🔍 Jordi Ribas, Microsoft Vice President: "Generative AI has started the era of copilots"
💡 In an exclusive interview, reflects on generative AI, European regulation, his relationship with Mustafa Suleyman and Satya Nadella, and Microsoft's role in leading global digital transformation.
Born in Manresa (Barcelona) 55 years ago, Jordi Ribas is one of the most prominent Catalan and Spanish figures in Artificial Intelligence. He is the Vice President of Microsoft and oversees its search engine and artificial intelligence. Ribas leads Bing’s global teams in engineering, product, and growth. His team launched Copilot, driving innovation and competition in search and AI to challenge Google. He is the author of over 50 technical and research publications and holds 20 patents. Currently, he reports directly to Microsoft AI CEO Mustafa Suleyman and maintains close connections with corporate CEO Satya Nadella. Suleyman, Microsoft’s star recruit in AI, directly supervises Ribas’ projects, including Bing, Copilot, Edge, MSN, AI models, and other products. Ribas continues to lead Bing and spearheaded the launch of the first Copilot version. With multiple Copilots now in Microsoft, his team continues contributing to Copilot while focusing primarily on Bing.
You’ve been in the United States for three decades. Are you eager to return home, or is this precisely the most exciting moment of your professional career?
Yes, I’ve been in the United States for nearly 35 years. I love returning to our country for many reasons, especially to see my childhood and university friends and one of my international teams based in Barcelona, which I try to visit whenever I can. However, my main team is in the U.S., and my close family—including my mother—lives with me in the States.
How do you perceive Europe, with so much AI regulation? Do you think it’s hindering innovation, or is it a reasonable position?
AI regulation is very important. Like any technology, it can be used for positive or negative purposes, and legislation can help reduce risks. For instance, facial recognition has helped reunite thousands of lost children with their parents in a project we worked on with an NGO in China, which is fantastic. However, the same technology can also be used for population surveillance, so it’s essential to have laws ensuring privacy rights.
Currently, AI regulation in Europe is intense and demands additional expenses and efforts. The key is effectiveness. The intent is good, but companies sometimes question whether it’s worth launching certain AI advancements in the EU. I believe the U.S. currently strikes a better balance between regulation and innovation.
With Donald Trump’s return, how do you envision the next few years in AI development in the U.S.? Do you think deregulation will dominate to compete fiercely with China, or will a conservative approach to AI prevail?
I think it’s hard to tell, as many factors will come into play, such as Elon Musk, who is currently a close advisor to Donald Trump. Musk has invested in and developed significant AI technology within his companies, and last year, he founded xAI to innovate and advance AI rapidly. In any case, I hope there will continue to be a good balance between regulation and innovation.
"It’s hard to predict how Trump’s return will impact AI. Factors like Elon Musk, a close advisor to Trump, could play a role. Musk aims to innovate and advance AI rapidly. I hope we maintain a good balance between regulation and innovation."
Microsoft has a center in Barcelona. With Spain and Catalonia betting on AI strategies, chips, and the Barcelona Supercomputing Center, can you see yourself working in Barcelona? Would that feel like playing in a lower league?
Barcelona is well-positioned in the public and private sectors with the Supercomputing Center and AI R&D hubs like ours. I established the first AI team at our R&D hub, and we now have over 200 Microsoft AI engineers. Their work is on the same level as what we do in the U.S. or in my teams in India and China.
At present, with my team distributed globally with thousands of members, it’s more practical for me to lead from the U.S. On a personal note, my wife is American, and my children live in Los Angeles and Chicago. That said, I don’t rule it out in the future. In fact, my wife is in love with our country and has already learned Catalan and Spanish.
If consulted by European, Spanish, and Catalan leaders, what would you recommend to ensure Europe and its countries not only stay on track but also compete globally?
I would recommend striving for a balance between regulation and AI innovation—more like in the U.S.—and investing in AI for the long term. Attracting foreign company investments like Microsoft, where Spaniards and Catalans can perform top-level R&D, is also crucial. I’m confident that in the future, some of these professionals will create their own major AI companies in our country. When Microsoft launched Microsoft Research in China, there were no companies like Alibaba, Baidu, or Tencent. What people often don’t know is that many of their executives were trained at Microsoft Research.
"Europe must strike a balance between AI regulation and innovation, similar to the U.S., and invest long-term. Attracting foreign investments like Microsoft is essential. I’m confident Spaniards and Catalans will create leading AI companies in the future."
Do you think Mistral AI is a one-off case, or should it become more common? How do you assess the relationship between Microsoft and Mistral AI? Is it an example to follow?
Mistral is an example to follow in terms of innovation. The AI models they’ve developed are highly competitive and are one of the options we offer on our Microsoft Azure cloud. We also use Mistral in some Bing components. What I’m less certain about is their business model if they continue with “open source,” as the investments required for the future of AI are enormous.
What is your relationship with Mustafa Suleyman? And with Satya Nadella?
Mustafa co-founded DeepMind and spent a few years at Google. Now he is my direct supervisor and the leader and CEO of Microsoft AI. He reports to Satya Nadella, the CEO of the entire corporation. I work closely with both of them, especially with Mustafa, and our relationship is very good.
If you had to name five key figures in AI on a global level, who would they be?
I’d have to include the pioneers of Deep Learning for their fundamental contributions to AI: Geoffrey Hinton, Yoshua Bengio, and Yann LeCun. I still remember how many traditional AI researchers opposed them. The three of them strongly believed in their methods and proved they were right. They received the Turing Award in 2018, which is equivalent to the Nobel Prize in Computing, and Hinton received the Nobel Prize in Physics this year.
At the same time, today’s generative AI wouldn’t have progressed as much without leaders who had the courage to invest on a large scale. For this reason, I would include our CEO Satya Nadella, who had the vision to invest in OpenAI when many doubted it was worth it. That investment sparked the revolution, first with ChatGPT, and later with our Bing Chat or Copilot with GPT-4. Other chatbots and models from other companies came later.
There are many other researchers, like Andrew Ng, or business leaders like Sam Altman, who are making excellent contributions to AI. But for El Punt Avui readers, I’d highlight Oriol Vinyals, who is an exceptional AI figure and also Catalan.
What profiles will lead AI in the coming years?
I think there will continue to be leaders with business and research profiles, and there will also be leaders in advanced AI product development, which is where I aim to contribute.
If I may simplify, are you more of a techno-optimist or a technological altruist?
I might lean more toward a technological altruist. My motivation when I wake up every morning is to try to improve the world with innovative products.
In Catalonia, there is concern about the future of the language. Is AI an opportunity or a threat for minority languages like Catalan?
I hope it’s an opportunity. From the start, GPT-4 has already worked in Catalan, and today chatbots function in many languages and can easily translate between them. Soon, we’ll be able to speak Catalan with people who speak any other language.
"I hope AI is an opportunity for languages. From the start, GPT-4 worked in Catalan, and today chatbots function in many languages and can translate seamlessly. Soon, we’ll speak Catalan with anyone, regardless of their language."
Your work and that of your teams have been pivotal in positioning Microsoft in the AI race. At what point, or with which project or decision, do you think Microsoft’s big AI move outpaced its competitors?
Microsoft has been a pioneer in AI for many years. In the 1990s, our founder Bill Gates already said, “Computers will be able to see, hear, and talk. They will recognize writing and gestures. They will be capable of learning and extracting information without a programmer’s help. They will understand us.” He then created Microsoft Research, which has always focused heavily on AI.
Our experience and knowledge in AI over so many years were crucial for Satya to have the vision and courage to invest billions of dollars in OpenAI, which led to ChatGPT and the revolution of generative AI on our Microsoft Azure cloud.
Personally, I had the privilege of leading Bing Chat, which we later called Copilot, and it was the first Microsoft product to integrate GPT-4. This started what Satya Nadella announced as the “Copilot era” at our company, which then made other companies realize the importance of LLMs and invest more in generative AI.
"In the 1990s, Bill Gates said, ‘Computers will be able to see, hear, and talk. They will understand us.’ His vision, combined with our deep AI expertise, enabled Satya Nadella to invest billions in OpenAI, sparking the generative AI revolution."
Search and AI, your responsibility, seems fascinating. Do you think search engines are on their way out due to the rise of chatbots, or are they mutually reinforcing? Bing has grown with Copilot and now with ChatGPT Search. How should we imagine the near future of information searches? And what about search advertising, whether in traditional engines or chatbots?
Search engines are evolving rapidly. As I mentioned, we were the first to integrate a chatbot into Bing in February 2023. We initially called it Bing Chat and later Copilot because we realized that, with GPT-4, search engines could understand much more natural and complex queries while producing more comprehensive answers without users having to visit multiple links. This was when the industry and the media saw that search engines would change profoundly, as highlighted by the Wall Street Journal article, “I Tried Microsoft’s New AI-Powered Bing. Search Will Never Be the Same.” This trend in search will continue.
What many may not realize is that new search engines still need traditional search engines. For instance, ChatGPT Search and Meta AI chatbots use Bing in a process called RAG (Retrieval-Augmented Generation), where they combine Bing results with LLMs like GPTs and Llama to generate responses. Without the traditional search engine as the foundation, LLMs are not up-to-date. In fact, the generative AI and LLM revolution makes traditional search engines even more critical.
Regarding advertising, we’ve integrated some search engine ads into our chatbots, but we believe it will also evolve and change. For instance, it will be logical and natural for brands to have their own chatbots.
Can Bing’s growth in the United States continue? What will it depend on? And in Europe?
In the U.S., Bing and search engines based on Bing, like DuckDuckGo, Yahoo, and more recently, ChatGPT Search, now generate nearly 30% of searches on PCs, according to the independent panel comScore. When I started, our market share in comScore was only about 8%. So, we’ve grown significantly, but there’s still much to do, especially on mobile. Bing is also growing in Europe, particularly on PCs.
"In the U.S., Bing and engines like DuckDuckGo or ChatGPT Search generate nearly 30% of PC searches. When I started, our share was only 8%. There’s still room to grow, especially on mobile, and Bing is gaining traction in Europe too."
To help users understand: Is the big bet on Copilot before Bing? Is it compatible with the alliance with OpenAI and ChatGPT? Isn’t there some confusion or redundancy?
Bing and Copilot are two complementary ways to search for information on the web. If it’s a direct search for a link, like “El Punt Avui,” traditional Bing search is faster and more practical. If it’s a more complex query requiring research or interaction, such as “I want an itinerary to visit New York in 5 days with my family. We have two kids under 10,” then it’s better to use Copilot. That’s why we’ve integrated Copilot into Bing since February last year.
Regarding OpenAI, our alliance is very positive for both companies. ChatGPT is a client of our Azure cloud, and OpenAI’s LLM models are available on Azure for all companies that want to use our cloud. This way, we’ve democratized access to the most advanced AI models. Another example is ChatGPT Plus, which uses Bing as a source for internet results.
On the other hand, Microsoft and OpenAI compete in some products. Collaboration and competition between companies are very common today, which is what we call “coopetition” in English.
What developments, innovations, or launches are you currently working on?
At Microsoft, we don’t discuss unannounced projects, but I can say that we continue innovating in all AI fields. One interesting area that all companies are currently working on is AI agents, which are a more autonomous version of chatbots that could improve many processes.
The major media outlets highlight the negative aspects of AI, and public perception tends to fear it. What would you say to people to help them relax or trust in the future?
AI has been with us for decades, and like any technology, although it has risks, its benefits for society are immense. We are living in an exciting moment. AI is already transforming every field, such as engineering, medicine, education, and industries, and now, with generative AI, this transformation will be even deeper. At Microsoft, we work alongside other companies, universities, and governments to maximize the benefits and minimize the challenges. I am optimistic about AI’s future.
"AI has been with us for decades. While it has risks, its benefits for society are immense. With generative AI, we’re seeing even deeper transformations in engineering, medicine, education, and more."
Your position gives you a privileged view to tell us: Are we closer to AGI or to the bursting of the 'AI bubble,' as recently warned by the European Central Bank?
There are many opinions on when AGI will arrive. People even debate its definition, but these often become almost philosophical discussions. I try to stay practical and focus on developing useful products with AI. Although we are seeing many advancements, I believe there’s still much work to do, and I don’t see any product that comes close to the traditional definition of AGI.
As for whether there’s an AI bubble, in my opinion, not in the long term. Everything is relative, though. I remember when people talked about the internet bubble in the late 1990s, and yes, some companies were overvalued at the time. However, if we look at where we are now, 25 years later, we see enormous growth in the industry thanks to the internet and a huge impact on society. I’m convinced the same will happen with AI.
Despite the robustness and efficiency of Microsoft’s tools, media perception seems to place OpenAI as the leader in AI with ChatGPT. What does Microsoft need to take the lead? What’s the strategy?
ChatGPT had a great advantage as the first tool of its kind to launch in the market. While it’s true we compete with ChatGPT with Copilot, as I mentioned earlier, we are OpenAI’s primary investor and a significant collaborator. If both ChatGPT and Copilot continue growing as they have so far, Microsoft benefits.
"ChatGPT had a first-mover advantage in this market. While we compete with ChatGPT through Copilot, as OpenAI’s largest investor and collaborator, Microsoft benefits as both grow."
The public has discovered generative AI, and every day there are exciting innovations in image, video, and audio generation. Multimodality is the big leap of 2024 for everyday users. However, Microsoft also offers AI solutions for companies and organizations that aren’t as well-known. Can you explain what Copilot and the advances of your teams mean for businesses?
Microsoft is primarily a B2B company, and approximately 90% of our revenue comes from businesses. Our mission is “to empower every person and every organization on the planet to achieve more,” which essentially means we only grow as a company when we help others grow. For this reason, we have a specific Copilot for businesses using our services, called Microsoft 365 Copilot, which can be tailored to each company. We also have a platform on Azure so that each company can develop its own Copilots. It’s all part of our strategy to democratize AI.
Are you concerned about energy consumption to keep developing AI? Will it be sustainable environmentally?
On our end, energy consumption will be sustainable. Microsoft is a company deeply focused on the environment, and we continue developing technologies to improve it through initiatives like “AI for Good.” Additionally, four years ago, Microsoft committed to being carbon negative, water positive, zero waste, and protecting more land than we use by 2030. In fact, just last week, we announced a new design for our data centers called “zero-water data centers,” which recycle all water without consuming it.
Are you concerned about the lack of data to train AI models? Will the internet become more essential or less so as progress continues?
At some point, we will reach a limit with data. However, there’s still plenty of room ahead, and there’s significant research into using data generated by LLMs themselves, which is showing substantial improvements.
"At some point, we’ll reach a data limit. However, there’s still room to grow, and significant research is exploring LLM-generated data, showing promising improvements."
How will AI impact the labor market? Which professions have a future?
There will be jobs that are lost, especially monotonous ones that AI can perform. This is inevitable, as we’ve seen throughout history whenever disruptive technologies emerge. However, so far, what we’re seeing is improved productivity across all fields, which historically has generated more and better jobs.
Is the idea of universal basic income reasonable as AI becomes more prevalent?
Perhaps this is a question better suited for economists.
Is personalization necessary to keep attracting AI users, or does it scare more than it appeals?
In some applications, personalization brings benefits to users, though I don’t think it’s always necessary. Currently, chatbot personalization is quite limited, and I hope that in the future, the best products will offer users control settings so they can decide the level of personalization they want.
The recent debate about X (Twitter) and disinformation attributed to AI—is it a temporary issue? How do you imagine the internet and social networks of the future?
Disinformation has always been with us, and there are ways to reduce it. Phenomena like deep fakes do worry me, as they appear real and can be created very easily with generative AI. It’s important to have provenance technologies so we can identify whether content was created by AI. For example, in our Bing Image Creation tool, we’ve implemented a provenance standard and other technologies to determine whether content is real or created by our AI. In general, there should also be legislation to ensure that products adopt AI responsibly.
Do you think the name Artificial Intelligence is appropriate? In Catalonia, Antoni Esteve advocates for the term Augmented Human Intelligence. How does that sound to you?
I like Antoni Esteve’s concept. The term AI is too often associated with science fiction movies, which can confuse and scare people. Ultimately, AI-powered products are merely tools under our control to improve productivity and enhance human intelligence.
Thank you very much for your time, which is incredibly valuable.
Thank you. It’s been a pleasure to share these reflections.
[This article from Transparent Algorithm #64 is also translated into Spanish, Catalan, French, and Italian.]