Culture of Experimentation

Mark Silis, Former CTO at MIT and Managing Partner @ Silis LLC

AI has captured the attention of business leaders everywhere, but according to former MIT CTO Mark Silis, technology alone is not what drives transformation. The organizations pulling ahead are those willing to experiment, take calculated risks, and lead through uncertainty. Drawing from decades of experience guiding large-scale innovation efforts, Silis argues that successful transformation requires more than a strategy—it requires a culture that embraces change, learns from failure, and keeps people at the center of every decision.

Are institutions moving too slowly or too quickly with AI? Or are they focused on the wrong things? I think it depends on the institution. They are all made up of many different types of institutions, each with its own culture and perspective. The institution I came from has a highly technical mindset and a strong willingness to embrace technology. Another place might view it very differently.

Right now, I think everyone is trying to answer a fundamental question: What is AI? People say they want to use it, take advantage of it, and build strategies around it. But if you ask leaders what AI actually means to them, you'll get an amazing range of answers.

A lot of people don't really know what it is yet, and I think that's actually an acceptable answer. Anyone who tells you they have the definitive AI strategy or know exactly how to fully AI-enable an institution is kidding themselves. Nobody has done this before. We're in completely uncharted territory.

When I was working in startups after MIT, we were helping companies become internet-enabled. Back then, nobody had all the answers either. Today feels very similar. Everyone is trying to figure out what this technology means and how it fits into their environment. We're all part of a giant experiment.

Can you speak to the balance of experimentation versus governance?

Experimentation is the right approach. One of the fastest ways to kill innovation is to declare an AI mandate too early. If you want to stifle experimentation, make it bigger than it should be before people understand it. The institutions seeing the most success are the ones willing to take chances, test ideas, and learn from the process.

At the same time, governance is important. But it's difficult to govern something you don't fully understand.

Look at the contrast between Europe and North America. Europe has taken a more regulatory approach to AI, while North America has generally allowed more room for experimentation. I'm not suggesting either approach is right or wrong—just that they reflect different philosophies.

The evolution of the internet 25 years ago followed a similar path. The institutions making meaningful progress today aren't operating without rules, nor are they being reckless. They simply understand that some degree of risk-taking is necessary because nobody knows exactly where this is going. Mistakes are inevitable, and that's okay. Mistakes are part of learning.

The institutions moving forward are the ones willing to try things, even if some projects don't succeed. A project that doesn't work isn't necessarily a failure—it provides information that makes the next effort better.

When calculators first appeared, there were serious debates about whether we should stop teaching the slide rule. People worried that students would lose fundamental engineering principles. There were passionate arguments on both sides, and ultimately education adapted.

We're seeing a similar debate today. Do we still need to teach everything we've always taught? Maybe we do. Maybe we don't. That's a conversation that will unfold over the next decade or two.

Beyond the classroom, AI will impact the operational side of business as well. Large institutions are supported by thousands of people performing countless tasks that keep everything running. AI will inevitably change many of those functions too.

When you look at the entire landscape, I still believe it's early. But it's also an exciting time.

What Is the biggest misconception leaders have about AI right now?

It's that leaders believe they can introduce a disruptive technology like AI in a way that keeps everyone happy, avoids discomfort, and doesn't rock the boat.

The reality is that technology is inherently disruptive. If you're in a senior technology leadership role, part of your job is managing disruption. That's not easy. Yet, success is often measured by whether nobody got upset, nobody felt uncomfortable, and everyone remained happy throughout the process.

I don't think that's realistic.

It's almost impossible to navigate a transformation of this magnitude without creating some level of discomfort. That's part of the process. Unfortunately, there are always people who show up with a neatly packaged plan and promise that they can deliver everything an institution wants while keeping everyone satisfied. Whether it's a consulting firm, a technology provider, or an outside advisor, the message is often the same: Follow this roadmap and everything will work perfectly.

That's simply not true.

What is your advice in the short term?

Organizations—especially in the public sector—have followed that path repeatedly, and it rarely delivers the results they expect. The reason is simple: Meaningful change requires disruption, and disruption is uncomfortable by nature.

We're seeing that dynamic play out with AI right now. Some leaders are looking for someone to hand them a strategy they can implement with minimal risk. They think, "If it works, great. If it doesn't, I'll blame the consultant and move on."

But that's not leadership.

If you want AI to have a meaningful impact on your institution, you have to own the responsibility yourself. You have to be willing to lead through uncertainty, make difficult decisions, and accept that not everyone will be comfortable with the changes.

Too often, institutions point to outside experts and say, "We're doing this because they told us to." That's not leadership. Leadership means taking responsibility for the direction you're choosing and helping your organization navigate the challenges that come with it.

The biggest misconception about AI isn't the technology itself. It's the belief that you can implement something this transformative without disruption, discomfort, and strong leadership.