Videos
Governance, Trust, and Innovation: Carmen Marsh on Making Responsible AI Operational
Insights
- The AI race has moved past who owns the fastest model, and now turns on what companies do with those models, how they govern the risks, and whether users trust the output.
- Building the framework meant getting stakeholders across dozens of countries to first agree on what being ethical, trustworthy, and accountable actually means in practice, a consensus that is harder to reach than the technical controls themselves.
- As digital and human workers sit side by side, every employee effectively becomes a manager, and the immediate task is reskilling people to run a hybrid workforce.
Governance almost always trails innovation, and AI has widened that gap fast. Carmen Marsh, Chief Responsible AI Officer and founder of the Global Council for Responsible AI, sat down with Jeff Kavanaugh, Head of the Infosys Knowledge Institute, at the Semafor World Economy Summit in Washington, D.C. Drawing on a career that began securing the early commercial internet, Marsh compares this moment to the dot-com boom: a familiar technology suddenly in everyone's hands, with the risks still poorly understood. Her core argument is that most frameworks fail because they stop at policy, when the harder work is embedding controls that teams can measure, adjust, and tailor by sector as regulation and technology keep moving. She adds that transparency into how those controls perform is what lets governance become a support for innovation, and that building change management in from the start is what keeps a framework from going stale.
Carmen Marsh:
A lot of the frameworks we have out there, they failed being, easy to implement, easy to configure and make it practical when you can measure it because you want to make, as you put your controls in place, you want to have a way to measure the performance of those controls and see is it actually this governance providing the value because one of the things that I always say is we don't want to stifle innovation. That would be the biggest mistake ever.
Jeff Kavanaugh:
So you're actually building a responsible AI framework that does not stifle innovation.
Jeff Kavanaugh:
Welcome to a series of conversations with leaders who are turning AI's promise into real business results. I'm Jeff Kavanaugh here at the Semafor World Economy Summit in Washington, D.C. Today, joined by Carmen Marsh, Chief Responsible AI Officer and Founder of the Global Council for Responsible AI. Carmen, welcome.
Carmen Marsh:
Thank you for having me, Jeff.
Jeff Kavanaugh:
What is the most surprising impact of the AI revolution, and how will it define how companies compete?
Carmen Marsh:
Great question. You know, often what happens is since I go way back when the internet first came to life in Silicon valley. and my background is information security. So we were trying to figure out the best way how to actually go about it securing. Now we've got all these new websites popping up, their information is online. How do we make it secure? And so people ask me, how do you compare that sort of a dot-com boom to what's going on with the AI, so to say, boom today. I think the only similarity is that it is something that's, it's not new, right? We know AI is not new, but it's becoming something that end users are now experiencing. They're using it. But I think for the first time in history, we are building intelligent systems and systems that will have intelligence that's bigger than us. We're not going to be the most intelligent, so to say system, right? We're human... in the world. Very interesting comparison and speaking about the competitive advantage even between what was going on in the dot-coms at that time and what's going on right now is...
Carmen Marsh:
I saw last year was all about who's going to have the fastest model, who is going to build the capabilities. I think this year is more of what are you doing with this models? Are end users trusting you, your product? How are you governing? What are you doing with the risks? Less about the model, more about the outcome. The adoption is more than what it was last year. So every year we see the bigger adoption, more companies and people are using AI.
Jeff Kavanaugh:
Of course you have to have trust to adopt as well and drive it, right?
Carmen Marsh:
Yes, trust is one of those things, especially in this world. We have a geoeconomical changes. There's a lot going on in the world, let's put it that way. And I think trust really becomes a competitive asset, I would say, because of, we need to know what's being put out there.
Jeff Kavanaugh:
And to build trust, you need to be perceived as being responsible. Including with AI.
Carmen Marsh:
Exactly.
Jeff Kavanaugh:
Hence the title and hence your focus. Tell us about this whole council of responsible AI.
Carmen Marsh:
So yeah, initially now, as I mentioned, my background is in cybersecurity, information security.
Jeff Kavanaugh:
Which requires trust.
Carmen Marsh:
Which requires trust. And it's always, say, like if you're in cybersecurity, you're a little bit paranoid all the time. When we started the council, it came from a risk perspective. Now we have this new technology. I call it new, which is not new again. But we have AI in a game. How do we really and that reminds me like I said back in going back to the dot-coms is that was one thing like how do we understand this tech and how do we make sure that we cover the risks?
Jeff Kavanaugh:
Just so people understand this council how many countries are involved?
Carmen Marsh:
We went from US to now we are in 77 countries. I wanted to create a framework that is really built with input from many different countries, different cultures, different backgrounds. I honestly didn't have any big expectations knowing that we had so many different cultures and diverse, stakeholders, right? We were all there. We wanted to have sort of at least a skeleton of the framework at the end of the workshops. But I didn't have too high of the hopes because I've never done this level of conversation and even for us to agree, what does it mean to be ethical? What does it mean to be trustworthy? What does it mean to be accountable?
Jeff Kavanaugh:
So you're laying out the ground rules of being ethical, trustworthy, accountable, all these foundational things that aren't just for countries, they're for companies.
Carmen Marsh:
They're for companies, they're for governments, businesses, academia, really just for users. So our framework is built across industries, across sector, across platforms.
Jeff Kavanaugh:
Did you say you launched it, not the UN, but where was it?
Carmen Marsh:
We were at House of Lords in London.
Jeff Kavanaugh:
House of Lords of London, right.
Carmen Marsh:
Yes, last February with 65 ambassadors and about 13 responsible global Governors.
Jeff Kavanaugh:
So you've got that big picture. How is it now being translated to where especially for like our audience? How are companies now saying great we have this framework. What do we do with it? How do we apply it and get some value from it?
Carmen Marsh:
I think the biggest difference between our framework and some of the frameworks out there is we really want to focus on making governance operational, making governance operational. Because it's not all about policies, right? You can't just put policies in place. You have to as you're building controls for your organization or the government.
Jeff Kavanaugh:
You have to make it practical.
Carmen Marsh:
You have to make it practical, configurable. And that's one of the things that a lot of frameworks we have out there, they failed being, easy to implement, easy to configure and make it practical when you can measure it because you want to make, as you put your controls in place, you want to have a way to measure the performance of those controls and see is it actually this governance providing the value because one of the things that I always say is we don't want to stifle innovation. That would be the biggest mistake ever.
Jeff Kavanaugh:
So you're actually building a responsible AI framework that does not stifle innovation.
Carmen Marsh:
No, that supports innovation.
Jeff Kavanaugh:
Because you're increasing trust so people will adopt it.
Carmen Marsh:
Exactly. And also the transparency, right? Because if you have visibility in what controls you have in place, how they're performing, where the gaps are. Because the more transparency you have, you're going to understand what works, what doesn't work. One of the things I think it was important building into our framework as well is change management. Because as we know, tech, technology is evolving. You cannot have policy base too strict of the controls that can never change.
Jeff Kavanaugh:
Significant uncertainty and rapid change require adaption.
Carmen Marsh:
Yes.
Jeff Kavanaugh:
How has this operating model or the framework, how has it responded to this uncertainty and rapid change?
Carmen Marsh:
Well, that's the thing and we've actually noticed that of course the innovation happens a lot faster than the governance, right? And the governance had been falling behind for the last I would say couple years as it came up to everyone. Oh, you know, we have this what do we do now? We should be responsible, but we don't want to fall behind on this this race that's going on so I think making it the way we build the frameworks. We did not make it to, how can I say, it's built at the higher level. When I mentioned across industry, you make the top layer very specific, but then each industry, like let's say the financial industry or let's take healthcare, which is of course very regulated, right? Then you can build additional governance around that and.
Jeff Kavanaugh:
They can tailor.
Carmen Marsh:
They can tailor it and also the way our framework implements it can actually embed with existing governance systems.
Jeff Kavanaugh:
All right. Talking about the framework.
Carmen Marsh:
Yeah.
Jeff Kavanaugh:
Let's talk about the human aspect.
Carmen Marsh:
Yes.
Jeff Kavanaugh:
Automation, AI have caused leaders to question how they think about talent. What changes are required to optimize this whole human plus AI in your framework and keep people at the center?
Carmen Marsh:
We have to completely rethink the workforce now. I listened to some of the other conversations yesterday on a panel specifically talking about the workforce and where is everything going. People concerned that jobs would be lost. Their jobs, their roles may disappear. Some, but I think it's really just reskilling now. It's... you know, and I have started building the training that is focusing on how do you manage hybrid workforce? Because be realistic. We are going to have a digital workforce and we're going to have a human workforce.
Jeff Kavanaugh:
Everyone's a manager. You have your agents.
Carmen Marsh:
Yes. and so you have to be able to manage both sides, right? And you have to teach your team, the humans, also how to basically now have the workflow and work together in a specific role. So it's all about just rethinking our workforce, rethinking our roles. How do we manage the hybrid workforce? That's what's coming, I think. That's the next big thing.
Jeff Kavanaugh:
Great. Carmen, thank you for your time.
Carmen Marsh:
Thank you for having me.
Jeff Kavanaugh:
You bet. I'm Jeff Kavanaugh from the Infosys Knowledge Institute. Until next time, keep learning and keep sharing.
