#485 Human Agency in a Digital World: Marcus Fontoura in Conversation with Niels Brabandt on AI, Cloud Computing and Strategic Decision-Making
Human Agency in a Digital World: Marcus Fontoura in Conversation with Niels Brabandt on AI, Cloud Computing and Strategic Decision-Making
Artificial intelligence is frequently portrayed as an autonomous force that will redefine industries, displace jobs and render human judgement obsolete. In this in-depth conversation, Marcus Fontoura joins Niels Brabandt to offer a more nuanced and strategically grounded perspective. The central question is not whether AI will transform business, but what role human agency will play in shaping that transformation.
Marcus Fontoura, Microsoft Technical Fellow and author of Human Agency in a Digital World, argues that we are at a pivotal moment in technological history. Automation has improved efficiency for decades, yet the current wave of artificial intelligence has intensified concerns about displacement and redundancy. However, as Marcus Fontoura explains, history suggests a different trajectory.
Using the analogy of chess, he recalls how IBM’s victory over Garry Kasparov in 1997 was widely perceived as the end of human relevance in the game. Instead, chess has grown in popularity and complexity. Human and machine collaboration has elevated standards of play. The lesson for business leaders is clear. The most powerful outcomes arise not from replacement, but from augmentation.
Niels Brabandt presses the question that concerns many executives. Should professionals in administrative or highly procedural roles fear automation? Marcus Fontoura responds with clarity. Roles dominated by repetitive and mechanical tasks are indeed susceptible to disruption. Yet as automation reduces manual workload, it simultaneously elevates the importance of analytical thinking, contextual reasoning and strategic judgement. Jobs will evolve. They will not simply disappear.
Cloud computing underpins this transformation. As Marcus Fontoura explains, the infrastructure enabling advanced AI systems depends on vast cloud architectures and digitised information at global scale. Without cloud computing, today’s large models would be impossible. Yet this power comes with consequences. Data centres currently account for approximately two percent of energy consumption in the United States, with projections suggesting significant growth. The sustainability challenge is real. Whether innovation in model efficiency can offset rising demand remains an open strategic question.
The discussion moves beyond infrastructure to governance and decision-making. Marcus Fontoura introduces the distinction between 40/60 decisions and 10/90 decisions. Some choices are reversible and incremental. Others are transformative and difficult to undo. Leaders must develop the discipline to distinguish between these categories. Investment in a new hardware ecosystem, for example, is rarely a marginal adjustment. It is a structural commitment.
For decision-makers in business, this distinction is critical in the age of AI. Investments in data architecture, cloud dependency or proprietary AI capabilities may represent 10/90 decisions with long-term strategic consequences. The ability to assess reversibility and systemic impact becomes a defining leadership competence.
Talent strategy forms another core theme. As production functions evolve and automation accelerates, organisations must prioritise adaptability, growth mindset and resilience. According to Marcus Fontoura, the leaders of tomorrow will not simply manage technology. They will cultivate teams capable of continuous learning and cognitive flexibility.
The interview concludes with a discussion of quantum computing. Contrary to popular belief, quantum computers are not merely faster machines. They are fundamentally different systems capable of addressing specific classes of complex problems. For executives, this is not a call for immediate deployment, but a reminder that technological literacy must extend beyond headlines.
Throughout the conversation, Niels Brabandt ensures that technical insights are translated into strategic implications. The result is a substantive examination of how leaders can preserve and strengthen human agency in a digital world.
In an era dominated by automation narratives, Marcus Fontoura and Niels Brabandt present a more demanding vision. Technology will continue to evolve. The decisive factor will be whether leaders harness it responsibly, strategically and with a clear commitment to human judgement at the centre.
Niels Brabandt
---
More on this topic in this week's videocast and podcast with Niels Brabandt: Videocast / Apple Podcasts / Spotify
For the videocast’s and podcast’s transcript, read below this article.
Is excellent leadership important to you?
Let's have a chat: NB@NB-Networks.com
Contact: Niels Brabandt on LinkedIn
Website: www.NB-Networks.biz
Niels Brabandt is an expert in sustainable leadership with more than 20 years of experience in practice and science.
Niels Brabandt: Professional Training, Speaking, Coaching, Consulting, Mentoring, Project & Interim Management. Event host, MC, Moderator.
Podcast and Videocast Transcript
Niels Brabandt
AI is everywhere. AI is omnipresent. The question is, which role does the human play here? And we have an expert on the matter with us here today. Hello and welcome, Markus von Thura.
Marcus Fontoura
Hello, Nils. Pleasure to be here.
Niels Brabandt
Thank you very much for taking the time. We'll get straight into the interview. You wrote the book which talks about human agency in a digital world. Most people at the moment think, "Okay, is this really about humans?" because everything we see at the moment is about how to make AI more efficient, how to make it more effective, how to maybe automatize things, or maybe even replace human beings, maybe not within their job, but maybe when they retire, not get someone new on board.
Niels Brabandt
So what is your take on human agency in a digital world in the times we live today? Because you're a Microsoft Technical Fellow, so you know from the technical side what you're talking about. What's your take on that?
Marcus Fontoura
Well, I, I think we're in an unprecedented pivotal moment now, like, because we've seen, like, technology and automation improve, like, wha our efficiency over time. And, and we've always been into into this dilemma that, like, will technology, like, eliminate our jobs or not? And probably this is true since like the early dawn of technology.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
Now I think it's probably accelerated, that feeling, with, like, some of the news that we see da-daily and some of the talk about AI. However, my personal opinion is that, like, we still need humans, and we still very much need to engage more humans in solving, like, important problems for society. And, and I think this is more relevant today than ever.
Marcus Fontoura
Like, we, we have, like, lots of still open problems that we'd like to solve, like climate change, distrib healthcare, distribution. Computers cannot do that by themselves. They cannot really solve these problems by themselves.
Marcus Fontoura
So, and Nils, if you take a parallel to chess, right, so the chess happen like, it's a field that, like, AI disruption happened like 30 years ago, like in 1997, IBM beat Kasparov, that was the reigning world champion. And you think, like, "Oh, then we don't have room for humans in chess anymore." But that's not true. Like, chess grew in popularity, and today you have more people playing chess than ever.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
And, and we can see that in chess, like, when you have humans and computers together, we can achieve more, and we can do more. The level of playing, like really increases. So that's what we're gonna see in the other areas with AI now.
Niels Brabandt
Hmm. Excellent. And in your book, you also talk about communication from extrasensory perception in information cascade. However, one of the chapters is also called "Do Not Study." And I saw on your CV you have a PhD in computer science, so you definitely did study. What do you mean with that? What do you mean with "Do Not Study"?
Marcus Fontoura
Well, I was just saying that, like, when you talk about with computers, language precision is very important. So, like, that was an anecdote. When I was in grad school we had a coffee shop, like, in my university building, and the, the sign there was "Do Not Do Not Study." And my advisor that was in formal methods, he was always saying, like, "That sign is wrong. It should be spelled 'Do Not Study Here' because, like, this is giving the wrong signal to students."
Niels Brabandt
Yeah.
Marcus Fontoura
But, but that is a very important anecdote because you and I have the context, right? So we know that "Do Not Study" in that context means "Do Not Study Here," but you still, as a student, are supposed to study.
Marcus Fontoura
However, when you, you have computers, the it's very hard for computers to understand this whole context. Although, like, AI models are very smart, they're AI model is equivalent to a computer program that read for 30,000 years, eight, eighty thou eight, eight hours a day nonstop and memorized everything, right? So we know a lot of things, but we still don't have, like, the human context that we have. And and it's very hard to capture that all that context. So that, that's, that's what I try to convey in the book.
Niels Brabandt
Hmm. Excellent. So when you now hear people who say, "I am very afraid that AI is probably going to replace my job because I work in general admin," would you say they need to be scared of that, or do you think it's something where no one needs to be frightened?
Marcus Fontoura
Well, I, I definitely think that AI is going to raise the, the bar, right? So we've seen, like, technology in general raise the bar, and we'll see new jobs being created by, like, for instance, like, with the example of chess that I gave before. Now we have the popularization of chess coaches online, chess platforms. There is, like there are influencers talking about chess in chess channels. So there are new jobs that were not present in 1997 when IBM beat Kasparov.
Marcus Fontoura
So I, I think today, like, we'll see that, like, AI will probably automate away some jobs that are very mechanical while, like, some other jobs will, will be created. And, and, and I think, like, we have a pretty good understanding of what, what are those jobs, right? If you are in a profession that all you do is, like, mechanical tasks that can be solved by a computer, then you are, you, you, you pretty much can consider yourself very, very susceptible for, for disruption.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
However, if you use your brains and you really need to analyze things and have a deeper understanding, understand the context, as we were saying before, and compare and contrast, like, option A with option B, then it's much harder to automate that job, right?
Marcus Fontoura
So I think what we'll see is that the jobs will become more interesting because people will have more free time to, to think, to, to devote to these analytical skills and less time needed in these manual tasks that we.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
That we see today.
Niels Brabandt
Excellent. Which role do you think cloud computing is going to play here?
Marcus Fontoura
I think cloud computing is already playing a, a massive role, and people don't notice that because it's kind of like under the covers, right? But, like, all these AI models run on cloud systems, like and they are only possible due to the vast amounts of digitized information that we have stored in the cloud in the clouds and in the cloud systems that we have across the world. So without the advent of cloud computing and the internet, like, we wouldn't have, like, this level of computational power that we have today and the amount of information that we have today to, to train these very large models that, that.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
That we see.
Niels Brabandt
Excellent. When we talk about all this AI and especially cloud computing, how do you think we're going to deal with the amount of energy we need to run this?
Marcus Fontoura
Yeah. If you look at the US today, about 2% of the, the energy, the power is for data centers. If you continue the trend, it will be about 9% by the end of the decade. So, like, it's kind of a five, five-fold increase, which is.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
Huge. And this is very much driven by th-this idea that, like, we, we need bigger and bigger models, right? As we said before, right, like "Do Not Study" is very different than "Do Not Study Here." If you want to, to capture all these details and all these exceptions, you need more and more parameters. You need to read for longer, and you need to store more details. And those details are encoded in, in parameters in the model. So if you keep hearing that GPT-3 has that many parameters, GPT-4 has a, a, a larger number of parameters, this means that we capture more detail about the world.
Marcus Fontoura
And the current bets, if you don't have any, any new innovation between now and the end of the decade, is that, like, to, to make progress, we need bigger models. And but, but I think, like, it's up in the air because there is research going on. And, and, and if you find if you find really find, like, better ways to represent knowledge, better ways of representing common sense and representing that you know "Do Not Study Here" is different than "Do Not Study," that we know that birds can fly, but then we it's very hard to capture all the exceptions, right? For instance, baby birds cannot fly, penguins cannot fly.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
These are easy, but, like, then birds involved in oil cannot fly either, right? So if you think about the amount of exceptions, we need bigger and bigger models. Unless we come up with a new way of representing knowledge that is more efficient, like, the trend for more power-hungry data centers will keep pushing us that direction.
Niels Brabandt
Hmm. Excellent. And in your book, you also talk about the 40/60 and 10/90 decisions. What do you mean by that?
Marcus Fontoura
Yeah. So in a company, when we are making decisions about which technology to invest, which systems to build, who to hire, and so on I claim that there are two types of decisions. One that are that are more inconsequential, like, I call this 60/40 or 40/60, like, me-mean-meaning that if you if you go the wrong way, like, you are, like you could be 60% more efficient, but then you're just 40% more efficient. And, and we have, like, lots of those decisions that we make, but and sometimes we over-index on those.
Marcus Fontoura
And I say that there are other types of decisions that are 90/10, and those are the ones that really, if you go one route, like, it's really hard to reverse, right? Like, for instance, like one classical example is, like, if Amazon decided to invest in, in, in Kindle, like, then it has to build, like, a whole new diver division of of hardware, building hardware. And that's, like, probably a very costly and hard to undo that decision.
Marcus Fontoura
So I was framing in the book that, like, when we are making decisions about the future of technology and especially, like what the impact of technology in society, we need to analyze, is this a reversible decision, like 40/60, or this is a decision that we really need to be mindful because we, we are in the 10/90 scenario.
Niels Brabandt
Excellent. So another cha another chapter is also about one thing, which is probably now crucial, how to attract and keep the right people.
Niels Brabandt
So how, in your opinion, especially as you, you work for a world-famous employer and with a PhD in computer science, you know how the competition is right now, especially around really co really competent talent, how do I know which are the right people to not only attract but also to keep them in my organization?
Marcus Fontoura
Yeah. And this is changing so much in New Zealand as we're seeing, like, the production function changing and all that we talked so far with, like more automation of the manual tasks. And we really need people now that, like, are willing to work hard, are willing to cope with change, right? They have this growth mindset because, like, probably the world in one year will be very different than the world today. So this adaptability, the, the will to work hard, the will to embrace change, and the, the will to to want to build a better future is, like, what we are looking for when we are hiring people.
Niels Brabandt
Excellent. And to wrap this interview up, and then I have one more bonus question after that what do you think the role of quantum computing is going to be? Because most people say, "Yeah, there are these quantum computers, and they are very, very fast," and that's usually what people know about that.
Marcus Fontoura
Yeah. And that's almost like an urban legend because quantum computers are not necessarily faster, right? They are in, in, in reality, they are slower. If you count clock speed, you know how fast a computer can process an instruction, quantum computers would be typically much slower if, um however, quantum computing is very different than classical computing because the types of operations that we can do in quantum computing are different.
Marcus Fontoura
Like in a because, because of quantum computers, we have this in quantum mechanics, we have superposition, right, so that either in compute in classical computers, you have a bit can be either zero or one. And but in quantum computers, you have a probability of being zero, probability of being one. So then actually what we have then is a vector of probabilities.
Niels Brabandt
Mm-hmm.
Marcus Fontoura
And because of this, we can represent represent a lot more states. So quantum computers are very, very good in representing problems that we need an exponential number of states. And, like, and these types of problems are are normally, like, what we call NP-hard problems or complexity in, in terms of complexity. They are hard to solve in classical computers, but they can more easily be solved in, in quantum computers.
Marcus Fontoura
And a lots of problems in this space include, like distribution, and then include quantum mechanics itself, includes biology, and so on. So quantum computers will not be a faster computer, but they'll be a computer that is very good in solving some, some very specific set of problems that are very important for us.
Niels Brabandt
Yeah. Excellent. And when now people say, "Hey, I think Marcus can be really of help for us, either as a speaker or a consultant or in any other role," how can people actually get in touch with you?
Marcus Fontoura
We have my website, fontura.org. I'm also very active in, in LinkedIn. And so th-these are the two primary forms.
Niels Brabandt
Perfect. I think these are the perfect final words. Marcus Fontura, thank you very much for your time.
Marcus Fontoura
Thanks, Neil. It was a pleasure being here. Thank you.