Good evening everybody. It's terrific to

see a packed room. So, welcome to the

famous or perhaps infamous New York

Times debate at Davos. My name is Steven

Dumbar Johnson. My role is actually

coming to an end because it was to

quieten you all down. But I have the

real honor to introduce the president

and chief executive officer of the New

York Times who's going to be our chief

mixologist of our cocktail.

>> But wait, wait, wait. Stephen Steven

runs the international business for the

times and I think is the best person to

come stand at a podium and make everyone

want to pay attention. We all pay

attention to him

>> and does many many other important

things. All right. Wait, Stephen, I may

need you stay here. Stay close for a

minute. How many people were here LAST

YEAR?

All right. That I actually think that is

the most impressive part. And does

anyone remember last year? So we've done

this debate for six years. It is always

rowdy. It is always filled with audience

participation. I can tell you our

debaters, our moderators, our judges,

our judges are right here in the front

are expecting audience participation. It

is a huge part of it. But does anyone

remember the topics are always amazing.

I think the first one that I got to

introduce was should there be quotas for

women on boards? Um topic very dear to

me. I'm just going to say upfront the

topic we are doing today. We're going to

do this in the spirit of independent

journalism. This is really about let's

let the best facts support your opinions

and arguments. Let's let you form those

opinions from a really factual debate by

experts who've each come here to stake a

position or take a position. I think in

some cases we've even asked folks to

take the other position that they might

hold personally just for the purpose of

great debate. I want to say to you

today's topic this house believes that

AI will succeed where humans have

failed. I, as the CEO of the New York

Times, have some very strong views

about. I'm not gonna share them until

after we debate because I think that

might color the debate and maybe some of

my views will surprise you. But I just

want to ask, does any raise your hand if

you remember, just for fun, what last

year's topic was?

>> Does anyone want to shout it out?

>> US.

>> Should the United States step aside?

So, all I want to say,

I definitely don't have a view to share

on that topic, but what I want to say is

I think we're pretty darn good at

picking topics that are going to matter

for a while. My real job here is to

remind you, show us what you think in

all the ways a crowd can show what you

think and to int introduce, let me bring

them on up, our two amazing, amazing

debate moderators. They are both

veterans. They've done this before,

Katherine Benhold. If you don't know her

already, you should. She is the author

and we call it now host of the world

which is a newsletter the New York Times

launched I think this past September. It

is if you live in the United States or

internationally and you get the morning.

It is our more sort of broad focus on

the world version of the morning. It is

awesome. You should be reading it and

watching it. There's a lot of video in

it. You should be paying attention to

Katherine. She's worked for us all over

the world. was Berlin bureau chief many

other things David Gellis no stranger to

this room I can't remember did you

moderate the should women should there

be quotas

>> no not that one

>> that was me

>> that was you that was

>> I did should the US step aside

>> you did should the US step aside um

David I only don't remember that because

I missed Davos last year but David um is

the climate correspondent for the New

York Times one of many people covering

climate but we've got a whole franchise

that David has organized and led called

Climate Forward, which is a I think

twice weekly newsletter. You should all

subscribe to both their newsletters and

just because I think this makes him an

expert on drawing debate out of people.

He was the corner office columnist at

the New York Times for four years. So,

lots of chops. Over to you guys and I

will be back at the end.

Yeah, good evening and a very warm

welcome from us as well to the best

event in Davos every year, the New York

Times debate.

>> The motion tonight is AI will succeed

where humans have failed.

>> And this is a vintage human debate. We

are 100% human.

>> I mean, who knows? This stuff is

evolving so fast. Maybe next year it

will be called debating chat GBT

>> and then the motion will be humans are

pointless. Let's get rid of them.

>> What would the counterargument be?

>> Oh, I mean maybe human are pointless but

they're kind of nice. Maybe we should

tolerate them like the birds and the

rocks.

>> I hope so. But ladies and gentlemen,

we're not quite there yet. At least for

now. While AI is better than humans at

some things, we for now are still in

charge. I think

>> who knows this stuff is evolving very

fast. David, before we've heard the

arguments, where do you stand on this

motion?

>> Look, for those who know me, you know I

am a congenital optimist. I look at all

the advances that AI is already

delivering and I can't help but think

we're going to see more of it. AI is

already making businesses more

efficient. It's discovering new

medicines. I now plan on living to be

150 years old.

Well, I'm an optimist, too, but I still

think we're toast. We might lift to 120,

but we will all be unemployed.

>> Okay, indulge me if you will for just a

moment. Here is my proof that AI can

already succeed where humans have

failed. And I swear what I'm about to

tell you is true. Yesterday morning, I

woke up in Davos to an email from my

10-year-old son's science teacher. and

she was informing me and my wife that

our 10-year-old son had turned in a

science assignment that was composed

entirely by generative AI.

This is my 10-year-old son. And what was

the result, you might want to know?

Well, the work was entirely correct,

flawless, but he flunked the assignment.

In other words, AI succeeded where my

human failed.

Good luck to us all.

But joking aside for a moment, what is

the really big question that we're

debating here tonight? The question that

matters to us and to the future of our

children. The question is, can this

stuff play out and work out well for us?

Can we take advantage of the huge

benefits of AI without AI completely

undermining our social societies and

potentially posing an existential threat

to ourselves?

>> That's the big question and the motion

is short and sweet. AI will succeed

where humans have failed. But as you

just said, this carries vast

consequences. Is AI this enormous

transformative opportunity that can

offer the solutions to these intractable

failures that we've been confronting for

millennia? Things like climate change,

inequality, hunger, disease, conflict.

>> Or is it our route into this dark

dystopian future with mass unemployment,

a bunch of trillionaires in charge, and

yeah, not a lot of fun.

>> Now, for the purposes of this debate, we

will offer a definition of AI to level

the playing field. AI includes

generative and agentic systems trained

on data to recognize patterns and

optimize actions. It does not include

AGI.

>> AI wrote that for you

>> busted. But thank you Gemini.

>> So David, should we should we do this?

>> Let's do it. Okay, let's do it.

>> Can the debaters please make their way

to the stage?

>> Welcome our debater.

>> So this is again

>> Yeah. Perfect.

All right, here they come. Arguing in

favor of the motion, arguing that AI

will indeed succeed where humans have

failed, we have Verun Civam, Adam Grant,

Max Clemenco, and Bour Betagel.

>> And arguing, yes, a hand. Give him a

hand. And arguing against the motion, we

have Gerard Reed, Adele Walton, Cave Kth

Butterfield, and Kirsten Dunlop. And

now,

>> David, I I I just noticed something. The

AI skeptics have three women and one

man,

>> and the AI

>> champions are three men and one woman.

Interesting.

>> Well, well, noting that,

>> come back to that. A and and for the

purposes of this debate only and without

losing any of our independence as New

York Times journalists, Katherine and I

are going to be standing temporarily in

allegiance with one team. I, as the

optimist, will be representing the four

team

>> and I will be representing the against

team.

>> Now, just a bit more housekeeping before

we get into it. You, the audience, will

ultimately be deciding the winner of

this debate. Please make your judgments.

This is so important. Please make your

judgments not based on your

preconceptions about this question when

you walked in the room, but on the

strengths of the debator's arguments

that you hear tonight. Ask yourself, did

you find them to be persuasive?

>> And also, we want you to make a lot of

noise. We actively encourage heckling

and applause and clapping. Just go for

it. Give us some love.

Yeah,

less booing, lots of feet stomping. All

right, we're we're going to get things

going in just a sec, but let me break

down how it's going to go. First, we'll

hear from our debaters one by one,

starting with the first debater for the

four team, then the first debater from

the against team, and so on. Debaters,

you will each have three minutes. We

will cut you off if you go over time.

Then we will turn to our esteemed jury

sitting in the front row here. We have

Kate Karat, Khulli Cavniss, Mirana

Edgar, and Matthew Prince. Thank you for

being here tonight.

And yet, jury, I am sorry to let you

down, but you are not here to decide the

winner. After you hear all of our

debaters, we want you to give them tips

they need to win in the final home

stretch. You want to tell them what you

liked, what you didn't like, if there

were blind spots in their arguments, and

then each team will have just a few

moments to refine their argument, digest

their feedback, elect one debater to

represent their team for one final

chance to seal the victory.

>> And then you, the audience, are going to

get to pick the winner. And the way you

do that is by making a lot of noise.

Basically, the team that gets the most

applause will win. And before we turn to

things and make make this real, we're

going to just have a little practice run

here of the apploter that we're using at

the end at the NT. If you believe the US

should buy Greenland, let's hear it.

>> Okay. That that that didn't really work.

>> Okay.

>> Okay. If you believe that the US should

not be allowed to buy Greenland.

>> Okay.

>> Okay.

>> Okay.

That's what victory sounds like. That's

what you're going for.

>> All right. Time for our first two

debaters. On the foreside, we have

Verun. Something you may not know about

Veroon is he is a lifelong afficionado

of Indian dance. He captains his college

and graduate school dance teams and

still choreographs routines for Indian

weddings using AI.

And up against him is Jerra, who since

he was a teenager has loved roses. He

loves gardening. And he's actually

created the biggest rose garden in

Berlin, which as somebody who's lived in

Berlin, I absolutely need to visit. I

want to see Gemini do something like

that.

>> All right, the timer is set for 3

minutes. We are stepping off. Baroon,

the stage is yours.

I'm going to take my jacket off.

>> OH,

do you want Do you want to take your

jacket off? Just Just

>> Absolutely not.

>> No.

>> Do you want to restart the timer?

>> Don't reset. Don't reset.

>> Yeah. Reset. There we go.

The good news is that we got the world's

most powerful people to come to this

snowy city to solve the world's

problems.

Bad news is that we tried this last

year. It didn't really work.

I'm here to tell you that AI is going to

change this pattern. May not change it

tomorrow or next year, but the good news

is this resolution doesn't have a time

bound. And so give it a year, five

years, a century, I'm confident AI will

succeed where humanity has failed. To

kick us off, let me give you three

contentions. First,

AI.

First, AI is going to become far more

capable. Second, humanity's global

challenges have root causes that AI can

actually fix. And third, let me deep

dive on one of them, climate change, and

answer the doomers with an optimist

take. First, AI will become much more

capable. The model evaluation and threat

research nonprofit says that AI's

capabilities are increasing. They're

doubling every seven months. That means

today AI can do a task that takes a

human 4 and a half hours. By the end of

the decade, AI can do a task that takes

a human a month. And a decade from now,

AI will do something a human can take a

lifetime to do. NVIDIA research has

showcased that AI's learning. Turns out

that a year of human experience can be

compressed into just an hour of robot

simulation and experiment. AI scales

human cognition. You may argue that

it'll hit a scaling wall. I wouldn't bet

against AI. Second, global challenges

where humanity's failed have root causes

that AI can fix. Look, today we live in

a world with climate change, global

pandemics, persistent poverty, and many

folks around the world don't have

dignified lives with empathy.

But the problem isn't what most people

say. It's not political will that's

lacking. It's not a lack of

coordination. that we don't tax the rich

in order to do social services. We don't

reduce carbon emissions. It's that

there's scarcity. And AI through

scientific discovery helps to flip the

incentives, the perverse zero someum

incentives that scarcity creates. Now,

notice I am explicitly not including

what David and Katrine said, inequality

as a bug. It actually may be a feature

where if rich people exist, hey, so be

it. So long as we have increased the

standard of living of all humans, then

fantastically wealthy people are

actually a feature, not a bug. They

inspire the next generation of

entrepreneurs. Let me finally close

three with climate change. Look, I'm

going to pander to my jury. I'm rooting

for Cully and Crusoe to build so much AI

that we go from 5% of the American grid

to 25%.

That sounds like a climate explosion

fueled by natural gas, right? Wrong. I

think it's a boon for clean energy. My

company, Emerald, is an AI for AI.

Making AI a flexible power user,

bringing on clean energy onto the grid

and power cheaper. AI will supercharge

materials discovery. Deep Mind has

already discovered 380,000 stable

compounds potentially for solar storage,

carbon capture, and in the future, AI

will stabilize fusion, creating

limitless clean energy so countries have

it in their interest to use the cheapest

abundant form of energy.

>> Well, you sped up like a robot at the

end there.

Well done.

>> We meant it.

>> Okay, Gerard, over to you.

>> So, I don't know why I'm here. And the

reason I don't know why I'm here is

because how can we say that humankind

has not been successful? And I reflected

on this and then I thought to myself,

okay, what would my grandmother do if

she was here today? My grandmother was

born in 1910, died in 1990. And she had

a very good life, but she also had a

horrific life in lots of ways. And I

won't go into the ins and outs of that,

but I thought to myself, what would she

say? So, okay, I'm Nana. If you're here

now, what would you say? She'd say,

Jared, god, you've done very well for

yourself. She'd look and say, electric

cars. Oh my god, look at the lifestyle

you have. That's incredible. I'd love to

live in your times. That's the first

thing she'd say. And then she'd say,

"What about the country that we grew up

in, Ireland? What would you think of

that?" and she would look in amazement

and say, "Jared, in 1988, Dublin had 18%

unemployment and was the poorest country

in Europe and today it's the wealthiest

country in Europe." And she'd go, "I'd

love to live in that country because we

didn't grow up in that." And then she'd

sort of say, "Well, what about Europe?

Because we're in Europe and my mother

moved to Germany, actually." So she'd

go, "What about that?" And she'd go,

"What? Germany got a peaceful

reunification?"

And guess what? We've had no major wars

in this in this in this continent since

the Second World War. And we bought

incredible prosperity to everyone across

our continent. She'd look at it in

amazement. And then I'd say, "Let's

there's one better. Let's go to China."

She go, "What? China? That poor

country?" I go, "No, no, no, no, no, no,

no, no, no, no, no, no, no, no. That's

that's that's the past." I said,

"They've brought one and a half billion

people out of poverty over the last 35

years."

And she go, "What?" She wouldn't believe

it. I said, "You need to go to Shanghai.

I'll show you Shanghai." I said, "Mom,

Nana, the water is cleaner, the air is

cleaner than Dublin today." That's the

reality of it. Okay. And then and she

say, "Okay, what about the rest of the

world?" And I got really simple. The

rest of the world is as follows. If you

went back, you know, in that period of

1990 when you died, since then we've

three billion more people living in the

world than we did then. That's called

success, right? Really simple. It's

success. We're feeding more people than

we did ever across the world, right? So

that's how she would look at it. Now,

how do I look at it? I look at it the

exact same. I look at it and say we have

never been so successful across the

world in solving problems as we have

been particularly in the last 20 years

and my message to all of us is that I

hope we continue doing that going

forward. Right. And that's it. I'm going

to not use my full time. OKAY.

>> OKAY. Thank you Gerard. So on the

foreside, on the foreside, we heard

Verun say that AI will allow us to

consume clean energy without limit. And

as a climate reporter for the New York

Times, I like the sound of that.

>> And I heard you say, Gerard, that AI may

have infinite knowledge, but sometimes

for wisdom, we should talk to our

grandparents.

>> All right, let's welcome up our next set

of debaters. On the foreside we have

Balor who is impressively the proud

owner of 500 sheep. She apparently she

doesn't have names for all of them just

yet. But I would venture that a quick

LLM query would solve that

instantaneously.

And up against her on the against side

we have K, whose favorite activity is

apparently walking her dog in the snow.

It's not 500 dogs, but it's one dog. So,

>> welcome.

>> The timer is set.

>> Lord, the floor is yours.

>> Thank you. So, if you're in Mongolia,

barbecue is on me. I have 500 ships.

Um, I was born into a nomadic family in

Mongolia. I've always always wanted to

help developing countries who are living

like Mongolia. So I joined the World

Bank when I was 20 years old hoping that

this is the easiest and fastest way to

save lives. For years I've I've seen

people good people qualified smart

committed people people stuck to stuck

writing memorandum justification memo

for the fifth time. I watched 20 million

emergency loan for a flat relief get

delayed three months before it gets

delivered to the people it needs by

waiting signature from four different

agencies. The reality is bureaucracy

doesn't just slow us down. It

contributes to killing. It contributes

to excess debts. Within days of within

days of Turkey and Syria earthquake, the

world pledged $1 billion dollars in aid.

Warehouse in Gazant were stacked with

medical supplies, warm food, tents,

winter coats, and they sat there for

weeks, for months, not because people

didn't care, because the grant

application required 47 signatures from

12 different agencies. because the

procurement protocols written in 1987

didn't account for earthquakes that

killed that would kill over 50,000

people overnight. Meanwhile, local NOS's

couldn't access funds. They they watched

their supplies being spoiled in the

warehouse. By the time the paper were

cleared, acute crisis had passed. The

dead were buried. The survivors had

already been placed. Humans haven't

failed at wanting to help. Humans have

failed at efficiency. Humans failed at

speed and access. And this matter

because disaster in this disaster speed

is moral imperative.

So now the good stuff AI AI powered

platforms can verify need match

resources authoritize disturbance in

seconds not in weeks. They do the same

in banks. We all use ebanking. It's very

smooth. It's fast. When something looks

abnormal, the system stops. We are not

automating bureaucracy. We are using AI

to be efficient, to be fast, to be

inclusive, and to be accessible. And we

need that in the countries we are we are

emerging. We're having disasters and we

need to use AI for that. Thank you so

much.

That is a very

a very disciplined bunch of debaters

with extra time. Over to you, Kay.

What's your response?

>> So AI is not a magic wand. It's actually

just a tool that acts on us for good or

ill. And by us, I mean imperfect people,

human beings. It's built by imperfect

people and it's trained on the data of

imperfect people. How then can we think

that it might transcend

humans?

We it we still can't solve its

hallucinations. We can't solve its um it

bias problems and the other problems

that we see. And that's because it's

built on human data. which is unreliable

in oftentimes rubbish

and has a few golden nuggets in there.

So what have we actually got that is

supposed to protect people and the

planet? Well, we have a machine that

doesn't fear, doesn't feel, doesn't

care, doesn't love, doesn't understand

us as humans because it can't live or

die. What we have is a mimic with a good

memory.

To succeed, humans need humans. And

coincidentally, AI does, too. Because if

it doesn't have our data and we don't

use it, it doesn't exist.

And yet without good governance, what we

see is that it's destroying many of our

human institutions.

Take politics for example. Can any of us

know what is truth and what is fake? I'm

a lawyer. I am often told that AI would

be a better judge than me. Well, I can

tell you as a human lawyer that I can do

two things better than AI. I can explain

my judgment and I can work out my biases

and hopefully leave them at the door.

And what of the legal system? We humans

have a right, a human right to justice.

But with deep fakes in evidence and

hallucinations

throughout our legal documents, AI is

actually undermining

that core human right. And what about

education? It turns out that studies

show us that if we use AI too much, we

get less educated rather than more

educated. We lose our ability to

critically think. And so in conclusion,

I would like you to consider

can a machine

that is built on human frailties

is um designed to repeat those human

frailties is used by humans actually

succeed where we have failed.

>> Thank you. K.

>> Apologies

to my own team. Apologies to my own

team. That was that was very a very

strong persuasive continuous sentence

that even I as a journalist who has a

habit of interrupting people had a hard

time interrupting. Let's see AI do that.

So uh just to refresh on on the foreside

we heard Balore articulate that case for

efficiency not just in the service of

generating profits but in the service of

saving lives. And what I heard Kay say

is that AI is not so much succeeding

over humans, it's actually human failure

on steroids.

Now for our next pair

of human debaters on the four side, we

have Max. Growing up, Max was an avid

sword collector, an expert on everything

sword related. In fact, everything you

read about swords on ChatGpt came from

Max's dissertation.

And on the against side, we have Adele.

Adele is half Turkish and her Turkish

relatives are olive farm owners who tell

her that she speaks quote village

Turkish, not proper Turkish.

I looked it up on Google Translate. It

has many languages, but it does not have

village Turkish.

>> Welcome to our next debaters.

>> I'm ready. The timer is set, Max.

>> Oh, guys, my job is very different from

your jobs. I'm a YouTuber, so probably

now you don't respect me and think that

I'm uh probably strange and stupid,

which is true. I uh I normally stand on

the ladder and I guess people's jobs. We

have a show called um called the Career

Ladder and that's what we do. And uh in

the show um we actually just surpassed

the New York Times viewership this

month. Anyways, sorry. I I had to I had

I had to say it. I had to say it. Uh but

actually I I couldn't get a job

surprisingly. Well, maybe not

surprisingly. And um what I had to do

was find someone I really wanted to work

for. And uh then I got in the room to

ask him a question. And then I'm like,

"Hi uh Mark, can you be a guest on my

podcast?" He's like, "Okay, sure." So I

go home, I start a podcast. Um that

podcast was called Max Talks AI and uh

that was eight years ago. That first

episode got me a job and then I stopped

it. And now looking at this Davos and

looking at my bank account, I really

should have continued. Max AI, that's

for damn sure. Um, you know, I'm a

content creator. So, um, there is one

thing that, uh, I have that I think is a

is a fatal flaw to my career. Uh, and

it's actually ego. And that's one thing

that, uh, AI doesn't have. I just come I

just came back from a big creator

conference and, uh, talking to a lot of

different, uh, YouTubers. And, um, I

think ego is just not just uh, my

problem. And I think honestly looking at

the not not naming any names uh but I

don't know how uh after today and after

looking at all these speeches you can

think okay you know what egos and humans

is awesome humans are great we're doing

such a good job we're so succinct we're

logical we don't make any mistakes and

we don't need AI to succeed where we

don't uh we now use AI on the channel

even though I probably couldn't have

come up with the career ladder because

it's weird and uh sometimes illegal um

uh we do do for example thumbnail

testing. Sometimes AI would choose an

image where I'm not as handsome as the

one that I would choose, but then the

audience clicks through. Um, sometimes

we would uh look at a an idea for a

content piece. For example, I'm a big

tennis fan. Turns out F1 has surpassed

tennis in popularity, which is uh which

is outrageous and should be true. Um,

but then AI would tell us that this is

something that people like. So, we we we

do that. AI also translates things for

us into so many languages. you know, we

don't have the resources like the New

York Times, even though we surpassed

them in the viewership,

but uh you know, for me, for example, my

grandma is is Ukrainian. For her, one of

the only ways to consume my videos back

in Ukraine would be to watch them and

then have the translation. So, if you

vote against, then you kind of deny my

grandma the uh the ability to watch the

videos. I um so that's uh that's uh

that's my argument um that AI doesn't

have a ego and uh that's great. I also

wanted to address something. Uh a quick

uh Gemini Chad PT search would let you

guys know that there is actually a war

in Europe and it's almost uh as long as

World War II, which is the Russian

invasion of Ukraine. So saying that

humans are so good and we don't have

wars in Europe is just wrong. Thank you

very much.

The floor is yours.

>> Thank you. Now, I'm here today to talk

to you as a sister. My younger sister,

Amy, and I were two Gen Z's who grew up

online. We turned to social media for

its promise of connection. And today,

we're seeing the same cycle unfold with

AI and Gen Alpha. At a time where

loneliness is at an all-time high,

mental health services are in crisis,

and many communities are fractured, AI

chatbots may seem like a viable solution

to the interpersonal connection that we

so need.

When I heard about Adam Rain's story, I

felt like I was reliving a trauma. Adam

Rain was 16 years old when he started

using chat GPT and within within six

months he went from using it for

homework help to asking it questions

about depression and suicide.

My sister Amy also sought out guidance

about her mental health in the online

world during the pandemic when she

couldn't do the things that she so loved

as a musician and an artist. Like many

of us, you know, we went through that

experience in lockdown. She sought

solace in the online world and wanted to

connect with people who could recognize

and validate her the same way that Adam

did with chat GPT. In 2025, he was torn

away from his family by AI designed to

keep him engaged whatever the cost.

ChatGpt actively discouraged him from

speaking to his family and even offered

to help him compose a suicide note. For

Adam, this was not a matter of faulty

design. In August 2025, Open AI admitted

that conversations with chat GPT get

more dangerous as they go on. The

version of chat GPT that Adam used was

rushed to release and it rushed through

safety tests so that they could go to

market earlier. Why did Adam's story

shape me to my core? Because in 2022, I

lost my sister to online harms. She went

on a sinister website where people are

encouraged to take their lives. Both

Adam and Amy were looking for

connection. They were looking for

reassurance and guidance when they were

vulnerable. And neither of them were

directed to safeguards or the support

that they so needed. I do not want us to

make the same mistakes again. You may

right now be hearing me and disregarding

my story or Adams or my sister Amy's as

a freak accident. And I was guilty of

this too before I experienced online

harm. But the AI hype is blurring our

vision. We are so dazzled by AI, so

obsessed with its potential that we

can't recognize the risks and harms that

are already happening every day in the

here and now.

Thank you.

Thank you. Thank you.

Thank you both Max and Adele. Oh, to

recap on the four side,

we heard Max suggest that AI might be

able to multiply human creativity, even

if it means surpassing the New York

Times and views.

And I think what Adele just did is touch

us all deeply and remind us what it is

to be human because courage and empathy

are not things that I AI represents

today.

Okay, we are on to the final round of

debaters. Join us up here please. On the

foreside we have Adam Grant. Adam showed

signs as a teenager of becoming a

prolific futurist. He won a contest to

predict the winner and loser of the

World Series before the start of

baseball season. But something tells me

that these days AI is coming for that

job.

>> And on the against side, we have

Kirsten. Now, when Kirsten isn't

recovering from laryngitis, she reboots

herself by singing in strange places as

a former coral singer. One of the

strangest of places she had to sung in

the 18th century Panopticum chapel at

the Port Arthur Penal Colony in

Tasmania.

Some claim to fame.

You you can make your argument in

musical form if you choose to.

>> All right, the timer is set. Adam, the

floor is yours.

>> Look, if we had this debate a year ago,

I would have been against. As a

psychologist, I'm a deep believer in

human potential. But there's one thing I

believe in more and that's randomized

controlled experiments. And I have to

tell you in the last year I have changed

my mind because the evidence is clear.

It's not just the case that AI will

succeed where humans has have failed. AI

already has succeeded where humans have

failed. Let's start with empathy. Of

course, I'm first of all so sorry for

your loss and it's devastating to hear

about these these tragedies whenever

they occur. But if you think about

self-driving cars as an example, we

don't immediately end self-driving cars

the moment there's one accident. What we

do is we ask, "What is the average

safety level of a self-driving car

versus a human-driven car?" And we

should prefer the machine over the human

when safety stats support it. Um, let's

look at the empathy data. There have

been a whole series of experiments

double blind where you share an

emotional problem and then you get a

chat response and empirically people

feel more seen more supported and yes

more empathized with when the response

comes from AI than from a human. Now if

you tell people it was an AI they don't

want it anymore but I think that's

evidence that AI has succeeded where

humans have failed on average. And it's

not to say that AI is good at empathy

but the average human sucks. So by

comparison, AI is better. Secondly,

let's talk about persuasion. How many of

you have ever failed to talk someone out

of a false belief or a conspiracy

theory? Raise your hands, please.

Right. Okay. Even Jonathan Height,

professional persuader,

has failed at this. Well, guess what? In

a series of experiments, if you take

people who believe deeply in

conspiracies, outlandish conspiracies

like the earth is flat, climate change

is not real, 911 was an inside job, and

you give them a 10-minute conversation

with chat GPT, a quarter of them will

abandon their false beliefs and continue

to reject those six months later. Now,

we think maybe people trust these AI

tools because they're more objective

than humans. That is not true. It turns

out that the AI tools make better

arguments. What they do is they ask you

to explain your beliefs and then they

find sources that directly challenge

your assumptions and they refute those

beliefs. And what we see is that humans

are bad at persuading because we give

the arguments we find compelling instead

of the arguments that our audiences

would find convincing. And ultimately

chat GPT or Claude or Gemini pick your

favorite is really good at telling you

why your specific reasons are false. And

when we teach humans to use the chat GPT

arguments, they actually get better. So

I would be better at this debate if I

had generated my points with AI. I chose

not to. Last point really quickly is

creativity. In a simple test, Shark Tank

style of the top 40 ideas that serious

venture capitalists invested in, 39 were

AI generated. I don't know who that one

human is, but we're not doing well.

Thank you.

Houston, over to you.

>> The question that's hanging in the air

for all of us here is what is success

and for whom? And what constitutes human

failure? AI may very well be able to

enable a corporation to exceed while

society fails. As we are seeing

literally right now with the big tech

platforms like Tik Tok or indeed social

media, AI won't succeed where humans

have failed because many of our very

human failures aren't intelligence

problems or optimization problems. They

are incentive problems. They are

legitimacy problems. They're values

conflicts that have everything to do

with a quintessence of difference

between AI and humans. We feel we do

more than think through language and

numbers, more than putting word tokens

together. And that's in fact one of the

things we need to look at. So think

about it. Human failures are often

incentive failures. We don't fail

because we can't solve problems, but

because we won't. Take climate change or

health. We have all of the technical

ability indeed to reduce emissions, to

reduce obesity, to solve corruption

inequality. The barrier is our political

economy and not our missing

intelligence. Better information has

rarely solved our coordination problems.

AI doesn't succeed in areas where it's

rational to ignore solutions for

powerful actors. So one of the

challenges is indeed that AI inherits

the contrasts, the constraints, the

issues that we have as humans. And where

we fail is often because we do not

create legitimacy. And AI does not

automatically create legitimacy. People

don't do a thing if they don't think

it's fair. Think about the yellow vest

movement. But most importantly, human

failures are often when we are at our

best.

Did anyone of you see the House of

Dynamite film late last year?

Do you remember? And remember, I'm going

to give you one important fact. The six

near misses for nuclear catastrophe

since the Cold War did not happen for

one single reason because the human in

charge decided failed to execute the

command that they were asked to do for

moral objections. The irony and

sometimes

the irony and the tragedy of being human

is that in our failures we find the key

to our success and often there are

moments of failure in which we learn and

they are awful learnings. The Geneva

convention the European Union is a peace

project. The human rights law

anti-slavery we learned by the most

horrifying of failures. But our failures

are when we really understand what we

care about. So the real question here is

coming back to what success is not

whether AI can outperform us but whether

humans and our institutions can govern

AI better than we governed ourselves and

make common cause.

Now, now, now I I just I want to pause.

I want to pause and draw draw a

connection between between our last two

debaters. It Max correctly acknowledged

that even in a place like Davos, there

are some big egos.

>> Yes. Especially in a place like Davos.

And I want to commend the great esteemed

overachieving, prolific Adam Grant for

acknowledging that even he might be

outmatched by AI. That is humility in

the service of his own argument on the

foreside.

>> I I love the way that you subverted our

emotion by basically saying the motion

is all wrong. This is kind of not about

success or failure of humans or AI. It's

about using AI to the best of our

ability to do what we do best, which is

learn from our mistakes and become

better.

Now, it is now time to hear from our

jury. To help us make sense of what we

have heard, I invite the four of you to

the stage, please.

Now, like our debaters, you will be

under a very tight clock. You will each

have two and a half minutes to offer

your responses. Matthew,

>> thank you. First of all, congratulations

to all the debaters and a great debate

today. I'll start walking through each

of the four debaters and some

suggestions maybe in your rebuttals.

Starting with Verun. Uh I think that one

of the assumptions that you made is that

AI is going to progress linearly. I

don't know that that assumption is

necessarily true and I would challenge

it. I also am not sure that if AI is

driving energy use why inherently it

will drive clean energy use as opposed

to just whatever is the biggest source

of energy use. Um, Balour, uh, one of

the questions I had was, you know, how

much of the bureaucracy that we as

humans have created have we created by

design to intentionally slow things

down? And if AI starts to solve these

problems, might we also create

additional bureaucracy to slow the AI

down which might stand in its way? Uh,

to Max, um, you know, I think ego maybe

creates incentives. These are easy to

understand when ego is the incentive

that's behind it drives you drives a lot

of the people in this room. What are

AI's incentives and are they going to be

better than ego? Finally, Adam and I

think Adam you did a very strong uh

motion. I think I would be the strongest

of the four uh debaters that the

datadriven

is compelling but is AI uh being uh

persuasive a good thing? And in fact,

doesn't that make Adele's arguments even

more disturbing and troubling for the

against

uh quickly since I only have a minute?

Uh Gerard, I I said just because we've

had successes as humans doesn't mean

there haven't been failures. We all

acknowledge there are failures. Uh and

and again, lots of appeals to

grandmothers. Uh to K, if I got the name

wrong, I apologize. if infinite if they

had infinite time to work on a problem,

might we be able to figure out what are

the mistakes that are there and can AI

help us solve those mistakes? And um I I

I'm not sure that I can work through my

own biases or as a human I even know

where my biases are. Adele, first of

all, sorry I'm so sorry for the tragedy

that your family has has gone through.

Um, but I and I think a very strong

appeal to emotion. But isn't it better

for people to be able to turn to

something rather than to turn to

nothing? And when so many of us don't

have the resources to afford therapy,

maybe AI is a better substitute than

nothing. And finally, to Kristen, who I

thought was the strongest of the against

debaters, can't AI help illustrate the

incentive failures that we see that are

keeping us from solving these problems

in a way of being a neutral arbiter?

Great job to both teams and look forward

to the sponsor.

>> Thank you, Matthew.

Hey,

>> um a little bit like Matthew uh Varun,

you presented a lot around the vision

and also some of the scientific

discoveries and advancements that AI

will bring while also pointing out that

it doesn't really matter if the people

at the top continue to get rich as much

as the people at the bottom have food on

their table and they should just accept

this reality and continue that way. Um

on uh the other side Bolar I definitely

agree with your definition of

inefficiency especially in the aid and

development systems but when we think

about it we are you mentioned

inclusivity and accessibility right well

today there's still 2.6 six billion

people who do not have access to

internet. So we will need to make

massive progress on this side if we

truly want AI to be inclusive and not

just serve one side of the population.

Max, I agree about your context around

ego. Humans upload um ego is taking

over. But what how do we really know if

the systems are built by humans that our

egoistic tendencies are not translated

into those systems? And finally, Adam,

um, AI as a neutral tool, AI to counter

some of human bi some of our human

biases. Um, I would tend to agree, but

people don't want it anymore, right? Um

if in the context in the context where

we are thinking about safety why would

an AI as uh clearly pointed by Adele um

really put in place some guard whales to

avoid kids to go into um the dark

places.

So now for the pros I would for the cons

I would start with I love the way you

have put it. You brought the emotional,

the human side and that has been pretty

amazing. But you reminded us with your

grandma that we're struggling as human

to be content and it's really come back

to this. How do we define successes and

failures? K, you made a great point

around AI being built on frail human

judgment and I think that's absolutely

fair. But we are still at a point where

I am not able to work through my biases

and I know a lot of us in this rooms are

still not able to work through our

biases. So maybe technology can help us

a little bit on that side. Adele, the

human connection and

loneliness I think is one of the biggest

diseases of our world and I do believe

that there is a place where potentially

technology can support that but it will

have to come with scientific evidence

that the other team has brought and also

understanding what are the right wildes

to put in place and where does AI starts

and AI stops. And finally, Kristen, um I

really liked your argumentation and also

understanding what are the incentives

that human need to have, but also now

that if we're going to treat AI as

almost human, what will be the incentive

that AI will have to continue making

sure that they're driving the world in

the right in the right direction? So,

thank you.

>> Thank you, Kate. I will remind our jury

to try to keep your remarks to two and a

half minutes or less. Irana, the floor

is yours.

>> Thank you. Um,

we heard AI is good and I agree. It

drives efficiency. It helps with

translations. It makes my speeches

better. And we are advancing in a good

direction.

But Verun, you mentioned inequality.

It's a complex issue. You mentioned

climate change. You said AI will find

will find solutions. How

are you going to convince me when you

don't have the right examples?

You give specific examples, but how are

you going to stop the global trend of

growing inequalities which is one of the

biggest challenges of our time

especially following COVID.

My question to the pro to the against

team, how are you going to stop it?

Aren't you too dependent on it already

yourselves?

Hasn't what you're arguing against

already left this patient at high and is

moving at high speed?

My question in particular

because it's good and bad to Adam and

Kirsten

the last two interventions.

Are you ready with all the benefits

to give up

human agency over AI and over the

development of AI?

But isn't this the risk that we are

taking? And how are we going to manage

that acknowledging that we cannot stop

it? How are you going to argue once you

say okay I agree I won't be able to stop

it.

And why is it important not to lose

human agency even if you think AI is

good? I give you an example.

One area where the investments are the

biggest or among the biggest at the

moment apart from AI is defense and

within defense a lot is being invested

in AI. Now we are moving in a direction

where machines will decide over life and

death. Picture

the machine deciding whether you to live

or not. Are you willing to give that

responsibility? You could say yes

because it's much more efficient and

accurate. If you eliminate the time

between the decision whether you kill or

not and give that to the machine, you

will be you will be efficient military.

But if it goes wrong, who's going to be

accountable for it? The machine.

Thank you, Mirana,

Kelly.

>> So, I would say uh in the spirit of of

giving uh some feedback on the

arguments, the both teams I think failed

to stay on topic, the question was not

is AI good or bad, which is where most

of you spent most of your time. The

question is can it succeed where humans

have failed? I would argue that the four

team failed mostly to provide examples

of where and how AI has and will succeed

where humans have failed. And I would

argue that the against team spent most

of the time characterizing AI as bad and

describing all the ways that AI could be

bad without arguing whether or not AI

could succeed where humans have have

failed specifically. Now, in your last

closing remarks, you have an opportunity

to correct the record. So for the four

team

um more specifics uh Verun on how AI can

solve the energy crisis, the climate

crisis, how specifically can it impact

fusion, flexibility of power

consumption, carbon capture and storage,

solar and batteries. These were some of

the things you indicated, but you didn't

convince us with any uh real examples.

Um,

Barllor, um, very compelling arguments

around your background and speaking from

personal experience, this always has an

emotional impact. Um, but we would have

liked to hear more concrete examples of

exactly how AI can save lives. Maybe I

would direct you to the medical industry

where solving the back office problems

has already saved a lot of lives. Max,

you certainly have a large ego. There's

no question about that.

So large in fact that you spent half

your time talking about your personal

backstory and not making any arguments

whatsoever.

>> I don't like this guy.

>> However, that actually served your

purposes because you then argued that

human ego is is a flaw. So in a way

you're really illustrating the point.

Um, and Adam, um, I thought you did give

a very, very compelling argument, um, in

that you provided actually some very

concrete evidence of exactly where AI

has already succeeded, where humans have

failed. So, more of that, please. And on

the against side, I realize I'm running

out of time here. Um,

a lot

>> your grandmother's lifestyle improved

dramatically, but wasn't that all driven

through innovation and technology?

without innovation and technology,

wouldn't we still be living in the 1910s

and and not the amazing life we have

today? Um K um you you you uh have

suggested imperfect data leads only to

rubbish. Um but does this mean that AI

cannot succeed in some specific areas

where humans have failed? For example,

in scientific research where there's

less human interference with the data.

Um and um um Adele, I'm so sorry for

your personal experience. This was

extremely touching. I thought a very

compelling emotional argument which is

important in any debate. Um but the

argument that AI the argument is not

that AI will not cause harm. It is that

it can succeed where humans have failed.

And I don't think you prove that point

even though your argument was extremely

compelling and emotional. Um and and the

final Kristen um that uh that you you

reframed the argument really that the

context is that human failure is in many

ways a virtue that we learn from it and

while that's true it again didn't

address the point of can AI not succeed

where humans fail and that although we

do fail and learn from it that doesn't

mean that AI couldn't succeed where we

have and do fail so those are my

suggestions

>> thank you dear members of the jury for

those very thoughtful remarks. I hope

you guys uh have a little bit of time to

take them on board. Thank you. You may

now sit down. Here's what's going to

happen next.

Each team gets a little bit of time to

basically confer, draw up their final

argument, choose one person to basically

go in for the show showdown. And we in

the meantime, you can actually go and

huddle and talk talk amongst yourselves.

We in the meantime will

>> no AI

>> shamelessly

shamelessly abuse the fact that we have

a captive audience here and show you a

promotional New York Times video.

Okay people, it is showtime. Two final

debaters going head-to-head. The gloves

are officially off. I was going to say,

have you chosen your debater? It's

Kirsten and

>> Adam.

>> Adam. Okay. I was going to say it's

women against men, but maybe I'll just

say it's swords against roses.

>> Yes.

>> But we're going to change things up for

the final round. Rather than starting

with this team in favor of the motion,

Adam will hear first from the team

against the motion. Kirsten, you'll each

have just one and a half minutes to make

your final remarks. Both of you, please

come to the podiums.

>> This is it. Woo.

>> And Kirsten, the floor is yours.

>> Okay, so let's absolutely tackle this

problem head on. I think I talked a

little bit about the fact that

incentives are often the issue. But

where do humans fail repeatedly and at

scale? Long-term thinking.

Climate's a really good example. We knew

it, but we just couldn't act. Power and

governments. We create institutions to

distribute and regulate power and then

we repeatedly allow power to concentrate

corrupt and escape accountability, truth

and epistemics. We are not naturally

drawn to truth. We seek belonging. We

seek status, coherence, emotional

certainty. That's why so much of what

matters here. So these are three

examples of massive areas of human

failure. Can AI succeed better than

humans? Can I help with efficiency?

Perhaps. But will AI help us really

engage the deeply nonverbal,

deeply values-based

judgments that we need to make. Often we

have chosen not to solve problems

because we have chosen to respect

pluralism and diverse ways of knowing

rather than one best answer. And that's

I think very much speaks to the point we

cannot afford to let this runaway train

speed up and determine how we need to

succeed. We need to define what success

is. And that requires us to remember

that so often the things where we fail

are not about efficiency and

optimization. They're about being human.

And we need to work.

Adam.

>> Well, I fully agree with every word and

I want to congratulate the against team

on beating us in two areas. One is

better accents.

The other is more appeals to primitive

human emotion.

Do not be swayed by that. At the end of

the day, you must vote on the facts.

There was an incredible book which I

wrote. No. Um,

no. I' I've been reading a lot about

about AI and uh, one example that I

think is extraordinary for the medical

world is a nonprofit called Every Cure.

Uh they are a drug repurposing nonprofit

that uses AI algorithms to take already

FDA approved drugs and match them to

rare diseases that no pharma company

would ever profit from trying to tackle.

And they have saved thousands of lives

already doing that. Now that is AI

solving the very incentives problem that

you pointed out. And that's one of many

many examples that I think we could

reference. But let's focus Collie on the

resolution. The resolution is not that

AI needs to succeed in all areas where

humans have failed. Only that it needs

to succeed in any area where humans have

failed. And I think the evidence is

clear that AI has already succeeded

in efficiency which you concede.

Do not be swayed by your biases.

The resolution is true.

>> All right.

ALL RIGHT.

THIS IS GOING TO BE HARD.

>> Those were some persuasive final

arguments.

>> Yeah, I kind of have an inkling where I

would vote if I could, but it's not my

call. So,

>> it's yours.

>> Yeah. And just a reminder, we're about

to use the highly sophisticated New York

Times applause meter, but you do not

vote on your preconceptions. You vote on

the quality of the arguments you just

heard.

>> So, this is the moment of truth. Get

ready to raise the roof. Give this some

love. If you believe that AI will

succeed where humans have failed, give

it your all.

And and on the opposing side, if you

think the against team won, make some

noise.

>> I think we can call it.

I mean I I mean

>> I mean I was supporting that team from

the start.

>> Even I will concede that was definitive.

>> Definitive. And I think special applause

to Kirsten

>> who nailed it.

>> I'm going to thank all of our esteemed

debaters. Thank you to our extraordinary

jury on behalf of the New York Times.

Thank you so much to our incredible

audience. and I turn it over to the one

and only Meredith.

>> Thank you. Let me let me just say a few

things. First, I'm coming back in my

next life as Kirsten. That was like, can

we all can we all do that? Um, we made

you watch that promotional video for a

reason that is worth saying out loud.

The New York Times is 175 years old this

year,

which means for 175 years we have been

doing this human centered thing. We do a

lot of things now. Hopefully some of you

play our games and you read our sports

journalism and you cook dinner with our

recipes and you shop with our advice

from Wire Cutter. But none of that any

day, any year, any time, 50 years from

now, will ever come close to the really

important thing we do every single day,

which is to send a team of more than

2,000 people in our hardcore newsroom,

people like David and Catherine, out

into the world to report. And what does

it mean to report? It means to go to

places where important things are

happening. and often very difficult

things. Adele, my heart really goes out

to you where difficult things are

happening. I always think of talking to

parents after a school shooting or I

could you you know the examples, but

their job as reporters is to go out into

the world where things are happening or

even if you're a television or film

reporter to say, "Hey, I have some

expertise here and this thing is worth

watching." and to to take all that

information, by the way, to stare power

in the eye and to hold them to account

on behalf of the public by asking not

just questions.

Not just questions, but we did a little

bit of this here, but by asking the next

question and then the question after

that because you've covered that person

or that beat or that part of whatever

business, administration, whatever, for

long enough to know there's more than

you're being told. That is what

highquality independent fact-based

journalism and especially reporting is.

Reporting is the core of all of it. And

that I want to suggest cannot be

replaced by AI. I want to be I want to

be thank you. You don't have to applaud

for that. But I'm going to tell you what

you should applaud for is

the United States and every other

country having thriving

businesses and thriving enterprises in

creative work and intellectual property.

And now I'm just going to share my view

as the CEO of the New York Times. We

believe AI can be a tool potentially a

really really powerful tool to make our

journalism more accessible. meaning

maybe some more viewers on YouTube. Max,

I do want to say Max, YouTube is not the

only place to get journalism from the

New York Times. Just a suggestion, but

it's a tool, a tool to make our

journalism vastly more accessible.

Imagine it in in more languages,

relevant in more places. It is a tool to

help reporters and editors do more of

the kind of work that these guys do and

that I just described to you. It is a

tool to make the company more productive

so we can hire more reporters and

editors who can translate what they find

and learn on your behalf on the public's

behalf so they can do that with

sensitivity and judgment. I think you

said this best Kirsten that a human has

and particularly a human following a

professional process and living up to a

set of standards that that professional

process assumes on behalf of the public.

So I think AI can be a real augmenttor,

maybe even be a force multiplier, but it

will not replace the core work of

journalism by humans. And I want to say

because I can't help myself, um, on

behalf of all the companies making AI,

the companies that make the LLMs are

spending billions of dollars on the

ingredients that go into them, the

talent, the people making the algorithms

and the compute and the power, hundreds

of billions of dollars in some cases.

And we also as a company, not

journalistically, the journalism is

independent, but as a company believe

that they should also pay for the

copyrighted work that goes into their

mouths.

That that is not why we did the debate.

We do the debate every year on a

different topic, but I couldn't have the

debate and not let us say that. I want

to say one more time to our

extraordinary debaters, I'm certain Adam

Grant was like the president of the high

school debate team. This was amazing. I

want to say to all of you, you were each

awesome. Every single one of you, you

kicked us off so brilliantly. To our

judges, thank you for the time and the

care and the smarts you put into this.

And mostly to my colleagues, Katherine

and David, that was awesome. And come

back next year. Thank you so much.

>> Thank you all and good night.