Category
🗞
NewsTranscript
00:00We have one of the most exciting panels here in Davos.
00:05We have an incredible group of people on stage.
00:08We have the folks who control the path of the next pandemic,
00:12much of the world's economy, whether the West survives.
00:16And based on the logs, how my kids are doing this homework,
00:19doing their homework this week while I'm in Davos.
00:21So we have a lot of influence right here.
00:24We're gonna talk about democracy,
00:25we're gonna talk about income inequality,
00:26we're gonna talk about climate,
00:27we're gonna talk about how companies are transforming.
00:29But first, we're gonna start with the Secretary General.
00:32First, let me just do quickly introductions.
00:34We have the CEO of Pfizer, CEO of Uber,
00:36CEO of Anthropic, Vice President of Alphabet and Google,
00:40and the CIO of the CEO of Salesforce
00:42and the Secretary General.
00:44We're gonna start with the Secretary General.
00:46Secretary General, how are you?
00:48I'm fine, how are you doing?
00:49Wonderful.
00:50And how's the group doing?
00:50Good to see you all.
00:51Wonderful.
00:52First question for you.
00:55A lot of technological change.
00:57It's been interesting to watch the war in Ukraine,
00:59where in fact, one of the most interesting technologies
01:01that has been most impactful has been drones.
01:05Tell me, as you see military changing,
01:07as you see AI coming,
01:08what to you is the most important change
01:11that will affect the balance of power in the coming years?
01:14Well, clearly at the moment, it is the drone technology,
01:16which is really changing the way we conduct these wars.
01:21You see now $400 Ukrainian drones
01:24taking out multi-million dollar,
01:28in terms of cost, Russian tanks.
01:30We just launched Baltic Sentry.
01:33This was a mission, is a mission,
01:35or an activity, we should say,
01:36in the Baltic Sea, in the East Sea,
01:38to fight off the Russians
01:40who are getting at our critical undersea infrastructure.
01:43We are using sea drone technology there,
01:45next to the more traditional technology with ships, et cetera,
01:49and aircraft, but also this drone technology here.
01:52And obviously, AI will also be transformative
01:55in the terms of how we fight our wars.
01:57Of course, the question I'm getting a lot
02:00is that I'm really pleading for more money
02:03to be spent on defense.
02:04And then people are worried that we will spend that
02:06in the same way as we did the last 100 years.
02:08But that is not the case.
02:11Yes, we have to spend more.
02:12We are spending now, on average here in Europe, 2%.
02:14That has to be much more.
02:16I think Donald Trump is right here
02:18that we are spending not enough.
02:20And anyway, we have to get it into a balance
02:23with what the U.S. is spending.
02:24But particularly here,
02:25we have to look out for the industry base.
02:29So what our defense industry is producing,
02:32and they are not producing enough,
02:34but also in terms of innovation,
02:36we are too slow in innovating.
02:38One of the problems here we have
02:39is that the better is the enemy of good.
02:42It all has to be perfect.
02:43But it doesn't have to be perfect.
02:45When Ukraine is fighting this war on a one to 10 scale,
02:48they can settle for a six or a seven.
02:50But in NATO, we can only bring something out
02:52if it is a nine or a 10.
02:53We don't have that luxury anymore.
02:55So speed is of the essence,
02:56not perfection to get these new technologies in.
02:59And if we don't, then we have to spend even more.
03:02So joint procurement, yes,
03:04we have to get the big contracts in.
03:06But also in terms of the innovation,
03:09if we not innovate faster at a higher speed,
03:12not achieving perfection,
03:14but getting speed and enough quality done
03:18in the right conjunction,
03:19then we have to spend even more.
03:21And it is already a big drain, of course,
03:23on society's defense spending,
03:24and we need more of that.
03:26So in that sense, technology is a crucial factor.
03:28It's a crucial factor.
03:29Well, the speed of transformation
03:31is actually a wonderful segue,
03:33because when I look at this panel,
03:34which covers so many diverse areas of the world,
03:37one of the things that most people have in common
03:40is they are running businesses
03:41that are going to be radically transformed by technology.
03:43So Mark, I want to go to you.
03:44One of the terms that people often say
03:46is creative destruction.
03:47How do you take a company that is very successful
03:49at one thing and radically transform it?
03:51You've been talking about doing that at Salesforce.
03:53Tell me, what is the hardest thing
03:55about the transformation you are going through right now?
03:57Well, I really appreciate it, Nick.
03:59I mean, I've never been more excited
04:00about my job and about my industry.
04:03I mean, it's incredible what is going on right now.
04:06It's really a phenomenon.
04:07This is just in a moment that none of us will ever forget.
04:10And I'm just thinking about this,
04:12talking to so many of my friends here
04:15at the World Economic Forum,
04:17and we were just in the IBC,
04:18the International Business Committee,
04:19which was 250 CEOs, and you were running a panel, Nick.
04:22And I was saying, I said to the group,
04:26we're going to be the last CEOs,
04:28we are the last CEOs,
04:31who are only going to be managing humans as our workforce.
04:34That from this point forward,
04:36like this is the marking that we will be managing
04:40not only human workers, but also digital workers.
04:43And that is just incredible.
04:45And I just look right here at Davos.
04:48Now, Salesforce runs all the information management
04:52for Davos, it has for more than a decade.
04:54So of course, when you're on the Davos apps,
04:57and you're going through all the lists of all the sessions,
04:59and who's attending and all that,
05:01and you figure out how to make
05:02the most productive use of your week,
05:04you're using all these Salesforce technology.
05:07But this is the first Davos that you'll notice
05:09that right on the Davos app, there is an agent.
05:13And this AI agent is there to help you
05:16to be your guide, to be your partner
05:18in making sure that you're able to have a great Davos.
05:21And what it does is it goes back
05:22and it looks at every session you've ever attended,
05:25because all of that is recorded
05:27as you badge through the conference,
05:29and every session that is available,
05:31and everybody that's here.
05:32And then that agent can say,
05:34yes, this is the thing you need to do in Davos
05:38to be really successful right now.
05:40And it uses the best large language models,
05:43but also the best machine learning,
05:44the best intelligence,
05:46it delivers the highest accurate result.
05:48It's delivering a 95% result for Davos attendees
05:52because it has access to so much data
05:56that has been collected here over so many years.
05:59And it's a great example of very practical AI.
06:04And I'll say another example is that
06:05we also met in the same meeting,
06:07Dave McKay, the CEO of RBC, World Bank of Canada.
06:11And he's got this incredible wealth management business
06:15that he runs, and he's a leader in this area.
06:17And they've built this agentic layer.
06:20Agentic layer means he's built agents around his business
06:24that augment his workforce.
06:26So in addition to his sales workers,
06:28and his service workers, and his support workers,
06:31he uses an agent force,
06:33just like we're using here to run the Davos.
06:35And so that's just fascinating to me,
06:37that we are at that moment that we're gonna say,
06:40this is when humans and AI are working together
06:45to create a higher level of success.
06:47I really appreciated your agent, Mark, actually,
06:49because I was out on the promenade at 2 a.m.,
06:51and I'd had seven drinks.
06:52And the agent, having based it,
06:54studied all my previous Davoses,
06:55said, stop, and sent me to bed.
06:58So thank you very much.
07:01Ruth, you have this company that has built
07:02one of the most successful products
07:04in humankind history, a search engine.
07:06It's gonna have to transform, in many ways,
07:08into an answer engine.
07:10Building a business model around that is hard.
07:13It will change the way you make money,
07:14and it will change the way a lot of companies
07:16that relied on the search engine make money.
07:18So explain the trade-offs as you make this transition.
07:22Well, first of all, it's great to be here with you,
07:24with everyone.
07:25Since inception, Google has been focused
07:28on long-term investing with a sense that technology changes,
07:31and we're gonna have to evolve throughout time.
07:33And if you think about it,
07:34Google started as 10 Blue Links.
07:36It then went to text search.
07:37It went to voice search.
07:38We're now living in a world with multimodal search,
07:40but we've been infusing search with AI for many years now.
07:45Billions of people are benefiting,
07:46whether it's as simple as autocomplete or searching photos.
07:49You just take it for granted.
07:51And multimodal search is exciting.
07:54What you can do with this product we call Lens,
07:57for those of you who haven't used it,
07:58you take your phone with the camera,
08:00you can take a photo of anything and search through that,
08:03find any information, product, availability, price,
08:06anything in nature, you name it.
08:0820 billion searches a month through Lens.
08:11So this has been a constant evolution,
08:14and where we are now, leveraging Gemini,
08:17is looking at AI overviews,
08:19which you can think of as sort of an umbrella,
08:22your cockpit, that then helps guide you
08:24and takes you to different places.
08:26And what we're seeing is more complex queries,
08:30longer queries that go into greater depth.
08:34It's expanding in terms of demographics,
08:37the youth in particular are looking at the ability
08:41to ask multi-part queries.
08:43And so it's continuing to enrich the search experience,
08:46which is what Google has been about all the time.
08:48And our investment, I'm frequently asked,
08:50consistently is around being ahead of
08:53and evolving where will search go,
08:55given the extraordinary experience
08:57that we can deliver for all of that.
08:59And this didn't start now, this evolution.
09:02We started Google Brain, our first AI entity,
09:05many, many years ago.
09:06We acquired DeepMind back in 2014
09:08with the view that at some point,
09:10we would be at this moment
09:12where we're all experiencing and living it,
09:14and it feels like a vertical lift.
09:15And what's exciting about being there,
09:17and I would say it's important for anybody today
09:20leading a company, leading a country,
09:23is we're all tech-enabled and have the opportunity.
09:26If you're not planning and investing for long-term growth,
09:29you're gonna be sowing the seeds of your own destruction.
09:31And so where we're sitting is an extraordinary position
09:34where we're leading in science and innovation.
09:37So we won, as an example,
09:39Demis Hassabis and John Jumper
09:40won the Nobel Prize this year
09:42for their breakthrough in science.
09:45We view ourselves as looking at science at digital speed.
09:48AlphaFold, which is the greatest contribution
09:51to drug discovery, that didn't come out
09:53of just recent research.
09:54They've been working on it for quite some time.
09:56Greatest contribution to drug discovery,
09:58the work we're doing in quantum,
10:00where we had a breakthrough in quantum computing.
10:02You can do a computation in less than five minutes
10:05that previously, or today, would take supercomputers,
10:08the best supercomputer on the planet,
10:1010 septillion, 10 septillion hours to do,
10:1410 and it's 24 zeros behind it.
10:16It's extraordinary what we're able to accomplish
10:19with science.
10:20And so we're already investing, to your question,
10:23on where the world will be next.
10:25Clearly, the majority continues to be in search,
10:28leveraging Gemini, what's that search experience?
10:31But we're also leaning into being the leader
10:33in science and innovation.
10:35I like that, I often think about how,
10:37at the Atlantic, we can diversify our business models
10:39for the age of AI.
10:40One of the things I had not thought about
10:41was building a quantum computer,
10:42but I will put that on my list.
10:44Albert, let's go to you.
10:46You have to change the way drug discovery works.
10:49You're gonna have to reinvent how the company works
10:51and you're gonna have to not be beaten
10:54by new drug development companies
10:56that are fully AI-based and are taking the technology
10:59that Ruth was just talking about
11:01and building companies from scratch based on AI
11:03with their agent employees.
11:04Tell me how you're thinking about this transformation.
11:07First, I need to tell you that I'm super excited
11:09about the prospects of life sciences
11:11exactly because of that.
11:12It's not only advancements in biotechnology,
11:14but also advancements in digital technology,
11:17that the two are colliding and they are going to create
11:19tremendous synergistic effects.
11:21I'm going to use an example of a disease that worries all,
11:24the emperor for maladies, cancer.
11:27There is a new biotechnology that we are using
11:29which is called ADC, antibody drug conjugate.
11:32I'm going to use, because we have marked here military terms
11:35to describe the war on cancer.
11:37The problem is that when you use the chemotherapies
11:40that are most commonly used drugs right now for cancer,
11:43you are not killing only the cancer cells,
11:46but you are killing also the healthy cells.
11:49So you need to be more precise.
11:51So that's why we developed something like
11:53a GPS-guided missile, a medicine that is very precise.
11:57We try, first of all, to identify a target
12:00that will identify that this is what we want to kill
12:05and should.
12:06So for example, a protein that it is more common
12:09in the cancer cell than in the healthy cells.
12:12One of the example, IB6, it's a protein,
12:15IB4, excuse me, which is a protein
12:17that is expressed in 90% of lung cancer cells.
12:21So now we know that this is the target.
12:22We know that if our missile sees IB4, we'll attack it.
12:28In order to identify targets, AI will make that process
12:33that usually was taking years, will take months.
12:37It's amazing, it's exponential, the improvement.
12:40Once we do that, now we need something
12:42that will be the GPS mechanism that will come
12:45and adhere to the target, this protein.
12:49So we need to bring an antibody that will go
12:52and click to this protein.
12:54And AlphaFold, which Ruth spoke about,
12:59and it is a good example because also the inventors
13:03took the Nobel Prize this year, it is helping us
13:07to accelerate the design of these antibodies
13:10instead of years, months.
13:12Once you have the antibody, now you need a warhead
13:15for your missile, and that will be nuclear
13:18or can be tactical, depends on what is the cancer
13:21you need to kill.
13:22So what is the result of all of that?
13:24You link the antibody with the warhead,
13:27you release the missile, and will not explode
13:30on any healthy cell, it will explode
13:32only on the cancer cells.
13:35AI can facilitate each and every of these prospects.
13:39Whatever we can do with ADCs in 10 years,
13:42we can do with ADCs in 18 months now.
13:46While you were talking about precision targeting missiles,
13:48I noticed that everybody on the panel was interested,
13:50particularly the Secretary General.
13:53I know how to get his attention.
13:56Dara, you're taking your company,
13:58you've got all these drivers, you've created incredible jobs
14:01for folks who need side gigs, working class jobs.
14:04You're about to move into a new era
14:06where a pretty significant percentage of your fleet
14:08will be robo-taxis.
14:09Pretty big change.
14:11How are you thinking about, actually I'm gonna ask you
14:13the same question I asked Mark.
14:14What is the hardest thing for you to get right
14:17during this transformation?
14:18Well, I think first the potential that we see
14:20in terms of AI.
14:21I think most persons, most of our experiences
14:24with digital experiences of AI, search, et cetera,
14:29where we work is the physical world.
14:31And we think that AI can have enormous impact
14:34on the physical world.
14:35And ultimately, the future of transportation that we see
14:38is going to be autonomous, it'll be electric,
14:41and eventually it'll be shared as well.
14:42So congestion on the streets is something
14:44that can be controlled.
14:48And for us, the applications of AI
14:50are essentially passenger vehicles.
14:51Instead of having a human driver,
14:53you'll have a robot driver.
14:55And we're working with Ruth and Waymo
14:57and a number of other companies in developing
15:01this technology and really bringing it to market
15:04so that everybody can experience it.
15:07And then also there are applications
15:09in terms of package delivery.
15:10So you may see here these little sidewalk robots,
15:13surf sidewalk robots that are delivering your food.
15:16And then of course in trucking,
15:18we have a business Uber Freight
15:20that's connecting shippers to truckers.
15:22Those trucks eventually will be driven by robots as well.
15:27And the ultimate promise of AI
15:29is literally to save lives, right?
15:31It's today as we speak, this year,
15:35there will be a million fatalities on the road
15:39as a result of human error.
15:41And this digital driver is going to have
15:45thousands of lifetimes of driving to train.
15:48It's not gonna be distracted in terms of by texting,
15:53by watching TikTok, whatever.
15:55It will be a driver that's completely focused on the road.
15:58It won't have two eyes.
16:00It'll have 10 eyes and many different sensor types.
16:04So by definition over a period of time,
16:06this driver is gonna get better and better and better.
16:09And it will start to be more of a reality in our network.
16:14Our view is we're working with a number of partners,
16:17Nvidia, Alphabet, Wabi, et cetera,
16:20to bring this technology to markets
16:23in a safe and responsible way.
16:25It's not gonna get there tomorrow.
16:28And there will be a very long hybrid road
16:31where sometimes you're gonna get a human driver.
16:34Sometimes you'll get an AI driver or robot driver.
16:37But fast forward 15, 20 years from now,
16:40the impact of AI in the physical world
16:43and the impact in terms of saving lives
16:46and avoiding fatalities is going to be enormous.
16:49And we want to play a big part in that transformation.
16:52Dario, Dario just said something.
16:54He said 15, 20 years.
16:55And talking to you here in Davos and talking to others,
16:58I think one of the biggest differences,
17:01you think things are happening much faster.
17:03Like most everybody in Davos agrees that AI is coming,
17:05things are gonna change.
17:06But when other people talk about years,
17:08it's changes that you talk about in months.
17:10And when other people talk about decades,
17:11it's changes you talk about in years.
17:13How fast is the world about to turn upside down?
17:17Yes, so I would say that I think the thing
17:21that most distinguishes AI
17:23from previous technological revolutions
17:25is exactly the speed at which it's happening.
17:28So I've been in the field for 10 years.
17:30I've been watching that curve for 10 years.
17:33And me and my co-founders at Anthropic
17:35were among the first to document
17:37what we call the scaling laws,
17:39which is the observation that as you pour more compute
17:42into these models with small changes in the algorithms
17:45according to which they're developed,
17:47they improve very fast on cognitive abilities.
17:50And so exponentials kind of start slow
17:53and very quickly pick up speed.
17:56And I think we're actually,
17:58I've become probably in the last three to six months
18:02more confident that we really are heading
18:04towards AI systems that are better than almost all humans
18:09at almost all tasks.
18:11My guess, and I'm still not certain of it,
18:14but I think with some of the things we've seen
18:16with models starting to be as good as PhD students
18:20in areas like math and programming and biology,
18:23it is my guess that by 2026 or 2027,
18:27we will have AI systems that are broadly
18:30better than almost all humans at almost all things.
18:35I see a lot of positive potential,
18:38that's an understatement,
18:39for how this kind of AI can be applied.
18:43You've heard about a lot of these areas, right?
18:45From military applications to in the workplace,
18:49to self-driving cars,
18:51to applications in biology and health,
18:55which are very near and dear to my heart
18:57because I used to be a biologist
18:58and I think that's maybe one of the opportunities
19:01to offer the most improvements in human welfare.
19:05In line with the thoughts on longer timelines
19:09for self-driving cars,
19:10I think what is gonna hold us back most
19:13in positive applications is the physical world
19:17and limitations on human institutions.
19:20So I wrote an essay about this transition,
19:23about powerful AI.
19:24If this really does happen, what will we be limited by?
19:28The term I used in this essay
19:30was marginal returns to intelligence.
19:33Economists talk about the marginal returns
19:35to capital, to labor, to land.
19:38A way we're not used to thinking about things
19:40is what if we have millions of copies,
19:43I called it a country of geniuses in a data center,
19:46of agents that are better than humans at everything.
19:51What limits?
19:53Do we just immediately solve all of the world's problems?
19:56I don't think that happens.
19:57For example, self-driving cars are difficult.
19:59So I think we're gonna be limited by the physical world,
20:02by deployment cycles, and often by laws.
20:05If we take the pharma case as an example,
20:09we still need to run clinical trials.
20:12There's still physics and biology,
20:14but I think the AI systems will get better
20:17at things that used to be late-stage clinical trials
20:19can become early-stage clinical trials.
20:21Things that used to be early-stage clinical trials
20:24can be done in vitro.
20:28Things that can increasingly be simulated.
20:35So I do think there will be a great acceleration.
20:39If I had to guess, and this is not a very exact science,
20:43my guess is that we can make 100 years of progress
20:47in areas like biology in five or 10 years
20:50if we really get this AI stuff right.
20:52If you think about what we might expect humans
20:56to accomplish in an area like biology in 100 years,
20:59I think a doubling of the human lifespan
21:02is not at all crazy.
21:04And then if AI is able to accelerate that,
21:06we may be able to get that in five to 10 years.
21:11So that's kind of the grand vision.
21:13At Anthropic, we are thinking about
21:17what's the first step towards that vision, right?
21:19If we're two or three years away
21:21from the enabling technologies for that,
21:24the way we've been thinking about it is
21:26all this describes productivity in the workplace, right?
21:29Even very high-end productivity for cures for diseases.
21:33So we're working on something called a virtual collaborator,
21:36which is the idea of having an early version of this.
21:39Not necessarily smarter than a Nobel Prize winner,
21:42but capable of doing relatively high-end tasks
21:46in the workplace that can open up Google Docs,
21:50that can use Slack, that can interact with its co-workers
21:53to perform tasks over hours to days,
21:57and that you only have to check in with every once in a while
22:00like a manager would check in with an employee.
22:02Mark said, you know, this will be the last generation
22:05of CEOs that only manages humans.
22:08We're trying to make that real.
22:10Secretary Genny, you want to jump in?
22:12Building on this, because what I'm seeing in my work,
22:15in my job now, is that what is holding us back
22:18in terms of speed of applying AI, exactly as you were saying,
22:21is the traditional structures, the bureaucracies.
22:24And basically, you have three types.
22:26One, of course, if there is no war,
22:28then the only way to leverage this new technology
22:31is to make sure that you work with smallest,
22:34medium-sized enterprises, with startups
22:36who are really leading edge here.
22:37So in NATO, we have all these systems,
22:39the IANA and the investment fund, et cetera,
22:41to make sure that you work with academia
22:44and with the startup sector.
22:45But it's not easy, because the system
22:47was used of building these huge tanks, huge fire jets, et cetera,
22:51and ships, what have you,
22:53and not implementing necessarily this latest technology.
22:57A second level is what we call hybrid threats,
23:01which can be quite scary, these hybrid threats.
23:04It can be an assassination attempt on the boss of Rheinmetall.
23:06It can be a cyber attack on the NHS in the United Kingdom.
23:11It can be the jamming of commercial flights in the Baltics.
23:14And, of course, getting at the critical undersea cables
23:18like between Estonia and Finland.
23:21And then there is a need to apply the latest technology fast.
23:25And I was happy to see that, at least within the NATO system,
23:27we were able to do this.
23:29But then, of course, leading edge here,
23:31cutting edge is Ukraine.
23:33What they have to do is to implement every two weeks
23:36the newest technologies to stay in front of,
23:39in terms of technology-wise, the Russians.
23:41But the Russians are copying
23:42what the Ukrainians are doing also in two weeks.
23:45These technologies, including the use of AI,
23:48and how we can speed up the bureaucracies to do this,
23:51is really key.
23:53And I'm absolutely optimistic, like all of you,
23:55when it comes to AI.
23:57But the question is,
23:58what is holding us back to do it even faster?
24:00It's AI and AI.
24:02Just one other question on AI for certain applications
24:06is the cost of an error.
24:10We, part of humanity, it's flaws.
24:14And we accept that humans are gonna make mistakes, et cetera.
24:17Again, you go to these road fatalities.
24:19I think one of the questions in certain AI applications,
24:22defense may be one of them,
24:23certainly AI in the physical world,
24:25is how much better does that AI have to be
24:28than a human being?
24:30Is 10 times better enough?
24:32Is 20 times better?
24:33You look at Waymo, for example,
24:34based on their safety statistics.
24:37Ruth, how much safer is Waymo
24:39than human beings, probably, at this point?
24:41Meaningfully safer, but your point is well taken.
24:44There's more forgiveness when it's a human driver.
24:47And so the question is, when a machine makes a mistake,
24:49when an algo makes a mistake,
24:51how does society look at that mistake
24:53and the cost of that mistake
24:54versus the benefits of AI coming into the fore?
24:58Well, society reacts very badly, right?
25:01As we've seen with self-driving cars,
25:03which are evidently safer.
25:04And the question is,
25:05will society change as more technology comes, right?
25:09Will there be a backlash?
25:10Albert, let me ask you this question.
25:12So what Dara is talking about
25:14is building much safer technology and facing a backlash.
25:16You built a product that was like the best product
25:19the world ever had and probably saved the lives
25:21of a bunch of people in this room,
25:22and there was a massive backlash.
25:24Is there anything you learned from that
25:27that will be helpful to people
25:28who are bringing in new technologies
25:30in the next years that will be useful?
25:33I'm sure you're referring to the vaccine,
25:35but it was a massive backlash.
25:36It was significant.
25:37It was maybe 10, 20% of the people.
25:40I can tell you at the same time,
25:41we received a lot of love from 80% of the people
25:45that felt that their mother or father
25:47is alive because of us.
25:48So let's put things into perspective.
25:50The fact that some of the voices
25:52are magnified through social media
25:54doesn't mean that they are the prevalent voice.
25:56But the example that you use is a good one
25:58because AI also will face the same problems
26:01and probably AI machines will create the problems to AI.
26:05They are the ones that will spread the disinformation
26:08about it, and why?
26:09Because it's a very powerful tool.
26:11And with every powerful tool,
26:13in the hands of good people, will do great things.
26:16In the hands of bad people, can do bad things.
26:18And I think that will be magnified.
26:20Every single mistake that AI will be doing,
26:22as you mentioned, will be magnified to the utmost degree,
26:28ignoring that humans are making even bigger mistakes
26:31and ignoring that the benefits
26:33that we got so far were huge.
26:35Ruth, can I ask you a big question
26:36that is quite relevant to this?
26:37So to me, the most interesting,
26:40I keep a list of unanswered, of questions in AI
26:43that I don't really know the answers to
26:44and that people disagree on.
26:45And the one that I'm most interested in
26:47is whether AI will make the world more equal or less equal.
26:49Will it increase inequality within countries
26:52or will it do the opposite?
26:53My view is that technology made the world less equal, right?
26:56Rose all boats, but the folks at the top did better.
26:59What do you think will happen with AI?
27:01Will it condense and collapse income inequality,
27:04as some studies show?
27:05Or will it exacerbate it as others do?
27:08Yeah, it's an absolutely critical question
27:09and one we're extremely focused on.
27:11I think there's a tremendous amount of upside to capture
27:16which will address this productively.
27:19But there are execution risks
27:20that we all need to be focused on.
27:22So what we get really excited about is it's in health,
27:24it's in education, it's the economic upside.
27:27We've talked about health already to an extent,
27:29but just to expand a bit, with AlphaFold, as I said,
27:34what the team did, what Demis and the team did,
27:36is predict the protein structure
27:38for all 200 million proteins on the planet,
27:41which is the building blocks for human life
27:43and that's why it accelerates drug discovery.
27:46They then open sourced it.
27:48And so we now have two and a half million scientists
27:51around the world, 190 countries,
27:53unleashed to use this extraordinary breakthrough
27:57for drug discovery.
27:58And so the ability to address disease,
28:00all forms of diseases, and do it at, as I said,
28:04at digital speed is absolutely extraordinary.
28:06And what does that do to address illnesses
28:09everywhere on the globe?
28:11The other, very much more direct to your question,
28:14is early diagnosis of disease.
28:16Early diagnosis is one of the most important elements
28:19in survival.
28:20And for me, this is personal, I've had cancer twice
28:23and I know I was one of the very fortunate ones
28:26who went to one of the best hospitals on the planet,
28:28Memorial Sloan Kettering,
28:29and I had early diagnosis.
28:31Not everyone does.
28:3240% of people in the United States will be diagnosed
28:35with cancer in their lifetime.
28:37Globally, there are other diseases that are actually
28:41killing people before cancer does,
28:42which goes back to my first point.
28:44Early diagnosis is key.
28:46Google, years ago, developed, with AI,
28:49the ability for early diagnosis of metastatic cancer.
28:52Started with breast, went to lung.
28:53I spoke to my oncologist about it
28:55and he said the only way to democratize healthcare
28:58is with AI, because it means anyone, anywhere,
29:01will be able to have the same high quality early detection
29:05that I had.
29:05And what we're already using today, through Google AI,
29:09the research that we've done,
29:11is early diagnosis for tuberculosis.
29:1330 to 40% of cases of TB go undiagnosed
29:16because they're in the global south
29:19or in poor communities across the United States.
29:22That is already happening today.
29:23That is not a tomorrow opportunity.
29:25Diabetic retinopathy, losing your eyesight
29:28as a result of diabetes.
29:29So, this very much is an opportunity to address inequity.
29:33Education's super exciting.
29:35In Africa, the average age is 19.
29:38Spoke to one minister who said,
29:40by definition, half of our population is under 19.
29:44We don't have enough teachers.
29:45And if you can't educate your kids,
29:48you can't address that opportunity gap.
29:50You know, I grew up, my parents always said,
29:52education is your passport to freedom.
29:54We firmly believe that.
29:55And what we can do, and what the minister pointed out,
29:58is AI is operating leverage for the teachers,
30:01and it also enables students to get taught through AI,
30:05which is extraordinary.
30:07The problem, the execution challenge.
30:10So, I talked about cancer.
30:11Well, cancer, early diagnosis,
30:13is one part of a broader ecosystem.
30:15In some places, this may be an equalizer
30:18in the United States.
30:19It will be an equalizer if you have a doctor
30:21who otherwise couldn't deliver,
30:22because you can then go somewhere else
30:24for chemo and radiation.
30:26But that is not true everywhere in the world.
30:27So, you need the full ecosystem to develop.
30:30The other really important point,
30:32probably the most critical,
30:33is that a third of the planet is still not connected,
30:36is still not online.
30:37So, they don't have access to the magic
30:39that we're talking about here today.
30:41And then you go to the economic upside.
30:43The economic upside is profound.
30:45The estimates are wide.
30:47They're in the trillions.
30:48But again, if you go under the cover,
30:49the execution issues are what we're really focused on.
30:53Now, it's encouraging that the data say,
30:56economists at MIT, at Stanford,
30:58that it is the early entrance to the workforce
31:01and lowest skilled jobs that may benefit the most
31:03because you have a tutor by your side.
31:05That's a positive.
31:07More jobs are created historically
31:08than destroyed in tech transition.
31:11That's the good news.
31:12The bad news is when your job is the one that's eliminated,
31:15it becomes very binary very quickly.
31:18And so, we're doing quite a bit of work,
31:20as are many others, on workforce empowerment
31:24for workforce training,
31:25getting people ready for what those new jobs are.
31:27And there are extraordinary jobs that are available.
31:30But that's something that we cannot take for granted.
31:32And the final point,
31:34wherever I travel around the globe,
31:35what I hear from global leaders
31:37is they want their country
31:38to be a part of the digital transformation
31:41because they see these upsides,
31:42the economic potential,
31:43the potential around healthcare,
31:45education, agriculture, you name it.
31:48And so, that's what we're trying to do
31:49is make sure that we are partnering up globally
31:52in the right way,
31:53and that execution definitely becomes key.
31:55And some of the comments today,
31:57I very much agree,
31:58at the end of the day,
31:59the upside opportunity is so profound
32:04that we need to collectively make sure
32:05we're mitigating on the downside.
32:06We can't miss it.
32:09This is a very optimistic panel,
32:10so let me try to change the mood a little bit, if possible.
32:14Mark, can I ask you about something
32:15that one of your passions,
32:17where maybe things aren't going as well as you might think,
32:19which is climate change.
32:20So, in the last little bit of time,
32:24we, the United States,
32:25has pulled out of the Paris Accords.
32:26We've frozen a bunch of spending
32:27that was going into climate investment.
32:29It seems, in fact,
32:30we're sort of transferring capital
32:31from climate to crypto,
32:33which maybe isn't the best use of energy resources.
32:37All this AI boom, incredibly carbon-intensive.
32:41Are we going in the wrong direction
32:43on an issue of paramount importance
32:44where we're already having a lot of trouble?
32:48Well, it's a pretty big context switch on the panel.
32:50I mean, we just heard the incredible benefits
32:53of unlimited workforces.
32:54The idea that AI is kind of becoming our partner
32:57to help us to run our lives, run our businesses,
33:01to help us to deliver a new level of productivity
33:06without a human workforce.
33:07And, you know, you're 100% right.
33:10We have to keep in mind
33:11that we have a vision for the world
33:15that is getting warmer.
33:20And the reason that it's getting warmer
33:22is because there is more carbon in the atmosphere.
33:24The first industrial revolution
33:26has really given us about 200 gigatons of carbon
33:29into the environment
33:30through various human levels of activity.
33:33But it's not our biggest worry.
33:35Our biggest worry is really deforestation.
33:37The planet had six trillion trees on us,
33:41and now more than half of those trees are gone.
33:46And for every trillion trees,
33:48we lose 200 gigatons of carbon banking.
33:51So just imagine that as three industrial revolutions
33:54that we've released into the environment
33:57through deforestation.
33:59I was very optimistic yesterday,
34:01and I guess we should go back,
34:04but we mentioned Trump five years ago.
34:07On this stage in 2020,
34:09President Trump announced the Trillion Tree Initiative.
34:12That is that our vision was to put a trillion
34:16of those trees back on the planet
34:17to sequester 200 gigatons of carbon.
34:20Of those trillion trees that we want to do,
34:22we actually now have commitments
34:24and underway on 200 billion.
34:26The biggest also happened on this stage last year,
34:30China coming in for 70 billion trees.
34:33That is a case for optimism,
34:35and another case for optimism
34:36happened just yesterday here at the forum.
34:39Not on this stage, but in another stage,
34:42the President of the Congo announced
34:43the world's largest forest reserve
34:45as part of the trillion trees.
34:47I think there's nothing more important
34:50than reforesting our planet
34:52if we want to make our world cooler
34:54and to counter what we've done in the industrialization.
34:58And number two is we can also see
35:00that the oceans are getting warmer.
35:01We've just had two back-to-back El Nino years,
35:05and I've heard now several reports
35:07talking to folks here from around the world
35:09that we see coral bleaching happening
35:11in areas where we have not,
35:13even just in the last six months.
35:15So whether it's coral bleaching, overfishing,
35:19whether it's creating more marine protected areas,
35:22getting to our vision of protecting 30% of the oceans
35:25by 2030, and getting the plastics out of the ocean.
35:30We're on the verge of the UN High Seas Treaty.
35:33This is probably one of the most fundamental things
35:36the United Nations can do this year.
35:38And the second is the UN Plastics Treaty.
35:41We have to get the plastics out of the ocean.
35:44So we can all think about this as a case for optimism
35:48in the environment, and AI is an accelerator
35:50on all of these ideas.
35:52So net-net, do you think that AI,
35:54because of its ability to help us model plastic flows
35:56in the ocean, for example,
35:57and figure out how to extract it more efficiently
35:59and maybe build carbon capture systems,
36:01net-net, AI will be positive for climate,
36:04even though it has huge energy costs?
36:06Net-net, I would say that the most optimistic thing
36:10that I've seen in getting the ocean plastic
36:13out of the ocean is, number one,
36:15making sure it doesn't get in the ocean.
36:17So robotics has advanced with AI,
36:21and we see entrepreneurs like Mark's colleague
36:24from the Netherlands, Boyan Slat,
36:27who has developed robots that sit in rivers
36:30in critical areas in Asia or in South America
36:35that prevent this plastic from getting in the ocean
36:37in the beginning.
36:38That just wasn't possible 10 years ago.
36:40Now we see the ability to extract millions,
36:44hundreds of millions, billions of tons of plastic
36:46that's already in the ocean,
36:47but we have to also keep it from going there.
36:50We also announced a plastic initiative
36:52here at the World Economic Forum.
36:54How do we get to bioplastics?
36:56This idea that the plastic that's in the ocean
36:59is kind of the nuclear waste of our age.
37:02It's gonna be there for a long time.
37:03We already see the infusion of that plastic into plankton.
37:07We now know the fish are eating the plankton,
37:09ingesting the plastic, and we know where that goes
37:13in our food chain right into our bodies,
37:15and I don't think any of us want what's happening,
37:17and I don't wanna go into the details
37:20because it just turns into,
37:22it'll take away the case for optimism on human health,
37:25but we want to have a healthier food chain
37:28and getting plastic out of our food chain is so important.
37:31Well, since in Davos I only eat Green Room granola bars,
37:34haven't had any plastic,
37:35but I do think this is an extremely important cause.
37:37Dario, let me go back to your essay
37:39which you mentioned earlier, Machines of Love and Grace.
37:42One of the most interesting parts,
37:44and I think something that ties into what Ruth
37:46was just talking about, something that has come up on stage,
37:48is the prospect for sort of stable democracy
37:53where people trust the government,
37:56and it's an extremely optimistic essay,
37:58but what's so interesting is there's a point
38:00where you're not as optimistic,
38:01and you say you're not sure whether AI
38:04is more enabling for democracy
38:06or more enabling for authoritarianism,
38:09so explain what you mean and explain what you think,
38:11and then we'll have the others comment on that
38:12because that's pretty important.
38:14Yes, so I think here we've mostly talked
38:17about all the positive applications,
38:19and I'm very excited about them.
38:21I tend to be optimistic about places where markets work well
38:25about human scientific advancement leading us forward
38:29and AI scientific advancement augmenting them.
38:33I think places where I'm less optimistic
38:36are around the political system.
38:40If we look at the last 100, 200 years of history,
38:43there has been sort of a trend towards democracy
38:47being more ascendant, democracy being more dominant,
38:50but it's been an extremely noisy trend, right?
38:53If the moral arc of history is long,
38:56this is perhaps the longest part of it,
38:58and so this is one of my deepest worries.
39:00If we go back to the country of geniuses in a data center,
39:03which I think we will likely have in two to three years,
39:07imagine what that country of geniuses in a data center
39:11could do in the hands of an autocracy.
39:13We take 10 million virtual minds
39:18that are smarter than any Nobel Prize winner,
39:21and let's say put them in the hands of China.
39:24What could they do in terms of a surveillance state?
39:27The power of dictatorships
39:29has traditionally been constrained
39:31by the need to have humans
39:33who carry out the will of the dictator.
39:35That has limited how terrible the dictatorship can be.
39:39There is a chilling possibility
39:40that AI could remove some of those limits
39:43and make possible something like a 1984 or darker.
39:47On the international stage,
39:49is it possible that AI could drive
39:52these very powerful drone fleets
39:54or this all-seeing ability
39:58in terms of intelligence analysis
40:01that could give autocracies an advantage against us?
40:08I think this is one of the most important issues
40:10for us to attend to,
40:11because I think unlike the positive benefits,
40:14which take work, but the wind is at our back,
40:17this is very serious,
40:18and I'm not sure it's gonna go right.
40:20Very quickly, I can see a couple people wanna jump in.
40:22Is the fear that AI in the hands of a dictatorship
40:26is more useful and powerful
40:28and entrenches them more than AI in the hands of democracy,
40:31or that there is something about the nature of the technology
40:33that creates instability
40:34that then leads to likely dictatorship?
40:36I think both.
40:39I'm worried, I think both about
40:42kind of the international side of things,
40:44like will autocracies get ahead of us?
40:46Will they be able to do this better than us?
40:48And I worry about internal constitution as well.
40:52So one worry I have is,
40:55is AI stabilizing to autocracies
40:58and destabilizing to democracies?
41:00I think we can fight that.
41:02Some of the ideas I've had,
41:05and probably over the coming year,
41:07we as an individual company will work
41:09to try and promote these,
41:10is can AI improve the quality of democratic governance?
41:15One area is public services.
41:17Many public services are not delivered
41:19in a technologically sophisticated way.
41:22Can we use AI to reinvent democracy,
41:25to inspire citizens that their government is doing things,
41:31I guess, efficiently, to use the modern language,
41:33but is also delivering robust services
41:37and making their lives better?
41:39Can it do better at enforcing and augmenting people's lives?
41:44Can it lead to better deliberative decision-making?
41:47About a year ago, we did some research
41:50with something called the Collective Intelligence Project,
41:53which was focused on collective decision-making.
41:57Can AI allow people to debate and discuss with each other
42:02and come to the truth better?
42:04Can it summarize positions?
42:06Can it encourage better processes
42:08of democratic deliberation?
42:10Can AI be involved in the system of justice?
42:14Again, we have to be careful with that,
42:16but can it be done in a way that enables us
42:19to give the same rights to everyone
42:22in a more uniform way?
42:24Can AI enhance the promises
42:27that democracies make to their citizens
42:31in order to do a better job
42:34of delivering on the ideals that democracy represents?
42:40All right, Ruth?
42:41I think there's so much important content
42:43in what Dario said.
42:44I just want to double-click and do a couple of points.
42:47First, it's imperative that the West stays ahead,
42:50that the US stays ahead with all that we're doing with AI,
42:55because the best approach
42:58is to have the strongest defense be ahead,
43:01and that's what we've done as an example with cybersecurity.
43:04You need to be ahead of those who might want to disrupt,
43:08take things down, and attack in the wrong way.
43:12And so that requires a pro-innovation regulatory environment
43:16that doesn't put sand in the gears.
43:18We need to be bold and responsible,
43:19but we need to make sure that we have the ability
43:21to maintain the lead currently.
43:23We think that the West, the US,
43:25is ahead in models by at least a year,
43:28in chips by at least a year,
43:29but if we were here a couple of years ago,
43:31that gap was wider, it is narrowing,
43:32and we cannot take for granted
43:34that we will continue to be in the lead.
43:35That is point number one.
43:36Point number two is a geopolitical point,
43:39and I would say there's an issue
43:40that's actually present here today
43:42that fortifies for the concerns that Dario's raised.
43:45I've already noted that wherever I go,
43:47what I hear from heads of state
43:48is they want to be part of the digital transformation.
43:51What I didn't say is they want to work
43:53with companies in the US.
43:54They want to be aligned with the West,
43:56the values, the products, the services,
43:58the uplift that we provide.
43:59But they're also very clear
44:01that in the absence of us being there,
44:03they will be part of the digital transformation,
44:05and they will partner elsewhere.
44:07And so I think it's imperative
44:09when we think about, for example, US policy,
44:12the AI diffusion rule that came out a couple of weeks ago,
44:15how can we engage and work with our allies globally?
44:18It's about the technical infrastructure
44:20that enables the product solutions
44:21that reflect Western values.
44:24It's about the education that you can then also provide.
44:27So I think those are really key,
44:29and I absolutely agree with Dario
44:31that we should each see anyone running
44:34the public sector.
44:35There's a massive opportunity
44:38to deliver better services more efficiently
44:41in a way that says we are working for you.
44:43And the number of examples in provisioning service,
44:46doing it around the globe
44:48where you can respond in a more timely way
44:52to needs, whether it's payments,
44:54whether it's information that says we are working for you
44:59and we are doing it efficiently.
45:01And one of my favorite stats
45:03that has actually a bunch of lessons
45:05that come out of it is Google, 20 years ago,
45:07started on this journey of Google Translate.
45:10We now translate in 260 languages.
45:13We added 110 languages in the last six months alone
45:16as a result of AI.
45:18The point is, wherever we are,
45:21even, for example, in the state of Minnesota
45:23in the United States,
45:24when they said we wanna be more productive,
45:27they asked us to deliver these services and solutions
45:30initially in four languages.
45:31They then added 26 more languages.
45:33We are there to serve.
45:35The other point is it's a vertical lift.
45:37So it underscores this 260 languages,
45:40110 in the last six months.
45:42Anybody who's on the front foot
45:43of leading a company or a country,
45:45you need to move fast because that's what AI is doing.
45:48And if we're not, the other guys are.
45:50All right, we're out of time,
45:52but Mark, I want you to give one final thought
45:53to wrap up how we're gonna save Western democracy here.
45:56Well, I would say that AI on balance
45:59is very good news for the Western democracy
46:00because it gives you access
46:02to a huge fund of wisdom and information.
46:05And many of my friends have shut off Google
46:07and are just using AI tools,
46:09even as a search engine
46:11to get to the latest insights, et cetera.
46:13So I do believe we can be generally optimistic.
46:15The reason why voters are not any longer voting
46:18for the centrist parties
46:20and we're flocking to the extreme right
46:21and extreme left wing
46:22is because the centrist parties,
46:24the liberals and the Christian Democrats
46:26and the social Democrats
46:27have not given answers to the two biggest issue
46:29of this time.
46:31And that is migration, one.
46:32And two, how to rebalance climate change
46:34with making sure that our economies stay competitive.
46:37And if we don't do that
46:39and the centrist parties are not allowing
46:41for answers to these questions,
46:43then it is not the voters turning in the wrong way.
46:46It is the centrist parties delivering a bad deal.
46:49And this has nothing to do with AI.
46:50This is just basic politics.
46:53You have to deal with the issues
46:54the voters want you to deal with.
46:57All right, thank you very much.
46:58Fabulous panel, fabulous discussion.
47:01Lots of big topics.
47:02Thank you all and now let's go have lunch.