that we can mitigate some of that and

and direct this into the the much more

hopeful and optimistic Direction why why

on that point though Evan is it is it

imaginable in the current scenario you

are around it all the time that a

research lab would cross a critical line

here without even noticing

I I'm I'm personally skeptic you know uh

Gary's written some wonderful points

about the fact that we are we are very

very far I believe from artificial

general intelligence and the Terminator

scenarios I I think we've got to be very

aware of right now is simply that this

technology is already right now today at

a state if it did not Advance any

further where its application is going

to profoundly change how we live our

lives how we work how we engage with

each other in communities how our

democracies function uh the impacts on

our democracies are going to be felt

right in the 2024 political cycle here

in the U.S that's what I think we need

to be talking about and preparing for

the scenarios of AI is like nuclear

weapons we have to ban it immediately I

think are much less applicable to the

the much more realistic uh changes that

are already happening around us right

now that are going to accelerate miles

um you've just come back from Washington

and I know that you've been talking to

policymakers about the specific issues

in fact the reason we're talking about

it tonight is because you tweeted no one

has a clue I I mean is that is it as

blunt as that that nobody really

understands it there is practically no

work being done on it

hi Christian you're you're spot on the

three biggest challenges right now with

policy makers are one this was

completely foreseeable there were some

of us in Washington talking about this

10 or 15 years ago policymakers weren't

paying attention and most of the think

tanks in Washington really failed to

start a conversation about the Practical

things that needed to be done to prepare

for the age of AI so we're behind the

ball from a policy making standpoint the

second thing I would emphasize what Evan

burfield just said there is a wave

coming and you can do two things when a

wave is coming you can get crushed by it

or you can ride the wave and to use

another analogy right now the discussion

in Washington is about whether to put

the genie back in the bottle or not that

shouldn't be the discussion it should be

what three wishes should we ask the

genie and that's the discussion that

should be had about how to handle Ai and

use it for good purposes and finally the

other problem is policy makers are not

thinking two steps forward on the

chessboard it's AI right now but in in

this decade AI is going to be

supercharged by other Technologies like

Quantum Computing that are going to give

machines genuine human-like emotion what

are we doing to prepare for that we

should be having that conversation now

there needs to be institutions in

Washington that focus on that so Jack

we've had a discussion about that in

this country and the UK government has

decided that it doesn't need a dedicated

UK regulator for AI so who's overseeing

it

that's a very good question I mean uh I

I was on stage last night with the

chancellor Jeremy Hunt and I asked him

about this you know he's the guy in

charge of the UK economy

um and he was really quite dismissive

he's he in in the sense that he said you

know this is something that is going to

happen and we have always embraced new

technologies in this country and we

should do so again uh it's full steam

ahead was the phrase that he used

um you know he was very very positive he

did he did not want to talk about the

possibility that people would lose their

jobs because of this technology he only

saw it as a purely positive thing and he

was not keen to talk about the way it

should be regulated now you know I'm no

expert on this still if I'm a politics

guy but what I do know is that is how

Westminster works and how um political

systems work and I can tell you now and

you'll know this Christian there is no

way our political system is set up to

deal with this challenge absolutely no

chance the speed at which decisions are

made in Westminster and I suspect in

other major political centers is far too

slow to cope with the pace at which this

technology is coming the policy makers

do not understand it at all this is just

something that is going to wash over us

and we're going to have to cross our

fingers you know the UK government put

out a white paper which is what they

call their draft strategy on AI the

other day I mean just the very name of

it white paper tells you how old school

this is yeah you know it's out of date

already and that thing took you know

years for them to put together

um we just don't have the sort of nimble

small system smart thinking people set

up to deal with this and I'd be very

surprised if that's different in the US

or indeed in many of the other big Power

centers well they clearly don't

understand it Professor Marcus do they

call you in to try and get you to

explain it to them

that I was talking to people in the U.S

and Canadian government yesterday I've

been called a lot lately

um I think there is an awareness that

people don't quite know what to do and

they are increasingly turning to me

um and also turning to all of my you

know academic colleagues and so forth so

I think that there's at least a

recognition and people know what they

don't know I do think that the UK white

paper saying that you won't have a

central office of AI is certainly for

all the reasons that were kind of

implicitly just said which is

um the government is going to be

ill-equipped to deal with the speed of

this and if you just leave it to 20

different Regulatory Agencies Each of

which don't have expertise you're asking

for trouble you're asking for a lack of

coordination and it's just not realistic

that all of those agencies are going to

be up on things so there needs to be I

think at least some Central oversight I

think the United States should consider

a cabinet level AI officer and you

should consider something comparable

um you need some people maybe like a G7

then we have a G7 meeting of foreign

ministers we need a G7 meeting of of

IAAI ministers is that is that

effectively what you're saying well I

mean I'm calling for something similar

which is a global organization uh kind

of like the IMF or an international

atomic energy agency

um where you have a lot of experts you

have a lot of people in government you

have a lot of people in uh the companies

and yeah you have regular meetings

you're like well this week the new thing

is this thing called this is a real

example called Auto gbt where you have

ai's training other AIS what do we do

about that how big a threat is it is it

a small threat big threat like if you

have a research arm then you can say

let's do some experiments here and try

to figure out what the limits are right

now instead you have like 193 countries

maybe some of them have read the news

about this major news Discovery some of

them haven't even aren't even aware of

it and there's like no coordination here

that just can't be the right way you're

nodding Evan because this is the key

issue it's miles as miles discussed it's

not human competitive intelligence it's

it's what happens after AI gets smarter

than humor intelligence right amazing

but you know I I can't go to a

conference I I actually live in

Washington DC most the time I can't go

to a dinner or a conference a meeting

without the word I being discussed and

they're all talking about chat GPT and

Gary's right is that even Auto gppt

there was a an experiment run last week

called chaos GPT where they took a

neutered version of Auto GPT and told it

to go out and figure out the most

efficient way to destroy Humanity it was

a it was sort of a test and it's set to

work doing it there's there's a lot of

this stuff is moving incredibly fast and

figuring out how you can educate policy

makers about how to mitigate regulate

bring transparency to some of those

threats while not preventing what can be

breathtaking advances in

um how we live our lives in much more

fulfilling and purposeful ways and

Society I think that's that's a lot of

the trick here to Echo Jack's point

though about white papers and the way

you know government moves I I tend to

agree you know miles may be more

optimistic than I am but I tend to agree

I think a lot of the big changes that

are going to need to happen probably

won't happen until uh there's some sort

of provoking event some sort of Crisis I

don't think that though prevents us from

starting to have the conversations at

least the Way Washington tends to work

at least you want to have the the policy

container the framework the ideas ready

some sort of consensus being built so

that when the opportunity presents

itself kind of like a a VC who sees a

great startup right when the opportunity

presents itself you're ready to jump on

it you're ready to move forward and I

think that that has to be happening

right now

go ahead

the opportunity that I see right now is

to build some Global governance I think

you have the governments are afraid of

the technology companies the technology

companies are afraid the governments are

going to shut them down as they did in

Italy and this means everybody has some

incentive to go to the table that's rare

and I think we should be seizing that

opportunity right now to try to do

something coherent that is dynamic

enough to cope with the speed of the

change to take advantage of the good

things and and to avoid the bad things

but we need that coordination now and we

can't just leave this to the usual

mechanisms it's just too slow miles one

of the more worrying things that you

said was that speaking McCarthy was

looking at an AI for for republicans and

you know one of the experiences we have

of recent years is that the Russians

were able to interfere in a democracy

and who knows arguably it's been debated

whether they were able to change some of

the results through what they were

putting onto the internet I mean we're

into a whole new ball game for democracy

if AI can put out misinformation and

propaganda

it's not if it's how much and when it's

going to happen probably in West 2024

election wow yeah there's no question I

mean this this coming election cycle in

the United States it's a big concern for

election security authorities it should

be but I yeah I got to go back to what

the other panelists said in order to

respond to it effectively we've got to

start with education and right now I

mean I've tried to brief policy makers

on this it's like explaining particle

physics to a chocolate chip cookie I

mean there's just not recognition about

what's happening if I was President Joe

Biden right now I will put I would put

the entire cabinet on Air Force One I

would fly them to Silicon Valley and we

would spend the week educate educating

them about what's happening because

there aren't just these security

implications for the elections as Evan

notes there's also really positive

implications I mean there's the ability

to address major health care problems

hunger homelessness and to do it in real

time and we are missing some

opportunities by policy makers not being

educated on the subject but of course

security has to come first and in order

to protect elections or anything else

it's got to start with you know policy

makers becoming technologists being

educated

fascinating conversation we're gonna

have to leave it there Evan burfield

Gary Marcus thank you very much indeed

for joining us