- 1 day ago
Category
🤖
TechTranscript
00:00Ruth Porat, I'm so happy to speak to you because AI could change everything for the good,
00:04maybe for the worse at certain parts, but where do you think we are in 2025 in terms of
00:09advancing technologies in AI but also adoption? First, great to be here. This is always a wonderful
00:14event, so delighted to be here. Look, I think where we are with AI is it is both moving really
00:20fast and really slow, and what I mean by that is when you focus on the science, the breakthroughs
00:27in science are breathtaking, the pace at which we're seeing advancement, and I think one of the
00:32best ways to think about that is actually with my colleague, Demis Asabas, who runs Google DeepMind,
00:37and the work he did to develop something called AlphaFold, which is the 3D prediction of protein
00:44structures, which is viewed as the single greatest contribution to drug discovery in our lifetime,
00:50and he went from working on that, he's been in neuroscience and computer science physics for
00:56his whole life, but four years ago started on this journey to take on what had been a 50-year
01:01grand challenge to predict protein structure of the protein structures for all known proteins,
01:07and in four years he went from an idea that people challenged, is it possible, and he said,
01:13why not, to a solution which has now been open sourced, 3 million scientists around the world are using
01:19it 190 countries, and the Nobel Prize. That is the speed of change, and we're seeing that across
01:24Google DeepMind. What is slow is actually the implementation in both the public and the private
01:30sector. We're still very early days. The excitement about AI is across the board, the economic upside,
01:374 trillion potential contribution to GDP by 2030, with proper application across industry and the
01:44public sector, what's called diffusion. It's the better delivery of health care, better delivery of
01:49education, and that is still very early days. AI is clearly a lot more than a chatbot. It's about a
01:55fundamental rethink of the processes that we have, and I would say for each one of us as leaders,
02:00we need to start on that journey.
02:02So why do you think root adoption is so slow, or slower compared to the innovation? Is it because
02:07it's capital intensive and chief executives or even countries don't really know where to go next?
02:12I think one of the key elements is the prioritization. Where do I begin? And I hear that time and time
02:19again. And the mental model I keep coming back to, the framework actually anchors on health for a host
02:26of reasons. One, I think it is so relevant to each one of us. I've had cancer twice, and so I look at
02:32the application of AI in health care, and there's so much reason to be optimistic. But when you then
02:38think through, what are the solutions? How do I prioritize? Actually, the insights from health
02:43are relevant across all industries. So first is saving lives. It's innovation. It's drug discovery.
02:50I've already talked about that with AlphaFold. But very importantly, what Demis said when he was
02:55awarded the Nobel Prize is, I want to take on the most intractable problems. And when he was challenged
03:00as to whether it was possible, his answer was, why not? And I think that's the way we each need to look
03:04at innovation, regardless of industry. The second is obviously in health care, early detection is the
03:11difference between survival and not, in many cases, or a really difficult course of treatment.
03:16And what is it? In cancer, Google pioneered the early detection of metastatic cancer, breast cancer,
03:22lung cancer, and more. And that is literally the proverbial needle in a haystack, finding that early
03:29sign of a problem. That is precisely what we do across many different applications. Cybersecurity
03:36is about finding the needle in the haystack. In fact, last year, Google DeepMind found with AI
03:43that they could see vulnerabilities, find vulnerabilities in software before they were
03:48breached. So again, it is fortifying what you were doing in cybersecurity. It's true for fraud,
03:54et cetera. And then finally, it's about operating effectiveness, shortage of doctors and nurses
04:00around the globe. What we found with AI is you can free them from about 30% of their work, the
04:06administrative, basically drudgery, so they can do what we want as patients, focus on you as a patient,
04:13focus on what they want to be doing. And again, that is true across public and private sector. And I just
04:19was with a CEO this week who said, yes, the application of AI customer support, what they're
04:24hearing from their employees is, thank you for freeing me from that drudgery. And they can do
04:28higher level work, and it unlocks capital that can then go into innovation. So I think prioritization
04:33is key. I think for everyone in this room, each one of us actually getting tactile with AI so that
04:38you can see how it's used. I learned how to vibe code. I think if I can vibe code, everyone in the room
04:44can vibe code. So it's really understanding and then prioritizing. How do you use it? Do you use
04:49it like on a day-to-day? I do. Well, first of all, we all do because at Google, we've been infusing it
04:54in search for many, many years now. And so I do. I think one of the most exciting applications I keep
05:00going back to is something called Notebook LM. If you haven't tried it, upload whatever report or series
05:06of reports, and you can actually, you can turn it into a podcast, first of all. But you can also
05:11simply synthesize and do cross comparisons, and it's your corpus. So you control the analysis and
05:17where you're going. The more we use AI, of course, the more energy it needs. How do you look at energy
05:23demand and infrastructure needs? I like the way you frame that because the upside from AI
05:29is profound, and we can only access it if we actually have the energy to power it. So it's
05:33important to keep those as two sides of the same equation. And clearly in the U.S. and in other
05:39places, the U.S., we've underinvested in the grid for quite some time, and so we need to catch up.
05:45I think what's important is there are both short-term and long-term solutions that are right in front of
05:50us. Probably one of the most important, but it's longer term, is continuing to make advancements
05:56in things like nuclear energy, advanced energy solutions. And this is clearly a priority for
06:02the administration. We very much agree with it. We think it's imperative that nuclear energy is a
06:08part of the energy mix. At Google, as an example, we were the first corporate purchaser for an SMR
06:15offtake from Kairos. We've invested in fusion. But this is critical. And when you look at what
06:20China's doing, they've been investing in nuclear for quite some time. They have about 30 gigawatts
06:25of nuclear in construction, another 200 gigawatts in development. The United States needs to avail
06:32itself of that, and I'm pleased the administration is very focused on how to fast-track it. Near-term,
06:38there's some really important solutions. And I think what is important, again, if we execute against
06:43these things that are right in front of us, we can increase capacity rapidly. We need it now. This is
06:50not a 2030 need. It's a now need. We can do it and enhance affordability across the country, which
06:55is clearly imperative, generate jobs and decarbonize the grid. And the two best examples are, number one,
07:03when you look across the United States, we have underinvested in transmission lines. The reality is
07:09there's about 2,500 gigawatts of capacity in development that is waiting to get on the grid,
07:16a staggering number. But we need the transmission lines to get it on the grid. To date, it takes about
07:23seven years in the U.S. to build these transition lines. With fast-tracking permitting and other work
07:29with states, you can cut that time in half, and some would say even more than that, which is a
07:34meaningful unlock. What's potent about that is that it also creates jobs, a meaningful number of jobs.
07:41And importantly, the vast majority of that is clean energy. So by getting it on the grid,
07:47you're decarbonizing the grid. The second area that I think is really exciting is actually
07:53modernizing the grid. There's another hundred gigawatts in the United States that the Department
07:59of Energy has estimated could be unlocked if the grid is modernized, applying what's called grid
08:05enhancing technology, something like conductors online so that you can get more capacity through.
08:10And again, that creates jobs. It unlocks assets that are here that are being underutilized because
08:18they have not been modernized. So I think that, again, there are a number of different solutions
08:21at Google. We're working on a whole host of others to improve energy efficiency in chips and models,
08:27delivery across the board. We can, if resolute, and we take each one of the available assets here
08:34and execute against them. We can unlock the energy.
08:37Is that part of the whole package in this AI race? So Google spent last, you know, quickly actually
08:43caught up with OpenAI in the last two years. How are U.S. companies doing compared to China,
08:48for example? And is the energy complex part of that race?
08:52So energy unequivocally is important. I think the U.S. is ahead in models and in chips and continues to be.
09:00I think what's very important is China, as we saw with DeepSeek, is a very active innovator and working,
09:08you know, assiduously on AI and the application of AI. So I think there are two critical elements for the U.S.
09:16What, three critical elements, really, for the U.S. and the West. Number one is you need to have the energy infrastructure
09:22to deliver the upside. And China is focused there. It's not just nuclear. It's also in solar.
09:27They're building out whatever they can to have that long-term infrastructure needed to deliver the upside from AI.
09:33We have tremendous resources in this country. If we avail ourselves of all of them, we can win in this regard.
09:40The second is the application. So the upside only comes with each one of us actually unlocking the upside from this,
09:47as I've already talked about. Anecdotally, we're hearing China is actually very focused on the application of AI across industry.
09:55It's imperative we each do that. It puts us in a better, each company, each country in a better competitive position.
10:01And then I think very importantly, it is about embracing working closely with our allies.
10:06And so one of the things I'm very proud of is the work we're doing around the globe.
10:10Every head of state with whom I meet says we want to be a part of the digital transformation.
10:14We see not just the economic upside, but we can address health care requirements.
10:19My oncologist, I loved a line he said to me, he said, the only way to democratize health care is with AI,
10:24because then you actually have a partner to anyone around the world who can deliver this.
10:29We're doing amazing work, I think, for example, in Africa on maternal care.
10:34There are so many areas where it is available to us now.
10:37So they're saying we must be a part of this.
10:39That requires, as we look at it at Google, engaging to unlock products and services,
10:45which means investing in technical infrastructure, and very importantly,
10:49skilling people so that they have the capabilities to actually evolve as the jobs evolve.
10:55I view that as a critical triangle, the services, the infrastructure, and the training.
10:59And I'm intrigued wherever I go how often I'm asked for, first, help me with the training.
11:03We just actually announced, which will be talked about here this week, a very, I think,
11:09exciting effort across Africa where we're building out subsea cables to ensure connectivity,
11:16resilience, and security across the continent, and really with four hubs, as well as incremental
11:21AI digital skilling.
11:24And this is really important because a third of the globe still isn't connected.
11:28And if you're not connected, you're not going to get the upside from what this offers.
11:30Ruth, we're almost out of time, but actually an important question on responsible AI deployment.
11:35What does that look like to you?
11:36What are some of the questions that the leaders here need to ask, also in parts of the world,
11:40but maybe the framework is not as mature as in other countries?
11:43It's super important.
11:45You know, we've talked about it for quite some time.
11:46With any technology as potent as this, tremendous upside, and we've talked about it.
11:50You've got to protect and mitigate against the downside.
11:53And part of that is about this labor transition question, which is why at Google we spend a lot
11:58of time on AI skilling and look forward to working with others, or that we did an electrical
12:04training program so that there are actually trade jobs that are available.
12:08But managing the transition well is really key.
12:11Ensuring quality of information is always key in ensuring access.
12:15I think what we're excited about is that, unlike the digital divide, we shouldn't have an AI divide.
12:22So much of the service is actually available on a phone.
12:25And so this is a way to address what historically has created the divide.
12:29And, again, another reason to be very excited about it.
Recommended
1:44
|
Up next
5:40
2:48
1:24
1:26
17:44
16:17
10:04
7:11
8:22
8:05
3:43
Be the first to comment