Knowledge is not experience
I see this pattern constantly.
Someone, otherwise perfectly competent, placed under pressure, suddenly performs at a level commensurate with a junior.
Their knowledge hasn't magically disappeared, just their ability to apply that knowledge effectively has deteriorated under pressure.
Why? Because knowledge is not experience.
You could have read every book on a topic, memorized every rule, and understood every best practice, yet still find yourself unable to make a sound decision when confronted with real constraints.
Knowing the rules is one thing. Knowing which rules apply, which can be ignored, and what the consequences are for getting those things confused is a whole other bag of chips.
This is the reality of working with living, breathing systems. They're complicated, unpredictable, and rarely fail in convenient or obvious ways. More often than not, they’re hostile to your mental well-being by default.
This is why experience matters. How effectively you apply your knowledge is heavily influenced by your ability to handle the situation in front of you. The more stressed, overwhelmed, or frightened you become, the harder it is to recall what you know, evaluate your options, and make reasonable decisions.
Air France Flight 447
Consider Air France Flight 447, which crashed into the Atlantic Ocean in 2009, killing all 228 people aboard.
After ice obstructed the aircraft's airspeed sensors, the autopilot disconnected and the crew was forced to respond to unreliable instrument readings. The pilot flying repeatedly commanded the aircraft's nose upward, even as it entered an aerodynamic stall. Instead of recovering, the aircraft remained stalled and descended into the ocean.
These were trained professionals who understood the principles of flight. Yet confronted with unfamiliar conditions, confusing feedback, and escalating pressure, they failed to recognize the situation and apply the appropriate recovery procedures.
At 02:11:10 UTC, the aircraft had climbed to its maximum altitude of around 38,000 feet. There, its angle of attack was 16 degrees, and the engine thrust levers were in the fully forward Takeoff/Go-around detent (TOGA), and at 02:11:15 UTC the pitch attitude was slightly over 16 degrees and falling, but the angle of attack rapidly increased towards 30 degrees. A second consequence of the reconfiguration into alternate law was that "stall protection" no longer operated. Whereas in normal law, the airplane's flight management computers would have acted to prevent such a high angle of attack, in alternate law this did not happen. (Indeed, the switch into alternate law occurred precisely because the computers, denied reliable speed data, were no longer able to provide such protection – nor many of the other functions expected of normal law). The wings lost lift and the aircraft stalled.
Subsequent investigations identified problems involving training, cockpit instrumentation, and the crew's interpretation of what was happening. Their knowledge of aviation wasn't enough to overcome their inability to understand the specific circumstances they were facing.
This applies everywhere. A driver who knows how to correct a skid can still instinctively overcorrect when their vehicle loses traction. A physician who understands the clinical application of a diagnosis can still make mistakes when overwhelmed by competing symptoms and incomplete information. An engineer who understands a system intimately can still make a catastrophic mistake when rushed into changing something without first considering its dependencies.
In each case, the problem isn't necessarily a lack of knowledge. It's a failure to recognize the situation, evaluate the available information, and apply that knowledge appropriately under pressure. Experience doesn't make someone immune to these failures, but it does provide a foundation for recognizing them before they become catastrophic.
This is why experience, understanding, and seniority matter.
Getting a quick answer from Google, Stack Overflow, or an LLM is not the same as understanding why that answer is correct (or may not be).
More importantly, having that unearned knowledge doesn't mean you can distinguish between an answer that's correct, one that's partially correct, and one that's complete nonsense in the context of the problem you're trying to solve.
Whenever I'm tasked with solving something genuinely critical, not merely perceived as critical, my first step is to silence everyone and everything. I create what I think of as a quarantined room in my head, free from external input and unnecessary distractions. Then I start digging.
Have I seen this problem before? If so, what was the cause and how was it resolved? Does that same reasoning apply here, or am I dealing with something novel? What assumptions am I making? If I change this part of the system, what else might be affected? And perhaps most importantly, what do I actually know versus what am I merely assuming to be true?
This is just the preliminary work. A real solution takes time, patience, and a willingness to admit when you don't yet understand the problem. There is no magic to it, even with an LLM at your disposal.
The point isn't to find an answer as quickly as possible, but rather, to develop enough understanding to make a decision without creating additional problems in the process.
The 'AI will figure it out' trap
Lately, anecdotally, I've noticed an increasing tendency to blame AI for bad decisions, unnecessary accidents, and broken software.
Certainly, these tools can produce convincing nonsense, and their tendency to express unwarranted confidence is a legitimate problem. But I think we're overlooking something far more fundamental: people are increasingly confusing the speed at which they can produce something with their ability to understand and manage it.
It's impressive to watch an LLM confidently rip through a codebase, implement a feature, and declare everything tested, safe, and ready for release.
That impression disappears pretty quickly, though, when the result turns out to be a broken pile of shit that someone now has to manually debug. The tool may have generated the mistake, but the decision to trust its output without sufficient understanding or verification was 100% pure, old-fashioned, homegrown human.
The funny part (in a dark sense) is that this situation isn't unique to AI. We've been making these mistakes for decades.
Rushed deployments, careless assumptions, poorly understood dependencies, and people making decisions they aren't equipped to make are hardly new phenomena. AI has simply made it possible to produce and compound these mistakes at a much greater speed.
What concerns me now is how readily we're treating this acceleration as progress.
We're removing the time traditionally spent thinking, questioning, and understanding, then acting surprised when the quality of our decisions deteriorates. Worse, we're introducing more information, more interruptions, and more pressure into environments where people are already struggling to think clearly.
We desperately need to pump the brakes (and for the initiates lacing up their matching Nike’s, no, I don’t mean pump the brakes on research—just application).
Experience teaches you when to move quickly, when to slow down, and when to stop entirely until you understand what you're dealing with.
Knowledge gives you the information necessary to understand a problem. Experience gives you the context to recognize how that information applies under real conditions. Competence is your demonstrated ability to put both to use, make sound decisions, and achieve the intended result.
Speed without understanding is a liability. Possessing one doesn't automatically grant you the others, and confusing them is how otherwise capable people end up making avoidable mistakes.
Ultimately, the measure of someone's ability isn't how much they know, how long they've been doing something, or how quickly they can produce an answer.
It's whether they can consistently exercise sound judgment when the circumstances demand it, using not just knowledge, but experience combined with knowledge to form a calm, considered response to any situation; no matter how benign or critical.