The Dark Cockpit: What Should Actually Light Up

By Martin Bergmann | AI Project Lab | August 2026

Captain Alexandre Gandini, who flies a Boeing 737 for GOL Airlines, explained in a Lean Enterprise Institute interview what happens when something goes wrong during a flight. Only one amber light, called Master Caution, lights up right in front of him. A panel then shows which system needs attention. For example, if the fuel pump has an issue, a low-pressure light appears in the fuel section, and that’s where he focuses. Nothing else on the flight deck distracts him.

The instinct that feels like diligence

Most PMOs do the opposite. They believe that tracking more metrics, updating them more often, and putting everything on one dashboard gives them more control. This way, senior management can see everything at once.

But that isn’t real oversight. It’s like the old days when pilots faced a wall of analog gauges, and some planes needed a flight engineer just to watch them. When one person tries to watch everything at once, it’s actually harder to spot what really matters.

What a dark cockpit is built to do

Gandini explains this difference with an example of a hydraulic failure. Before EICAS, the crew-alerting system, pilots had to spot hydraulic problems by noticing symptoms like flight-control issues or the autopilot turning off. They would then manually check all the pressure and fluid levels to find the problem. With EICAS, the plane now tells the crew exactly which system failed. The crew can go straight to the checklist and focus on their real decision: whether to keep flying or divert.

This doesn’t mean a 737’s panel is empty. During the flight, airspeed, altitude, and engine readings are always visible. Only the caution system, which is meant to get the pilot’s attention, goes dark when there’s nothing urgent. PMO dashboards usually don’t have this kind of caution layer. Instead, everything can cause an interruption, because every item is treated like a caution light, even if it doesn’t deserve that level of attention.

What’s actually worth a light

A recent CEO interview made a similar point from a different perspective. A portfolio might show mostly green, but that doesn’t really tell a leader anything. Green compared to what? The plan at kickoff? The business case for the budget? What does leadership need to know this quarter? These are all different, and combining them into one status light is the problem. The CEO’s solution was simple: “fewer indicators with stronger meaning.”

Having fewer indicators doesn’t mean having none. A metric should only be included if it answers one of those important questions, not just because it’s green. If a status has been green for three months straight, it’s not answering anything. It’s just a light that’s been on so long that no one remembers what would make it change.

That same interview also highlights the real value of AI: not writing status reports faster, but spotting problems sooner. This is similar to EICAS, but not quite the same. EICAS is certified, so an engineer has already confirmed what a low-pressure reading means before it ever shows up in the cockpit. AI that monitors forty metrics doesn’t have that level of certainty. It can point out what seems off, but it can’t be trusted like a sensor. Treating AI like a sensor just adds more noise. As that same interview put it: “AI can generate an option. It cannot accept responsibility for the consequences.” AI is best used to monitor all forty metrics, so a person can review the ones it flags rather than trying to check all forty manually and missing something important.

Give it a try this week: take a quarter’s worth of status updates for one project, have AI flag every metric that hasn’t changed, and then review the list yourself.

A dark cockpit isn’t an empty one. It’s one in which every light that’s on means something.

---

Martin Bergmann is the Director of an IT PMO and founder of the AI Project Lab. He writes about AI, project management, and the future of intelligent work.