The Fastest Cars Have the Best Brakes: Why “Speed vs. Oversight” Is a False Dichotomy in AI

Years ago, my husband and I rented a powerful little car in France. I found myself gripping the dashboard on a winding coastal road, all sharp curves and dramatic drop-offs. The sun was bright, the sea flashed blue in my peripheral vision, and behind me an impatient local driver was so close I could practically feel his front bumper nudging me forward.

There was curve after curve, with barely a shoulder to spare and I spent every moment being thankful for what the rental office had told me as they handed over the keys: “We just put in new brake pads and rotors. Enjoy yourself!” 

Every time we rounded a tight corner, I was very, very glad that the car had excellent brakes.

When you’re moving fast in a high-stakes environment, control is what makes speed possible.That stressful, beautiful, memorable drive is what comes to mind when leaders tell me that meaningful AI oversight will slow innovation down.

I understand the concern. Nobody wants to bury a promising initiative under layers of unnecessary process. But I think the framing is off. “Speed vs. oversight” is a false dichotomy. In practice, trust is a competitive advantage. Accountability architecture is infrastructure for speed.

Organizations that treat oversight like a drag on progress may move quickly at first. But often, they’re not really moving fast. They’re moving exposed.

Why This Framing Matters

When leaders talk about AI oversight as if it naturally gets in the way of innovation, the assumption is usually that there are only two options: move fast or govern carefully.

But that’s not how durable institutions work.

The organizations that earn lasting confidence—from boards, regulators, employees, customers, and investors—are usually the ones that communicate risk clearly and early, before it escalates. That’s a theme I’ve returned to often in my work: Credibility is built before the crisis, not during it. The same principle applies here.

If your organization can’t explain 

• where a model is being used

• who owns its output

• what decisions it influences

• when humans step in

• how concerns get escalated

  
  

then your speed is fragile. It may look impressive for a while but it probably won’t hold under pressure.

By contrast, companies that build traceability and accountability into their AI systems are often able to move faster where it matters most: in regulated, high-stakes, and reputation-sensitive environments where confidence is a prerequisite for action.

Trust reduces friction

One reason I push back on this tradeoff is simple: trust reduces friction.

When ownership is clear, review processes are defined, and escalation paths are understood, decisions move faster. Legal is less likely to become the department of “no.” Boards are less likely to be blindsided. Teams are less likely to burn time debating governance only after something has already gone wrong.

That matters as AI adoption accelerates across nearly every function. According to McKinsey’s State of AI, organizations continue to scale AI use, but many are still maturing their governance practices unevenly. That gap creates hidden drag. Rapid deployment without accountability often leads to the very delays leaders were trying to avoid: rework, internal distrust, reputational damage, regulatory scrutiny, and expensive cleanup.

A simple rule of thumb applies here: The cost of proactive oversight is almost always lower than the cost of a preventable failure.

We’ve Already Seen The Price Of Weak

Oversight

Regrettably, there’s no shortage of examples of companies allowing their AI to operate with “weak brakes.” 

In 2024, Air Canada was held responsible for inaccurate information provided by its chatbot. A customer relied on the bot’s answer about bereavement fares, and the airline later argued that the chatbot was effectively a separate entity. That argument didn’t hold. If an AI-enabled system is speaking for your company, it is your company in the eyes of the customer.

We saw a different kind of failure when Google paused Gemini’s image generation feature after public backlash over inaccurate historical outputs. Yes, there were technical issues involved. But there were also leadership and governance questions underneath them. How was risk being evaluated? Who had the authority to challenge assumptions or stop the rollout? What should’ve been surfaced earlier?

And now we’ve seen a similar pattern with Meta. Just days after rolling out a new feature that let users create AI-generated or altered images from public Instagram content, the company pulled the feature following backlash over privacy and consent concerns. Users had effectively been opted in by default, which meant people’s likenesses could be used without their knowledge or permission. Meta admitted it had “missed the mark.”

These are reminders that when organizations move faster than their accountability architecture, they often end up doing exactly what they were trying to avoid: stopping abruptly, absorbing criticism, and rebuilding trust after the fact. Which slows things down and results in less trust when they do finally relaunch. None of these are arguments against innovation; they’re an argument against expensive, preventable mistakes.

Accountability Isn’t Bureaucracy

This is one of the points I spoke about earlier this year at The Digital Economist’s event, We The People: Reclaiming Accountability in the Age of Intelligent Systems. Too often, AI accountability gets framed as a technical issue that should stay with technical teams. I don’t see it that way; it’s a leadership issue.

Speaking at The Digital Economist's event in DC this spring.

And leadership accountability doesn’t mean creating endless layers of approvals for their own sake. It means getting clear—early!—on a handful of essential questions:

• What is this system being used for?

• What level of risk does that use case create?

• Who’s accountable for the outcome?

• What happens if something goes wrong?

• How are decisions documented and communicated?

  
  

In fact, organizations that skip these questions in the name of speed often end up creating more delay for themselves later. They generate confusion, spark conflict, and lose time trying to figure out who owns what after the fact. If everyone’s accountable, no one’s accountable.

The Leaders Who Move Fastest Are The Ones Who Prepare

The strongest executive teams understand that trust is a performance variable.

It affects how quickly initiatives get approved, how candidly risks are surfaced, how confidently people act, and how resilient the organization is when something doesn’t go according to plan. It also affects talent. Strong leaders want to work in organizations where innovation is ambitious and responsible, where the system is mature enough to support bold moves without asking people to ignore obvious risks.

That’s especially true in healthcare, financial services, legal, government, and other high-consequence sectors. In those environments, oversight is part of execution. The leaders who earn lasting institutional confidence are the ones who build the mechanisms that make real control possible.

A Better Question For Leaders

So instead of asking, “How much oversight can we afford before we slow down?” I’d ask What accountability architecture do we need in order to move quickly, credibly, and repeatedly?

The goal isn’t to eliminate risk and no serious leader believes that’s possible. The goal is to create enough clarity, ownership, and traceability that your organization can innovate without constantly increasing the odds of a preventable failure.

When I think back on that drive up the French coast, I certainly remember the sparkling  water, the wind in my hair, the curve of the road hugging the cliffs. I also remember the feeling of knowing that when the next turn came, that zippy car, my husband, and I could handle it.

That’s what good oversight does. It doesn’t take away your momentum, it gives you the confidence to keep going. That, to me, is what mature leadership looks like in the age of AI. The fastest cars have the best brakes and the most trusted organizations do too.

If your team is thinking through how to build trust, accountability, and governance into AI adoption without losing momentum, I’d be glad to continue the conversation through a keynote, advisory engagement, or leadership session.