Media Briefing: Unlocking the North Star for AI Adoption, Scaling and Global Impact / DAVOS 2025

Summary

This media briefing at the World Economic Forum in Davos focused on AI adoption, scaling, and global impact across industries. The panel, featuring experts from the World Economic Forum, Siemens, Dow Jones, and Cohere, discussed the transformative potential of AI and its current state of implementation.

Cathy Li from the World Economic Forum highlighted key findings from their white paper series, noting significant projected market growth in various sectors due to AI adoption. The panelists emphasized the rapid pace of AI development, with Aidan Gomez of Cohere expressing surprise at how quickly the technology has advanced beyond initial expectations.

The discussion touched on several key themes, including the importance of privacy and data integration in AI implementation, the need for trust and ethical considerations, and the potential for AI to enhance productivity and sustainability efforts. Cedrik Neike of Siemens stressed the importance of AI in addressing global challenges like climate change, while Almar Latour of Dow Jones highlighted AI’s impact on media and content creation.

The panelists acknowledged that while AI adoption is still in its early stages for many businesses, there is growing optimism as familiarity with the technology increases. They emphasized the need for collaboration between industries and the importance of addressing talent gaps through reskilling initiatives.

Challenges such as misinformation and the energy consumption of AI systems were also discussed, with panelists stressing the need for responsible AI development and deployment. Overall, the discussion painted a picture of AI as a powerful tool with the potential to revolutionize industries, but one that requires careful consideration of its societal impacts and ethical implications.

Keypoints

Major discussion points:

– The rapid growth and adoption of AI across industries, with projections for significant economic impact

– The importance of trust, privacy, and responsible AI governance as adoption increases

– The need for integration of AI with company-specific data and systems to realize its full potential

– The transformative effects of AI on workflows, product development, and business models

– Challenges around talent gaps, scaling AI implementations, and addressing risks like disinformation

Overall purpose:

The goal of this discussion was to provide insights on the current state of AI adoption across industries, highlight opportunities and challenges, and discuss how businesses can effectively implement AI technologies. The panel aimed to share perspectives from different sectors on AI’s transformative potential.

Tone:

The overall tone was optimistic and forward-looking. Panelists expressed excitement about AI’s capabilities and potential positive impacts, while also acknowledging challenges that need to be addressed. The tone became increasingly positive as the discussion progressed, with panelists emphasizing the exhilarating pace of progress and AI’s potential to solve major global problems.

Speakers

– Anais Rassat: Moderator

– Aidan Gomez: CEO of Cohere

– Cathy Li: Head of AI, Data and Metaverse at the World Economic Forum

– Almar Latour: CEO of Dow Jones, publisher of the Wall Street Journal

– Cedrik Neike: CEO of Digital Industries and Managing Board Member of Siemens

– Audience: Various unnamed audience members asking questions

Additional speakers:

– Kazuo Teranishi: Journalist from Japanese newspaper Asahi Shimbun

– Christoph Kackmann: Reporter with Handelsblatt German business daily

AI Adoption and Global Impact: Insights from World Economic Forum Panel

A media briefing at the World Economic Forum in Davos brought together experts from various sectors to discuss the current state and future prospects of artificial intelligence (AI) adoption across industries. The panel, moderated by Anais Rassat, featured Aidan Gomez (CEO of Cohere and co-author of the Transformer paper), Cathy Li (Head of AI, Data and Metaverse at the World Economic Forum), Almar Latour (CEO of Dow Jones), and Cedrik Neike (CEO of Digital Industries at Siemens).

Economic Impact and Industry Transformation

Cathy Li highlighted key findings from the World Economic Forum’s white paper series on AI adoption. She presented data showing that early adopters of generative AI have achieved up to 2.4 times greater productivity and cost savings of 13%, underscoring the technology’s transformative potential across multiple sectors.

Cedrik Neike emphasized AI’s role in improving efficiency and sustainability in industrial processes. He drew a parallel between AI adoption and the introduction of robotics, suggesting that AI could empower workers rather than replace them. Neike stated, “We need to make AI an empowerment of the shop floor worker because it’s the same which happened when robots came. Everybody was afraid they’d lose their jobs. Now, the countries with the most robots have the most industrial jobs. The same will happen to AI in the industrial space.”

Almar Latour provided insights into AI’s impact on media and publishing, highlighting how it is revolutionizing workflows and products in the industry. He stressed the importance of protecting intellectual property rights in AI training and usage, particularly for smaller publishers who may struggle to safeguard their content.

Challenges and Considerations in AI Implementation

While the panelists were generally optimistic about AI’s potential, they also addressed several challenges and considerations for responsible implementation.

Cathy Li emphasized the need for responsible AI adoption and governance frameworks. This sentiment was echoed by Cedrik Neike, who stressed the importance of building societal trust in AI technologies. Neike highlighted the complexity of leveraging proprietary data while respecting customer rights, stating, “We have more design data on how things are being built at Siemens than anyone else, but it belongs to our customers. So we need to find something that between you, our customers, etc., we can build it.”

Aidan Gomez addressed barriers to AI adoption, including privacy concerns and the need for customization. He also raised the issue of potential misuse, such as disinformation and social media manipulation, emphasizing the need for effective defensive measures against AI-generated misinformation.

Almar Latour proposed commercial collaboration as a potential solution to address intellectual property concerns in the media industry. He stated, “Without that value being recognized, you run the risk of really damaging journalism. And so there is a two-pronged answer to that from the industry at this point. One is, at Dow Jones, we like this particular path the most, which is commercial collaboration.”

Future Outlook and Technological Advancements

The panelists expressed excitement about the rapid pace of AI development and its future potential. Aidan Gomez highlighted recent technological advancements, particularly in the realm of AI agents. He explained, “We’re starting to see the first production systems of agents start to enter the sphere, and with that is a technological shift. There’s two components that are crucial for agents. One of them is reasoning… And the second piece is learning from experience.”

Cedrik Neike discussed Siemens’ work on training AI models with CAD data and time series data, emphasizing AI’s potential as a critical tool for solving major global challenges, particularly in the realm of sustainability. However, he also noted the projected increase in energy consumption due to AI systems, highlighting the need to balance AI advancement with sustainability goals.

Collaborative Approaches and Future Directions

A key theme that emerged from the discussion was the need for cross-industry collaboration to advance AI responsibly. The panelists agreed that no single company could solve AI challenges alone and emphasized the importance of partnerships in developing and implementing AI solutions.

To this end, Cathy Li announced the launch of the MINDS (Measuring Impact of New Digital Systems) program by the World Economic Forum, aimed at highlighting real AI use cases across industries. She also mentioned the AIGA (AI Governance Alliance) initiative, which focuses on developing governance frameworks for AI.

The panel noted a significant shift in AI discussions at Davos compared to the previous year. Cathy Li observed that conversations had moved from theoretical possibilities to practical applications and real-world impact, indicating a maturing understanding of AI’s potential and challenges.

Unresolved Issues and Future Challenges

Despite the overall optimistic tone, several unresolved issues were identified. These included:

1. Effectively addressing AI-generated disinformation and social media manipulation

2. Balancing AI’s energy consumption with sustainability goals

3. Closing the skills gap to enable wider enterprise adoption of AI

In response to a question from a German reporter, the panel discussed challenges of AI adoption in businesses, including the need for specialized knowledge, data quality issues, and the importance of change management in implementing AI solutions.

In conclusion, the panel provided a comprehensive overview of the current state of AI adoption across industries, highlighting both the transformative potential of the technology and the challenges that need to be addressed for responsible and effective implementation. The discussion emphasized the need for collaboration, trust-building, and ethical considerations as AI continues to reshape the global economic landscape.

Anais Rassat: Hello, welcome everyone. Welcome to this media briefing on Unlocking the North Star of AI Adoption, Scaling and Global Impact, here at the World Economic Forum Annual Meeting in Davos. The theme of today’s media briefing is AI. So we have AI as one of the most important technological changes in the 21st century, growing rapidly, and especially we have generative AI impacting how society, individuals, industries are interacting, communicating and working. But we find that businesses are not necessarily prepared for the change that is coming today. So today the World Economic Forum has published a series of white papers on industries in the intelligent age that provides solutions and use cases. And we also have this series of white papers that is part of the AI Governments Alliance, so AIGA, at the World Economic Forum, a broader initiative. So joining me today is a series of panellists. So we have to my left Cathy Lee, Head of AI, Data and Metaverse at the World Economic Forum. Hello. Cédric Neike, CEO of Digital Industries and Managing Board Member of Siemens. Almar Latour, CEO of Dow Jones, a publisher in the Wall Street Journal. And Aydan Gomez, CEO of Cohere. Hello. So thank you. We have journalists and participants in the room. And for those of you following online, or if you’re sharing on social media, please use the hashtag WEF25. Let’s start perhaps with Cathy. If you can give us an overview of this series of reports of white papers. So what are the insights that we’re having on AI in these series? Thanks, Agnes.

Cathy Li: Thanks for having me. So first of all, just a very quick overview. The work is done not by one organisation, the forum, it’s not done by one firm. It’s really a community of 463 organisations are part of the AI Governance Alliance that we’ve launched a year and a half ago. So this work is really an accumulation of all of the knowledge and insights with the forums, different industry communities together. So today, I just wanted to share some of the key stats through the research. For example, we do find AI is transforming many of the ecosystems. Take the global health systems as an example. We do see a lot of the applications in predictive care, personalised medicine, and operational efficiency, driving a projected market growth of 43% annually to reach 491 billion by 2032. Financial services also have invested heavily, probably one of the most heavily invested markets in AI, with spending expected to grow from $35 billion in 2023 to $97 billion by 2027. And they focus very much on fraud detection, operational efficiency, and ethical governance frameworks. And I can go on and on, but I do encourage you to take a look at our reports. In terms of the total economic and productivity gains, there are significant outcomes to be projected when it comes to AI adoption. The early adopters, particularly of generative AI, have achieved up to 2.4 times greater productivity and cost saving of 13%. The contribution to the global economy is projected to reach between $7.6 and $17.9 trillion by 2038. Workers’ readiness emerges, as you can imagine, as a key focus area, with 74% of companies reporting challenges in scaling AI due to talent gaps. The reports emphasise the need for rescaling and upscaling initiatives to empower workers and effectively engage with AI technologies. Finally, I wanted to emphasise sustainability. Sustainability goals are definitely highlighted throughout the report, with AI playing a pivotal role in optimising energy use and supporting decarbonisation. At the same time, we also need to address the rising electricity demand of AI systems, as AI-related energy consumption is projected to grow by 50% annually through 2030. This data centre electricity use alone is expected to reach over 3% of total global demand by the end of the decade. Only by aligning the technological innovation with environmental stewardship, we can address the AI energy paradox. Last thing I would say, we are launching the work on industry and communities, but let’s not forget the other parts of the work of the AI Governance Alliance, which very much focus on governance and ethics, trust and cybersecurity fairness and transparency. So we’ll continue to push for those efforts and encourage the society and industries to adopt AI responsibly.

Anais Rassat: Thank you, Cathy. I’d like to hear from our other panellists. Friedrich Neike, your company Siemens has been a leader in integrating AI into industrial processes. Can you tell us how this white paper series of reports informs your sector? And maybe you can also give a little introduction to exactly what your company does. We’ll do that.

Cedrik Neike: I mean, since the 70s, we’ve been working on predecessors of AI and using this technology. Just for everybody, Siemens is mainly in the area of infrastructure, which is buildings, trains, energy systems, or a lot of the factories. Most of the factories run on Siemens infrastructure. But what we are doing is called OT, operational technology. The world loves IT, but it actually runs on OT, your factories, your power plants, et cetera. And the idea was these environments create lots of data, but we haven’t used it in the past. And as we said, is this finally we can use it? Because with a view of what Cassie said in sustainability, we have to do more with less. And that technology enables us to do it. And I think we’ve done a lot to, you can now, if you designed lots and lots of products, you can take this and with AI, the best, most sustainable product can be shared with you. Because 80% of sustainability is defined in the design phase. You can run your electricity system, your factory more efficiency, thanks to AI. You can do a lot of things and we’re applying it and we’re putting it in place. And I’ll just give you a couple of examples. We’ve been launching a co-pilot, which with more than a hundred customers and together with Microsoft, because AI is one of the things you cannot do alone. You do it with others to help, for example, it helps you design the products. It helps you run the machine efficiency and it lets you talk to the machine. So somebody from the shop floor can. And say, what’s wrong? And the machine would say, well, I know what’s wrong. This part is missing. Please repair me. And you have to do these and these steps in multiple languages. That’s a huge capability. Or I compared myself to the other 100 machines. That’s what you need to do. So we’re really putting this into action. And since there’s more machines and infrastructures out there, there’s going to be more demand for AI. The key thing is going to be trust, societal trust. When I ask my customers, a lot of them say, I love AI. And then I said, what do you think society does? Well, they’re afraid of it. And when you’re on the shop floor, we need to make AI an empowerment of the shop floor worker because it’s the same which happened when robots came. Everybody was afraid they’d lose their jobs. Now, the countries with the most robots have the most industrial jobs. The same will happen to AI in the industrial space. So there’s a lot happening. We’re leaning in. We need to make sure there’s trust. But basically, the whole analysis we’ve done together with the World Economic Forum is

Anais Rassat: proving that if you do it cleverly, you can have a huge advantage. Okay, huge advantage and impact. And impact, absolutely. Excellent. Thank you. Almar Latour, you work in the context of media and digital services. So how is AI transforming that space? And again, you can also, if you want, give an introduction of your role also. Yeah. Thank you. It’s perhaps easier to ask, how is it not transforming this space? Because it’s really touching just about everything.

Almar Latour: A quick word on Dow Jones and the Wall Street Journal, we’re an information service and business news company, Wall Street Journal, one of the leading news outlets for the business community globally. And so I would call out three areas where there is a tremendous impact industry-wide. First of all, intellectual property and making sure that archives and the proprietary data and high-quality journalism that has been invested in over the years, has accumulated over decades, that the value of that is recognized by the platforms. Without that value being recognized, you run the risk of really damaging journalism. And so there is a two-pronged answer to that from the industry at this point. One is, at Dow Jones, we like this particular path the most, which is commercial collaboration. Let’s understand what the value is. Let’s have the market determine what that value is and have a commercial arrangement. But if that value is not recognized, if IP rights, in our view or in industry’s view, are abused, then it ends up in a legal battle. And you see multiple pieces of evidence for that. So that’s a huge category. We can come back to that later. One additional element to that is there are large publishers, large media houses, and large players like us can actually afford to do battle with large companies that play in the AI space and can also have leverage in negotiations and commercial negotiations. If you are a small publisher in, say, Germany or in Asia or even in the U.S., that’s much harder to do. We have had for years a platform called Factiva, where we actually pay royalties for usage of content. There are about 15,000 different publishers, a billion and a half different articles on that. And we intend to convert that platform into also a vehicle, a platform that can help with payments for AI rights for publishers. And so that is critical to us. So that’s IP, I would say. Without that foundation, you cannot really build further from there. Two areas that I’ll touch on quickly. Workflow improvement. This goes for almost any industry, but think of the creation of headlines, think of the creation of publishing in 100 different languages, right? There’s super labor-intensive things that now can go much faster. That’s transformative. It will be, I think, commercially incredibly beneficial. And then third is new products. If you think of Dow Jones, but many companies could do the same, as a Rubik’s Cube. Inside of it, a lot of information, and you can rearrange it, and each tile is a way to get information out of our company. And the rearranging is what the AI can help us with tremendously, and I think that will result in a lot of co-creation with clients, and a lot of creation of new products that will also lead to new revenue streams. And at the same time, it will create a lot of new competitors, and I think that’s really healthy as well.

Anais Rassat: Thank you. Very interesting. I’ll turn to our last panelist, Aiden Gomez, so you’re CEO of Cohere. And as we’ve discussed at the beginning, generative AI is really transforming industries at an unprecedented pace. Your organization supports companies in that transformation. Can you tell us a bit more about your company, and what businesses need in this transformation? Yeah, of course.

Aidan Gomez: So we build large language models and deploy them inside of enterprises, and we’ve watched this. We’ve been doing it for five years, so before it was a big thing, and we’ve watched the market really shift over the past year. It went from a lot of experimentation into production, but there are a set of barriers to this. If you really realize the value in adopting generative AI, the two things that you need, the first is privacy. So these models are only as capable as the information and systems that they have access to. If you’re compromising on privacy, they can’t reach all of that data, and so they can’t be as useful as you want them to be. The second piece, you can call it either customization or integration. The thing is, they’re trained on the web, and so they don’t know anything about you, your company, your specific domain, and so we’re very focused on bringing your data into the fold in an extremely private and accessible way, and making that process easier. Because right now, the barrier to adoption is the fact that the models simply don’t have access to the information to answer the questions that you need them to, and they don’t have access to the systems that your humans use in order to help augment work or automate work. That brings me to another major point, which is the shift towards agents, which has been a big one and is coming to fruition now. We’re starting to see the first production systems of agents start to enter the sphere, and with that is a technological shift. There’s two components that are crucial for agents. One of them is reasoning. So this notion that humans, when we’re presented with a problem, we need to think through how to solve it. We can’t immediately blurt out the answer. But in all previous generations of models, that’s been the expectation. So we’re seeing a shift now to models that can break down problems, solve components, and piece them together. And the second piece is learning from experience. So previously, you would train a model, maybe you spend hundreds of millions of dollars training this model, you push it to production, and it’s frozen. That end of training is the end of when it learns. And so that will start to shift, and we’ll see models that learn from experience, and that leads very naturally into personalization. As you interact, as your employees interact with the model, it learns you, and it learns your preferences, what you want it to do, and it learns from you. So I think with these two technological shifts, there’s a whole tier of productivity that we’re going to gain. I was surprised, and pleasantly surprised, to hear the 2.4x gain in productivity. That’s the floor. Because right now, it’s extremely rudimentary.

Anais Rassat: And what’s coming is much more advanced. So I think there’s lots to look forward to. Can I ask a question? Yes, of course, Cathy. I think also Aiden has been modest.

Cathy Li: He was also one of the co-authors of the Transformer article back then that lays the foundation for the generative AI hype and the whole economy we’re talking about. So actually, question to you is, Aiden, because you were with Google, I think, in that time. You drafted the paper with other co-authors. Did you expect that this would happen the way it has today? No way. No way. I don’t think anyone did. I mean, there’s no way.

Aidan Gomez: And there were definitely indications that it was a promising architecture for doing this type of work. But it’s exceeded my expectation. Today, where we are technologically, if you asked me seven years ago, I would have said it was coming in a quarter century, half a century. It’s a huge shock to me and everyone else. I’m going to break the model, but I’m going to have another question.

Cedrik Neike: We were looking at the LLM models, and one of the key things is a lot of large language models. But we’re actually starting to train them on CAD data, on time series data, on very industry-specific environments. And we see that these models, even with very small data sets compared to where you train them on, are extremely useful. So we can now, on a picture, just find out, okay, how would I have to have the CNC machine work on it? How do I design it? So we see that the key access is not only privacy on personal and that sort of workflow, but also in the design phase and industrial sort of production phases. Yeah, one of the big barriers to progress in LLMs is the fact that that sort of data, it doesn’t exist out on the web.

Aidan Gomez: And so you can’t actually experience it as a model. And so via partnerships, I think we’re going to see another big step up in terms of application.

Cedrik Neike: We have more design data on how things are being built at Siemens than anyone else, but it belongs to our customers. So we need to find something that between you, our customers, etc., we can build it.

Anais Rassat: And that’s in the paper. We will not solve this AI topic on our own. It will be partnerships which will enable that. Thank you. I’d like to ask one more question to Cathy Lee, and then we’ll take some questions from the audience. As part of the forum’s wider and broader initiative, there’s a new program called MINDS that was launched also today. Can you tell us more about this program and what it’s doing to support AI adoption?

Cathy Li: Yeah, absolutely. So this MINDS program stands for Meaningful, Intelligent, Novel, Deployable Solutions. What do we mean by that? Because you see a lot of the AI application and use cases, but we also find out that I think more than only 30% of AI projects within existing organizations. Actually, 30% of AI projects within the existing organizations fail to progress beyond the proof-of-concept stage. I have to say I’m a little bit surprised. I thought that number is actually a lot higher. So what we really wanted to do is to really separate the noise from what’s actually happening, what are the real use cases, and how do each of the organizations actually implement the technology, integrate the technology into different processes, your change management strategy, your people strategy, everything. So that’s why we’re launching this program today and opening up to application, and we’re hoping to really work with all the key leaders from different industries and sectors and be able to highlight some of the real use cases that can be shared across different sectors and to promote the kind of cross-industry collaboration that you’re talking about. Thank you, Cathy. I’d like to invite now some questions from the audience, from the media, and specifically questions related to the topic of this media briefing.

Anais Rassat: Yes, can you please introduce yourself and the name of your media and then ask your question? Thank you.

Audience: Hello. My name is Kazuo Teranishi. I’m from Japanese newspaper Asahi Shimbun, and I have two questions. The first one is to Ms. Lee. In this white paper, of course, disinformation is one of the highest risks in the risk report you published last week. According to this report, this white paper, 50% of enterprises will adopt product or service to address disinformation, which is only up from less than 5% in 2024. It’s only up by 2028. So it is only 5% compared to in 2024. It’s only 5% more companies try to adopt some media to address the disinformation risk. What is your message? Is it only 5% or is it enough? What is your message?

Anais Rassat: Great question, and I might turn to Alma later just for that. From the research perspective, we definitely see, yes, there’s a slight uptick in terms of deploying technology also to tackle some of the challenges that are potentially brought by technology, such as misinformation and misuse.

Cathy Li: We always say technology is a double-edged sword. You can use it to do good things. You can use it to do bad things. The good thing is the technology is progressing. It’s able to detect machine-generated information a lot better just by predicting the text pattern and in images as well. Obviously, some technologies have not really caught up in terms of – as we move into multimodal AI models, obviously there will be more challenges. But I do think the technology is progressing, and this issue will need to be addressed. It’s not a can or not. It’s necessary. But I may turn to Omar just to comment. As was mentioned earlier, trust is at the heart, I think, of AI’s development in all industries, and that trust has to be gained. In journalism, investing in quality journalism, this is a low-tech thing, but making sure that there is actually a supply of high-quality, reliable news and information is critical before we even get to AI. So, a critical issue for us, yes. I can give an example. An industry is very simple. If you get an article wrong or if you have an IP right, something happens.

Cedrik Neike: If an industry AI makes a mistake, somebody dies, somebody gets injured, we need to be extremely careful. So, especially on the LLM model, I mean the black box approach, we need to be able to actually understand what has happened, how it’s happening. So, the trust aspect is a lot of things we’re working on. So, there’s one about the ethical aspects, and the other one is the replicability and predictability, so that we understand that whatever comes out needs to be extremely correct and right in the right way. So, that’s a lot of where we have more than 1,500 AI engineers, and the focus is really on how to make it trustworthy. Aidan, your company supports businesses. Do you see this increase of need towards solutions to address disinformation?

Aidan Gomez: Yeah, we get that request a lot. I think the very first thing on all C-suites’ minds are risk and what happens if things go wrong as we deploy these systems. And so, that’s a pretty common question. Just generally on the disinformation piece, the thing that I’m nervous about is definitely astroturfing on social media. So, bots are becoming dramatically more compelling. We’re sort of past the Turing test in terms of whether you can distinguish just by reading a post whether it’s human or AI-generated. And so, implementing defensive measures against that, I think, is a crucial priority for these social media platforms. Thank you. Okay, maybe we’ll take your second question, and then the person behind you. Thank you.

Anais Rassat: Can you speak closer to the mic? Okay, I want to ask. So, for my impression, last year, this AI is one of the main topics.

Audience: And then, I really want to know what is the difference of the direction of this discussion in Davos about the AI. So, my impression is that now, this year, the more positive aspect you are now focusing on of this AI, can AI, so because it grows revenue and then maybe the cost efficiency, as you mentioned, the transparency. transformation. So my impression is that you are now focusing on the positive aspect of the GNI. So could you explain the difference between last year and this year?

Anais Rassat: So maybe, yes, all of our panellists discussed the rate of change which is unprecedented. So maybe is there something this year that has, compared to last year, is even more impressive in terms of that rate of change? Cédric?

Cedrik Neike: The main thing is, as I said, AI has been around for a long time until the transformation paper was written. And the one big difference is that the transformer models enable us, we would need three months in a factory to have one AI model sort of deployed that was very specific. And this goes much, much faster in this way. So the rate of adoption is being defined that the learning mechanisms are there. So the one thing is we tackled also the whole issue on how do we make it more trustworthy. So we passed the just, wow, this works, to how is it useful. And we’re seeing, I mean, we have more than 100 customers which are trying to experiment it. And we have so many issues in industries. We don’t have enough workers. We have complexity which is increasing. And AI, and simple to use AI, for me it’s a bit like the internet when you had the web browser coming up. It’s now available to way more people. I think that makes a huge difference. So the difference is it’s now useful. It’s much, much easier. We understand the risks. We know that we train them specifically and not generically. These are huge sort of increases. And that’s why we are embracing it so much because I think it’s going to, as the internet has redefined the world, AI will redefine the world. And we just need to make sure it happens. And the second point, and then I’m going to hand over, it’s also now merging to the real world. My kids don’t only use it to do their essays or whatever they do. The engineers are basically thinking how can I build something which goes better? And we, especially in robotics, the idea of combining automation and robotics now with AI, robots were pretty dumb in the past. I’m sorry for anyone brailling robots, but they’re becoming super intelligent and adaptive and the whole humanoid robot is sort of the next extension. So not only is it more useful, it’s actually reaching out into the real world, solving problems, which we had on not enough labor force, et cetera. And that’s why I think this year is probably more positive than it was last year.

Anais Rassat: Thank you.

Almar Latour: I think the rate of progress is just astonishing. I mean, we’re all experiencing this, but sticking with a simple example that I mentioned before, translation, a year ago, maybe certain models were very good at a certain language. And now that that scoreboard, that leaderboard has changed entirely. And so that’s accelerating us reaching not just millions of people, tens of millions of people, hundreds of millions of people that in the past would have taken decades to achieve, but really never really happening. The same dynamic is happening in audio and the notebook LM that, for example, from Google that launched allows you to absorb different sources of information and then come out with an AI generated podcast with two hosts that interact as if it’s an NPR type podcast. And the same thing is happening in video. And then so every aspect of media, there are major shifts that help scaling, that help the proliferation of more competition, lower the threshold of entry. And so it’s exhilarating. Obviously it comes with some warning labels as well, but it’s mainly exhilarating. Thank you. We have a question at the back.

Audience: Thank you very much. My name is Christoph Kackmann. I’m a reporter with hundreds about German business daily. Here at World Economic Forum, there’s lots of optimism around the progress of AI. You can see all the posters here in the city. But if I look at the studies conducted by a couple of consulting companies, there seems to be only quite a small number of companies that really use the technology at scale. I mean, there are lots of proofs of concept, but there aren’t that many companies that are really able to use it as a core part of the business model. So what do you think, what keeps them back and what has to happen to take these hurdles? So are we under utilizing AI? I’m happy to start. Yeah, I think we’re definitely still in the early phases of adoption. And

Aidan Gomez: a big piece of that are the two points that I spoke about earlier, which is integration with the data sources. It’s not going to be useful to your business if it doesn’t know about your business. And so that’s actually a project and a journey that you need to go along as a company. And enterprises don’t move as fast as consumers. So we’ve seen it enter the consumer world, and pretty much everyone uses it. Well, the younger generation uses it every day. I think part of that is like the skills gap, right, and trying to teach people how to use it. And just to the last point around optimism, I actually think it is increasing familiarity and experience with the technology that’s driving the optimism. It’s less scary, it’s less this unknown thing that people keep talking about, and it’s something right in front of you that you can use. So the more people have gotten close to it, the more they’re excited and they can start to see themselves opportunities for driving productivity or integrating into their life. So yeah, I agree with you. I think we’re still early. Last year, 2024, was definitely the breakout year in terms of production. But this is a ball that’s moving and it’s not going to stop. It’s just going to get faster.

Almar Latour: I think we’ll see many new milestones in years to come. If you draw a parallel, for example, to how telecommunications evolved from 2G to 3G, there were great expectations, but that took some time for adoption. And then when finally the mobile internet landed, yet then it had an upending effect on just about anything that we did in any industry and in our personal lives. And so this is, I think, a development of greater magnitude than that. It’s moving at faster speed and going from 2G to 3G, but we are very much at the beginning. And I think there will be many moments where we look back and say, oh my God, that was the moment. And yet you get surprised almost every day with there’s a new moment. Some closing remarks, Cathy or Cedric, on optimism and AI adoption. I would just compliment what my colleagues are saying by just stressing building the trust within the company, within society is extremely important. And eventually, yeah,

Cathy Li: as generation progress, definitely a lot of the changes will happen that way naturally as well. Collaboration is key. AI is only as good as the data that empowers it. So the more collaboration on data that can be enabled by privacy and security and safety, I think we’ll see more progress in terms of both the development of the technology, but also the integration into society as well.

Anais Rassat: Thank you, Cathy. Cedric? Yeah, I mean, the reality, I’m a glass half full person and my view is that we have many, big problems at the moment. Sustainability and 1.5 degrees is one of them. So we’re running out of time to solve those problems. And the only way for us to cheat time, to be faster, to solve those problems is going to be using AI in the right way. So I think it’s actually a tool at the right time to help us to be competitive, to help us to be more sustainable, to help us to be more inclusive, as long as we take in account how to take society with us along. That actually seems like more of a glass half full. That’s what I was saying. Sorry, if I said I’m half empty, I meant half full. Excellent. Thank you very much to our panellists for this session. Thank you for those attending in the room and those following online. Goodbye. Thank you. Bye. Bye.

Agreement Points

AI’s transformative impact across industries

speakers

– Cathy Li
– Cedrik Neike
– Almar Latour
– Aidan Gomez

arguments

AI transforming multiple sectors with significant economic potential

AI integration in industrial processes improving efficiency and sustainability

AI revolutionizing media and publishing workflows and products

Rapid technological progress enabling new AI capabilities

summary

All speakers agreed that AI is having a significant and transformative impact across various industries, leading to improved efficiency, productivity, and new opportunities.

Need for responsible AI adoption and trust-building

speakers

– Cathy Li
– Cedrik Neike
– Almar Latour

arguments

Need for responsible AI adoption and governance frameworks

Importance of building societal trust in AI technologies

Protecting intellectual property rights in AI training and usage

summary

Multiple speakers emphasized the importance of responsible AI adoption, building trust, and addressing ethical and legal concerns in AI implementation.

Similar Viewpoints

Both speakers view AI as a crucial tool for addressing global challenges and stress the importance of collaboration in advancing AI responsibly.

speakers

– Cathy Li
– Cedrik Neike

arguments

Need for cross-industry collaboration to advance AI responsibly

AI as a critical tool for solving major global challenges

Both speakers express optimism about AI’s potential and highlight the rapid progress in AI capabilities leading to new opportunities.

speakers

– Almar Latour
– Aidan Gomez

arguments

Increasing optimism about AI’s potential as familiarity grows

Rapid technological progress enabling new AI capabilities

Unexpected Consensus

AI’s role in addressing sustainability challenges

speakers

– Cathy Li
– Cedrik Neike

arguments

AI transforming multiple sectors with significant economic potential

AI as a critical tool for solving major global challenges

explanation

While the focus was primarily on economic and productivity gains, there was an unexpected consensus on AI’s potential to address sustainability challenges, highlighting a broader perspective on AI’s societal impact.

Overall Assessment

Summary

The speakers generally agreed on AI’s transformative potential across industries, the need for responsible adoption and trust-building, and the importance of collaboration in advancing AI. There was also a shared optimism about AI’s future capabilities and its role in addressing global challenges.

Consensus level

High level of consensus among speakers, implying a unified vision for AI’s future development and implementation. This consensus suggests a collaborative approach to addressing challenges and leveraging opportunities in AI adoption across various sectors.

Different Viewpoints

Unexpected Differences

Overall Assessment

summary

The speakers demonstrated a high level of agreement on the potential of AI, its challenges, and the need for responsible implementation.

difference_level

Low level of disagreement. The speakers presented complementary perspectives that reinforced the main points about AI adoption, its impacts across industries, and the need for responsible implementation. This alignment suggests a growing consensus on the importance of AI and the key considerations for its development and deployment.

Partial Agreements

Similar Viewpoints

Both speakers view AI as a crucial tool for addressing global challenges and stress the importance of collaboration in advancing AI responsibly.

speakers

– Cathy Li
– Cedrik Neike

arguments

Need for cross-industry collaboration to advance AI responsibly

AI as a critical tool for solving major global challenges

Both speakers express optimism about AI’s potential and highlight the rapid progress in AI capabilities leading to new opportunities.

speakers

– Almar Latour
– Aidan Gomez

arguments

Increasing optimism about AI’s potential as familiarity grows

Rapid technological progress enabling new AI capabilities

Key Takeaways

AI is rapidly transforming multiple industries with significant economic and productivity potential

Key challenges for AI adoption include privacy concerns, need for customization, and building societal trust

Responsible AI governance and ethical frameworks are crucial as adoption increases

Cross-industry collaboration and partnerships are essential to advance AI capabilities and applications

AI is seen as a critical tool for solving major global challenges like sustainability

Resolutions and Action Items

Launch of MINDS program by World Economic Forum to highlight real AI use cases across industries

Dow Jones plans to convert Factiva platform to help with AI rights payments for publishers

Unresolved Issues

How to effectively address AI-generated disinformation and social media manipulation

Balancing AI’s energy consumption with sustainability goals

Closing the skills gap to enable wider enterprise adoption of AI

Suggested Compromises

Commercial collaborations between AI companies and content publishers to recognize value of data/IP

Developing AI models trained on industry-specific data while respecting customer data ownership

The early adopters, particularly of generative AI, have achieved up to 2.4 times greater productivity and cost saving of 13%. The contribution to the global economy is projected to reach between $7.6 and $17.9 trillion by 2038.

speaker

Cathy Li

reason

This comment provides concrete data on the economic impact of AI adoption, highlighting the significant productivity gains and cost savings. It sets the stage for discussing AI’s transformative potential across industries.

impact

This data point shifted the conversation towards the tangible benefits of AI adoption, prompting other panelists to discuss specific use cases and applications in their industries.

We need to make AI an empowerment of the shop floor worker because it’s the same which happened when robots came. Everybody was afraid they’d lose their jobs. Now, the countries with the most robots have the most industrial jobs. The same will happen to AI in the industrial space.

speaker

Cedrik Neike

reason

This insight challenges the common fear of job displacement due to AI by drawing a parallel with robotics adoption. It reframes AI as a tool for worker empowerment rather than replacement.

impact

This comment shifted the discussion towards the human aspect of AI adoption, emphasizing the importance of trust and societal acceptance in successful AI implementation.

Without that value being recognized, you run the risk of really damaging journalism. And so there is a two-pronged answer to that from the industry at this point. One is, at Dow Jones, we like this particular path the most, which is commercial collaboration.

speaker

Almar Latour

reason

This comment highlights the critical issue of intellectual property rights and value recognition in the context of AI and journalism. It introduces the concept of commercial collaboration as a potential solution.

impact

This insight broadened the discussion to include the challenges and opportunities AI presents for content creators and publishers, emphasizing the need for fair partnerships between AI companies and content providers.

We’re starting to see the first production systems of agents start to enter the sphere, and with that is a technological shift. There’s two components that are crucial for agents. One of them is reasoning… And the second piece is learning from experience.

speaker

Aidan Gomez

reason

This comment introduces the concept of AI agents and highlights two crucial technological advancements: reasoning and learning from experience. It provides insight into the future direction of AI development.

impact

This comment shifted the discussion towards more advanced AI capabilities, prompting consideration of how these developments might further transform industries and increase productivity gains.

We have more design data on how things are being built at Siemens than anyone else, but it belongs to our customers. So we need to find something that between you, our customers, etc., we can build it.

speaker

Cedrik Neike

reason

This comment highlights the importance of data ownership and collaboration in AI development, especially in industrial settings. It introduces the complexity of leveraging proprietary data while respecting customer rights.

impact

This insight led to a discussion about the need for partnerships and collaborative approaches in AI development, emphasizing that no single company can solve AI challenges alone.

Overall Assessment

These key comments shaped the discussion by moving it from general observations about AI’s potential to specific challenges and opportunities across different industries. They highlighted the economic impact of AI, the importance of human factors and trust, the need for fair partnerships in content creation, the technological advancements driving AI capabilities, and the crucial role of data and collaboration. The discussion evolved from focusing on AI’s promise to addressing practical considerations for its successful implementation and societal integration.

How can we address the rising electricity demand of AI systems while pursuing sustainability goals?

speaker

Cathy Li

explanation

AI-related energy consumption is projected to grow significantly, potentially conflicting with environmental objectives. Understanding how to balance technological innovation with environmental stewardship is crucial.

How can we build societal trust in AI, particularly in industrial settings?

speaker

Cedrik Neike

explanation

There’s a disconnect between businesses’ enthusiasm for AI and societal fears. Addressing this trust gap is essential for widespread AI adoption and acceptance.

How can smaller publishers protect their intellectual property rights in the age of AI?

speaker

Almar Latour

explanation

While large media companies can negotiate or litigate with AI companies, smaller publishers may struggle to protect their content. Finding solutions for this is important for maintaining diverse media ecosystems.

How can we improve AI models’ access to specialized industrial data while respecting privacy and ownership?

speaker

Cedrik Neike and Aidan Gomez

explanation

Access to specialized data (e.g., CAD, time series) could significantly enhance AI capabilities in industrial settings, but privacy and ownership issues need to be addressed.

How can we develop more effective defensive measures against AI-generated disinformation on social media?

speaker

Aidan Gomez

explanation

As AI-generated content becomes increasingly difficult to distinguish from human-generated content, protecting against widespread disinformation campaigns is crucial.

What strategies can be employed to accelerate AI adoption in enterprises?

speaker

Audience member (Christoph Kackmann)

explanation

Despite the optimism around AI, many companies are still in the early stages of adoption. Understanding and addressing the barriers to widespread implementation is important for realizing AI’s potential benefits.