Williams

Almost $20 trillion in revenue—and a record-breaking $1.87 trillion in profits. Introducing the 2025 Fortune 500, the definitive list of the biggest companies in America.

Fortune 500

Industries included in the sector are mining, oil and gas exploration and services, pipelines, refining, and utilities.

Fortune Sector Leaders: Energy

The Fortune 500, in its 70th year in 2024, ranks the biggest U.S. companies by revenue.

The Fortune 500, in its 69th year in 2023, ranks the biggest U.S. companies by revenue.

This year’s Fortune 500 marks the 68th running of the list.

This year’s Fortune 500 marks the 67th running of the list.

F500-2021-379-Williams

The oil and gas company was founded in 1908 and grew quickly to manage nationwide pipelines, which it still maintains today.

This year's Fortune 500 marks the 66th running of the list.

F500-2020-387-Williams

The Tulsa-based energy company focuses on natural gas and petroleum exploration and production, with assets concentrated along the Gulf of Mexico and the Eastern Seaboard.

This year's Fortune 500 marks the 65th running of the list.

This year's Fortune 500 marks the 64th running of the list.

This year's Fortune 500 marks the 63rd running of the list.

This year's Fortune 500 marks the 62nd running of the list.

Business insiders have spoken: Asked which companies they most admire, more than 4,100 executives, directors, and analysts chose these top 50 All-Stars.

World's Most Admired Companies

This year's Fortune 500 marks the 61st running of the list. Wal-Mart claims the top spot again. See who made the list.

Key Financials (last Fiscal Year)

Revenues ($M)

Profits ($M)

Assets ($M)

Total Stockholder Equity ($M)

Profit Ratios

Profit as % of Revenues

Profits as % of Assets

Profits as % of Stockholder Equity

Earnings Per Share (last Fiscal Year)

Earnings Per Share ($)

EPS % Change (from prior fiscal year)

EPS % Change (5 year annual rate)

EPS % Change (10 year annual rate)

Total Return

Total Return to Investors (last year)

Total Return to Investors (5 year, annualized)

Total Return to Investors (10 year, annualized)

Q2 2025

Q1 2025

Q4 2024

Get insights into Williams including business profile, stock information, news, and rankings from Fortune.

Williams (WMB): Company Profile, Stock Price, News, Rankings

legacy

williams

Companies

Williams | Fortune

Williams (WMB): Company Profile, Stock Price, News, Rankings | Fortune

Fortune Editors

Fortune 500 Power Moves: Which executives gained and lost power this week, May 3-9, 2025

/article/fortune-500-power-moves-may-3-9/

Fortune 500 Power Moves: Which executives gained and lost power this week, April 19-25, 2025

/article/fortune-500-power-moves-april-19-25/

Fortune 500 Power Moves: Which executives gained and lost power this week, March 8-14, 2025

/2025/03/14/fortune-500-power-moves-march-8-14/

Wayfair thrived during the pandemic, scaling to over 30 million products and rising to #217 on the Fortune 500. But after a post-COVID slump, it’s rethinking its strategy. In 2024, the company opened its first large-format retail store and rolled out new AI tools to reimagine how people shop for their homes. Here’s how Wayfair is betting on technology and in-person retail to power its next chapter in a crowded home-goods market.


How Wayfair’s Betting on AI to Future Proof its $12B Empire 

Wayfair is betting big on AI and physical retail to reconnect with shopper

Tech

Wayfair thrived during the pandemic, scaling to over 30 million products and rising to #217 on the Fortune 500. But after a post-COVID slump, it’s rethinking its strategy

How Wayfair Built a $12 Billion Home Retail Empire 

Wayfair built its multi-billion dollar business by selling home goods online, eventually scaling to more than 30 million products. Today, it's the largest online home goods retailer in the United States, but after a pandemic boom in sales, growth slumped, and so did the company's stock price. Now Wayfair is pivoting. In 2024, it stepped into physical retail with its first large format store. At the same time, it's betting on AI to streamline operations and reimagine how shoppers discover furniture. So we've been using the sort of text to image generation capability to generate an AI to be able to showcase a whole bunch of different images that our customers can then react to and say, oh, I like this. Here's how Wayfair built a $12 billion home goods empire. And how it's using AI and big box retail to fuel its next chapter. Founded in 2002 as a network of niche furniture sites, Wayfair evolved into a sprawling online marketplace. It now offers over 30 million products from around 20,000 suppliers. It also runs a lineup of lifestyle and luxury brands. The main basic difference between Wayfair and some of its other home furnishing competitors is that it is a marketplace platform, so they don't own any of their inventory. Everything is just forward position. on their site so that consumers can use it from the many vendors that they work with. Unlike many of its competitors, Wayfair doesn't manufacture or own its inventory. Instead, it connects shoppers with third party suppliers. This dropship business model allowed Wayfair to scale quickly, offering an enormous product catalog while keeping fixed costs low. Having these relationships with some of these smaller vendors is really unique in that it's a More visible way for the vendors to reach a much broader total addressable market. It helps elevate the business of small vendors while it also helps provide more choices to consumers for Wayfair. During the pandemic, that model helped the company thrive. Sales surged by 45% in 2020 as consumers redecorated during lockdowns. The next year, Wayfair shot up the Fortune 500 rankings, climbing more than 100. 30 spots to land at number 217. When we look back at COVID, that clearly accelerated e-commerce adoption across a number of categories, things like grocery and furniture. We saw the share price really rise to around $350 a share, but post-pandemic demand cooled as interest rates climbed and inflation pinched household budgets. Spending on big ticket furniture slowed. Wayfair's stock lost over 80% of its. between 2021 and early 2023, investors realized things are going to reopen. People are traveling. They're eating at restaurants again, and it wasn't idiosyncratic to Wayfair. It was across the industry, right? All of the peers really saw that phenomenon. Major home retailers like Bed Bath and Beyond filed for bankruptcy in 2023, shuttering hundreds of stores around the country. Despite declining sales post pandemic, Wayfair was. A to weather the market volatility. You know, if you look at somebody like Bed Bath and Beyond, there were liquidity issues. Interestingly, despite the fact that really Wayfair isn't generating a whole lot of profitability, the way that the capital structure works is that they don't really need that same sort of liquidity. But as the online home retail market has become more crowded, Wayfair is looking for ways to improve its e-commerce model and its customer experience. So we've been applying AI for quite some time, probably over a decade. The more recent advancements are allowing us to take that to yet another level. Fiona Tan is Wayfair's chief technology officer. Over the last few years, she has helped lead the company's investments in AI. It's a category where it's hard for our customers sometimes to describe what they're looking for. And then sometimes they don't really even know what they're looking for, so we've been using the sort of text to image generation capability to generate a AI to be able to showcase, you know, a whole bunch of different images that our customers can then react to and say, oh, I like this. I didn't know, I don't know what it's called, but it speaks to me. In 2025, Wayfair debuted a generative AI platform called Muse. The tool allows shoppers to curate AI generated home designs in certain styles. From there, the platform will recommend Wayfair products that fit the look. And then you can even tailor it a little bit more and just say, OK, well I like this, but I prefer green chairs instead of beige, and as you interact with muse, it'll generate you another image based on what you're saying. As the customers are inspired and engaged more with us, we're able to discern better what the intent and what their style preferences are, maybe what the budget preferences are, and then it allows us. And then recommend products for them. Waver says that level of personalization may help reduce return rates by giving customers more confidence in their purchases because the customers are more confident of the products that they buy or they've been able to get more information about it, ideally what you also see is there are fewer returns, right, because you're able to help them make um more. Informed decisions. Behind the scenes, Wayfair's legal, HR, and finance teams are also using AI to process data faster and streamline operations. For example, our legal teams using it to be able to look at, you know, user reviews to look for items that may have caused any kind of injury, and then you'll have examples in the technology department. Our accounting and finance team is also leveraging it. So it's really across the enterprise. While AI is helping to streamline Wayfair's digital operations, the company is also testing a more traditional strategy opening physical stores. In May 2024, Wayfair opened its first large format retail store in Wilma, Illinois, just outside of Chicago. The 150,000. A square foot store features showrooms, interior design services, and a cafe. Instead of spending marketing dollars on digital spend, you're spending your marketing dollars on physical spend. While competitors like William Sonoma benefit from built-in visibility through their retail locations, Wayfair has had to rely more heavily on advertising. every, every home. In 2024, they spent about 12% of their net revenue on advertising, while competitors like William Sonoma spent about 7% and Restoration Hardware only spent 4%. Wayfair leans into advertising a little bit more because it's hard to remain visible when you're an e-commerce only retailer, right? And so, You know, reminding consumers that you exist isn't free. Wayfair's new store location sits strategically along a major commuter highway. Since the opening of the store, Wayfair says sales in Illinois have jumped around 15%. Based on early results, Wayfair plans to open two more large format stores, one in Atlanta, Georgia and another in Yonkers, New York. Still, when it comes to home goods, Wayfair operates in a competitive and fragmented landscape. While the company holds the top spot for largest online home goods retailer in the US, it's only by a small margin, very fragmented, very hard to have a competitive advantage. We think Wayfair is one of a number of competitors that are really struggling. To avoid discounting and reinvent themselves in different ways to remain relevant to consumers. Big box retailers like Walmart, Target, and IKEA all sell home goods with the added advantage of physical stores, tight logistics, and brand familiarity. Other high-end home goods companies like RH and Pottery Barn offer in-person shopping that build a more luxury customer experience. It's a lot easier to be top of mind when you are a pottery barn in a mall or a RH that has a freestanding store and a restaurant, and it's an experience, but we. moves into brick and mortar stores and AI are helping to close that gap. So one of the capabilities again of AI and also generative AI is our ability to understand, have a deeper understanding of the products that we sell, discerning the color, the size, dimensions. Some of it's in the product details, sometimes there may be errors and it can also augment and enrich the information. That we get from our suppliers, so we're doing a lot in the customer experience. I do think AI will help young people sort of get their head around how they want to plan their smaller spaces, but also, you know, the accessible price points that is that Wayfair offers also helps bring those consumers in, and I think that a lot of that is their their main consumer.

Brian Cornell, Chairman and CEO, Target Corp.
Steven Williams, Chief Executive Officer, PepsiCo Foods North America
In conversation with: Diane Brady, Fortune

Fortune Global Forum 2024: Connecting Culture and Performance

Conferences

Dr. Matthias Röder, Chief Executive Officer, Karajan Institute; Co-founder and Managing Partner, The Mindshift; Director, Beethoven X – The AI Project

Brainstorm AI London 2024: How Deutsche Telekom Used AI to Finish Beethoven’s Unfinished Symphony

Photography by Joe Maher/Fortune Brainstorm AI

Thank you, Jeremy. Hey, everyone. 200 years ago, a mildly grumpy, middle aged man with wild hair, stepped out of his apartment in Vienna briskly walked down the street, mumbling, humming and singing from time to time standing writing something in his notebook and continuing on his way. Yes, you guessed right. I'm talking about Ludwig Van Beethoven at the time, the most well known composer of music in Europe and almost completely deaf. At this point. Later at home, he would take out the melodies he found during his walk and copy them into a big sketchbook that was lying on his desk. This is where the real work started, where he started to work on the musical problems that the material posed to him. How should I develop this melody? How should I combine it? Which key should it be modulating to? What is the setting and the structure that is good for this type of material. This was the actual work and Beethoven had amassed many of such books and he would go back and forth. This was his creative Bible, the 10th symphony that he was working on at the time that didn't materialize Beethoven died two years later, having left the work unfinished. 195 years later, I got a phone call not by Beethoven, but Mi Michael Schulz from Deutsche Telekom asked me, Matthias, can we finish this 10th symphony of Beethoven with a I, I don't know, Michael. Uh we got to put a team together, we got to start working on it and we got to see how far we can get well do it. Two months later, I was pretty certain that it was possible to do something whether it was the big thing, I don't know. So Beethoven left 45 sketches for 1/10 symphony. Some of which were a little longer, others were shorter. A musicologist named Barry Cooper a couple of years ago used the longer ones to finish one potential version of the 10th symphony and that worked pretty well. But with the short material, you know, no human could ever possibly take that and turn it into something bigger. Why? Because you would naturally put your own ideas, your own style, your own way of composing into the material. I called Ahmed El Gamal from Rutgers University who is a computer scientist specialized on creating artworks with the help of A I and asked him if he wanted to be part of the project. He said yes, but I've never done anything in music. So I sat down with him, taught him everything I knew about the creative process of Beethoven and how composition works. And he said, let me think about it, called me back next week and said, Matthias, I thought about it. It's quite straightforward. Composing music is just like email, autocomplete. Who I, you can imagine that moment. I thought like, ok, well, maybe the wrong person and this is never going to work. But then he told me no, but look, every note is a function of all of the notes that came before. And it, it makes sense when I have a piece of music, the next note is related to what came before. We can actually try it out. Now together, I'll sing you a melody. I'll stop at some point and you the note that I left out. OK. Are you ready? OK. Ta ta, ta, ta, ta, ta ta. Yeah. You see every node is a function of all of the notes that came before. So we set out, we said, well, let's use this technology, the transformer architecture that Google invented in 2017 and uh create a G BT for music that's trained on late 18th century, early 19th century music and then fine tuned with the works of Beethoven. And then we use the sketches that Beethoven wrote down as prompts. And we let the machine continue these ideas. And then a group of musicologists combine the material, go through it, improve on it, give it back to the machine and this way iteratively create one possible version of 1/10 symphony at the time. Most people thought, oh, they are crazy. This was never going to work. Even some Google folks said that to the CEO of Deutsche Telekom. But we were pretty certain that at least, you know, we would create something that would pass as a potential musical work of that time. So I want to play you an example of that tense symphony which starts with one of these sketches that Beethoven had written down. It's basically 1 to 1 what he wrote down in his book. And at some point, the A I will take over and I invite you to play this game for with me where you raise your hand when you think this is the moment where the A I comes in, of course, I'm going to give you um you know, the solution afterwards. But this is one minute of music, close your eyes, don't get distracted by the people around you and just be bold, raise your hand whenever you think this is where Beethoven stops and A I begins was generated by this algorithm. Uh Granted it was then, you know, going through other algorithms as well. We not only produced an A I that could continue the sketches, we produced one that would find the suitable accompaniment, one that would stitch together pieces of A I generated music and then one final A I that would take the output of all of the other AIS and say this note will be played by that instrument. This note by this instrument, an orchestration engine. This is where we really reached the end of the line because that A I wasn't that successful. So we actually hired a human in our team to orchestrate the final version that you just heard being performed by a real orchestra with real humans. This collaboration started in 2019. As I mentioned in 2020 we were supposed to premiere it. Uh but COVID came along and so we had to postpone, we had one additional year to work on it. And in that time frame, I got a second phone call from Deutsche Telekom. Hey Matthias. Uh Robbie Williams has heard about your Beethoven A I and he thinks he would like to compose with Beethoven. Is that possible? I said, well, you know, what can I say? I don't say no if Robbie Williams wants to use an A I that we produce, so we'll find a way to make it happen. Of course, the question was, how could we do that? I flew to London. I explained it to Guy Chambers, the composer that worked very closely with him and Robbie's manager exactly what we were doing. And they said, well, we'll think about it and they called me back the next day. We want to do angels. Can you recompose angels in the style of Beethoven? And I said, I don't know, I really don't know if that's possible. Um The, the problem was, you see, we had this A I that was trained on 18th and 19th century music and that could very well deal with Beethoven's style. But how do we get Robbie Williams in there? This is when the composer on our team WV, I had this amazingly genius idea. He said, well, we'll take a famous melody of Beethoven and we combine it with bits and pieces from Robbie's Angels. And this combination is what we then use as a prompt. And so by prompting the A I with something that had already the main motific uh content of angels in it. It was capable of producing music that combined Robbie williams' Angels with the style of composing of Beethoven. There was one problem we thought, OK, well, our A I is pretty good at creating long stretches of music, but Guy Chambers told us no. Um you have uh four seconds for the intro and you have another 30 seconds for where the solo is and everything has to happen in there. So we would try out different types of uh combinations of music. And finally, we found one that worked very well. We combined uh the Moonlight Sonata with uh with some people are laughing. Yeah, uh with, with angels and I have another excerpt for you that uh is, is a little mash up of what we did with Robbie. You can listen to the whole thing on his newest album. Um And please let's listen to Angels with Beethoven A I. The point of this was really to collaborate and we started out to create this one A I that could, you know, do something very, very specific. And we ended up providing a tool set to one of the great artist of our time to collaborate with. And I think this is the, the move that we see in a lot of what's happening in generative A I and music right now, we see tools that become collaboration partners and that are really um instances of inspiration of transformation of working bodies for composers. And if you ask me what the future of all of this is, then I'd say all of the tools that right now are sort of text to finished song where you basically say, I want to have a track which is 15th century style with green sleeves and heavy metal combines and then you get finished track. This is, this is all nice and very good, but it will be superseded by tools that will be directly integrated into the software where we compose music where we have our inspirational A I tools. Now, I want to leave you with another excerpt this time from the Beethoven 10th symphony. It's actually a work of 25 minutes and I condensed it for you here to, to a couple of minutes for you to get some sense of how the piece sounds when you hear it performed by an orchestra, we can roll the tape. Um This was done a couple of years ago in Hamburg and the Philammon, this is the Beethoven orchestra conducted by Dirk Kaftan. And there's a little surprise there, an instrument that you wouldn't expect in a Beethoven symphony. So enjoy.

Emran Mian, Director General, Digital Technologies and Telecoms, Department for Science, Innovation and Technology (DSIT) Robyn Scott, Co-founder and CEO, Apolitical Guy Williams, Head, UK Defence, Palantir Technologies Moderator: Alan Murray, Chief Executive Officer, FORTUNE

Brainstorm AI London 2024: AI’s Impact On The Geopolitical Landscape

Photography by Joe Maher / Fortune Brainstorm AI

Emran Mian,, Innovation and Technology (DSIT), Robyn Scott, Apolitical, Guy Williams, Palantir Technologies, Alan Murray, FORTUNE

Thank you. We have a lot to talk about and a very short time to talk about it. Enron, I'm eager to talk to you about what can government actually do to mitigate the downside risks of A I, Rob? And I'm eager to talk to you about how can government use A I to just be better at what it does. But I, I think I have to start with you uh because you're in the defense business, uh you work for pun here but work closely with the UK government in the last 48 hours we've all seen on television, a pretty extraordinary uh event that Iran sent more than 300 drones, ballistic missiles, cruise missiles aimed straight at Israel. And we're told that 99% of them were shot out of the sky before they hit the ground, which is if true. Extraordinary. What is this event that we've all witnessed in the last 48 hours? Tell us about the changing face of national defense. I think the thing that's really highlighted for me is um if you go back in time, only, not even 10 years, but say four or four or maybe even three military operations are all about the speed of decision making decisions at the speed of relevance I think is a term often used. And historically, we've seen things like if you go all the way to before something is launched, it might be looking at satellite imagery to understand what things have moved, what changes have taken place. Now, with the introduction of computer vision models, you're able to take a two kilometer by five kilometer image. And within a second, identify all of the things that you might be interested in, whether it's changed or whether it is something that you, you recognize historically, that might have taken analyst hours to get to that point on one single image. And now it would take thousands of 300 of them with a pretty limited time. Yeah, exactly. And so the idea of identifying where these things might be coming from and deciding these are the ones of interest and then maybe tasking things, the collection assets that you might have, which could be sensitive or what have you and then passing that structured data through a model to then understand what are the interesting things do you care about all the things that might be good indicators? And more so it's a secret, it is a sea change. Totally, the compression of decision making times has been, you were directly involved. My understanding is you were directly involved in working with Ukraine to very much the same weaponry, by the way, both the offensive weaponry and the defensive weaponry, uh that Russia was using against Ukraine. What did you learn in that experience? Uh The Ukraine example is, is fascinating and very unique in the fact that you have seen, um, a complete tech sector nationalized into the defense sector and the speed of change that they've been able to deliver in terms of transformation and the use of A I and other technologies has been incredible. Well, they had some help, a little help. Yeah, but I mean, again, one of the, one of the earlier um commentators are talking about the, the getting your data in a healthy place and then run the models across it. And that's a very important part. The thing is we're fortunate in both of those cases that the defensive capabilities were so strong, but these are going to empower offensive capabilities as well, right? I mean, it's gonna be, it's gonna be a race. Yeah, I think the idea of strategic strategic competition is definitely, definitely with us and I think in terms of governments investing into future technologies and A I is, is gonna be pretty um pretty game changing, deterministic uh Emma and I, I want to back up a little bit and, and get to the broader question of what A I means for governments, uh which is the question you wrestle with every day in your position with the UK government, I guess. The thing is someone said this from the stage earlier today is this is moving so fast, so fast. How can government hope to keep up with the pace of change and mitigate the negative consequences in any way? Yeah, so a big part of our approach has been to do things differently. So we've been much more open to bringing in expertise from outside of government than perhaps sometimes we are as so we brought in lots of expertise from A I firms from academia, including Ian Hogarth who chairs our A I Safety Institute I think was speaking earlier in the conference and we've been much more open to the expertise and sometimes we are as government. We've also moved really, really quickly and in order to move as quickly as that, you really need political support all the way up to the top of government. And we've absolutely had that we've had a prime minister who's chosen to prioritize this set of issues. And in particular, he lent into saying that the UK would host the first A I Safety Summit which we ran in November last year. And I think that's another important lesson for us in terms of dealing with some of the risks or some of the potential downsides of artificial intelligence. We've got to do it with other governments. We've got to do this together. We've got to build international governance around this and that's why it's been so important to us to kind of begin to lead that conversation. So we hosted the first summit last year, we're co hosting the next summit with Korea next month, we are also trying to collaborate really extensively on A I safety testing with other countries. So our A I safety institute just signed a partnership agreement with the same institute in the US. And we're talking to these capabilities as they begin to be developed in other countries as well. So kind of that collaborative approach I think is really fundamental. And so what is sort of the top three things that you think you can accomplish with this A I safety effort? So I think the first thing is to bind this set of countries together, to build that connective tissue between the countries which have frontier A I firms and which you would expect to be at the frontier of A I being deployed across the wider economy. I think the second thing I'd say is there's still a lot to do in terms of understanding the science of A I safety. So there's a lot of discussion about what's the right approach to regulating frontier A I. And our view has been that the best thing to do first is to figure out what does good A I safety testing look like and set standards for that and to do that in collaboration with other people who are at the frontier of doing that work. And that's what the A I Safety institute been focused on. We want to understand what really good safety testing looks like, do that in collaboration with the firms and do that in collaboration with other countries. And is the end result, you give a seal of approval to certain models and not to others. What does this look like? Five or 10 years down the road? I think it's hard at this point to say what it looks like. Five or 10 years down the road. I think our approach at the moment is let's establish the basic science, let's figure out what good safety testing looks like. And alongside that, let's use our existing regulatory system for the harms that are manifesting potentially here. And now, and I think that's another important part of our approach. The digital markets that we're talking about here are markets that we already regulate in important ways. We have just passed legislation to allow us to better regulate for harms that occur online. Um Many of those harms will now also be created with A I, but we've, we've just regulated for that. So we should use that regulatory framework to deal with A I generated harms in the same way as we were planning to use it for other things. Yeah. Robin. Uh uh you're working uh from a different angle on these issues. You're trying to figure out how can A I make government better at what it does. Can you talk a little bit about what you're doing and where you're seeing results. Yeah. So the perspective I bring to this is running a political, which is a global network used by about a quarter of a million public servants. We also do online training and as part of that, we have a government A I campus and we've trained already about 10,000 public officials around the world. So that's our lens. The headlines are the potential is immense. The Turing Institute came out with a report recently which speaks to speak to this in the UK context, but it's generalisable. There are about a billion transactions a year between government and citizens. About 100 and 40 million of those are complex but repeatable tasks. And of those uh the the institute estimates that about 84% can be uh automated with A I. So incredible potential to cut into the work of government, improve citizen services, improve the return on taxpayer dollars. There's also a secondary reason for doing this and for adopting A I, because better users of A I are better regulators of A I and they're better buyers of A I. And I think, you know, the UK really has shown leadership on the general adoption front in government as well. And is there a willingness to do this gets back to the retraining issue? We were talking about a few minutes ago. Is there a willingness in government to really embrace the effort? Yeah, absolutely. So it's very interesting to look at government's current data sort of position on A I. First of all, somewhat surprisingly, your average public servant, we've done a lot of polling on this is more positive than negative on the potential of A I that doesn't hold necessarily for prior technologies. And given all the talk about concerns, we've been surprised how generally optimistic they are. There is much more optimism in emerging markets versus developed markets. The conversation in emerging markets is all about this is going to solve personalized education. This is going to bring great health care. The conversation in developed markets tends to have much more of a bent to work towards safety and harms and economic nationalism is also uh becoming a more and more important uh force, sorry A I nationalism rather. And we've seen big plays recently in the UAE and India and expect to see many more around the world. So all those are context for adoption, but there is general willingness to use it. The vast majority of public servants have already played with generative A I sometimes I think under approved guidance, probably sometimes somewhat off the books and that's only going to increase pretty dramatically. Yeah, I'm glad you mentioned economic nationalism because one of the things that makes this conversation about geopolitical effects uh interesting and challenging is the existence of China not represented on this, on this stage, China is certainly an A I superpower. There are some people out there who seem to think it will become the leading A I superpower. Uh and it has a very different background set of values that could be on the other side of any uh uh actual uh uh warfare. Um It, it takes a very different approach to many of the questions uh to use it a simple example. Uh The UK has said very clearly, we're not going to use facial recognition or other technologies to pick people out in a massive crowd crossing the street. I believe China has done the exact opposite so that we can make great gains in social cohesion if we have the ability to do that. So how, how does China's strength in A I affect these conversations, Emran, I'll start with you. Yeah. So our approach has been to include China in the international conversation about safety. Uh So China will represent it at the A I Safety Summit in Bletchley Park. And we think that's important because China produces a lot of fundamental research in A I. Uh and it has a very busy and a very active and a very interesting A I ecosystem. Um So having China as part of this conversation, we think is important. But at the same time, we've been really clear as a government that where China is engaged in activities that affect or harm UK interests, we will say so clearly and plainly. Uh so we've done that in the past few weeks we have called out where China has been engaged in uh offensive cyber operations in the UK. And we think that those two things are entirely consistent. Well, there are a lot of people in the US these days who think maybe they aren't consistent and that should we be sharing all our research with China if we're going to be in conflict over whether it's cyber or actual warfare? Yeah. And that's why our approach will not be to share all of our research with China. We're very clear on there are areas of research that you, that we will protect and that we do protect Robin. How do you think about this? We're looking very much at the practice, the applied side of A I in government. And one of the interesting general admissions, I think there's a general admission around thinking around capability in government. There's a lot of talk about, we must do this, there's a lot of talk about adoption, but there isn't the focus on building the kind of diffusion capability. So you can actually roll it out across government. So we look at that and then we look at how the exchange of best practice happens because the field is moving so fast, you are never going to win. If you just take top down guidance or uh you know, directives, you've got to talk to others who are in the field. And I think one of the interesting dynamics is actually the vast majority of countries are quite good at collaborating and sharing best practice. We see a lot of that on our platform, China kind of doesn't take part for the most part explicitly or at least in the same way in that sharing of best practice. So the question I'm interested in is is there going to be a siloing around how A I is rolled out in the Chinese government versus elsewhere? And should we be trying to get more permeability so we can, so we can be mutual learning around non controversial things? Like how do you get great services to citizens? How do you bring down the carbon footprint of these A I technologies? We heard about that earlier. Yep and Guy in defense, how, how do you think about China? Um Well, I mean, as you point out that facial recognition is reportedly like fantastic, but when it comes to things like large language models, apparently there's a slight limitation because they um restrict elements of the internet available to specific areas. So there is a, there is a theory that they might be slightly behind but to pick up Robin's Point there, I think what's really important is almost not ignoring it but just more internal focused of looking at an organization and in parallel to the regulatory and the safety side is like, how do you really push the use of these technologies to win? Like what are you doing day in day out to make the most of this tech to actually improve your, improve your work. And there's often this concern about losing jobs, but we like to view it as um in a marvel context of Team Iron Man versus Team Ultron. And it's how do you bring this technology to humans so that they can be 1000 times better at their job? And with the view to how you actually win in the long run, was that a national defense answer? Yeah. Well, I, I think so. I think if, when you're an analyst working in the defense in defense intelligence, if an operational commander is, how do you, how do you bring? But you, but you have to be thinking about, do our cap are our cap, do their capabilities match ours are our capabilities, period. That's the nature of defense. And I think that's the important thing of, of balancing the innovation and driving innovation forward in parallel to the safety side rather than letting the safety side restrict what you do in terms of innovation. Ok. Last question, I'd like to ask all of you to answer. If A I ran the government, what would we be doing differently today? What would we be doing that we aren't doing? And I'm going to make you go first. Um, we'd be doing meaningful work, not busy work. It would be my blunt answer. Still too much busy work in government. I think, I, I think it, it's very hard and, and A I can help us get rid of it. I hope so. Yeah, Robin. So there's been some data suggesting this is an aside that a majority of people, I think it's younger people in Europe actually want A I politicians. So they want A I to run the government as opposed to the politicians, as opposed to the flesh and blood runs of the coming, coming to government sooner than you may think. But in terms of what that would look like, assuming it's benevolent and aligned A I, I think the the potential the dream is that you get much better, much more personalized services, hyper personalized services to citizens and you can invest a lot more in those because you are saving money because of so many efficiencies that you're also finding Iran final word. Yeah, I'd add, I think government is a complex activity and it's too often con noted as being an elite activity. And I think A I gives us some real opportunities to, to deal with that. I think the way in which A I can help to simplify and explain things to citizens, I think has real power to it. I think the way in which it can help to simplify interactions with citizens, I think has real power to it. Um So I think breaking down that elitism that I think a lot of people experience in their interactions with government, I think, yeah, I can help us with that. Well, thank you all, you've given us a lot to think about. Thanks for being part of brainstorm A I.

Williams

Williams in Fortune Rankings

Williams stock chart

Videos