Twenty years ago, you Googled it.
Ten blue links. You scanned them, discarded most, opened two, cross-checked one against the other, and formed a view. It took ninety seconds and required a small amount of judgment. Not much. But some.
Now you ask, and it answers. Not links... an answer. Complete, confident, formatted, cited if you want citations. The distance between not knowing and sounding like you know has collapsed to about four seconds.
Let me be clear before I go further. I use these tools every day. They are extraordinary. The value is not theoretical and it is not overstated. When I need the history of a regulation, or a first pass at a framework, or the name of the thing I am circling but cannot retrieve, the speed is genuinely delightful. Deliriously so, some days.
But something got compressed, and it was not the search.
It was the interval. The gap between the question forming and the answer arriving. That interval used to be unavoidable, and inside it, quietly, we did things. We noticed what we already believed. We felt the shape of what we did not know. We occasionally discovered we were asking the wrong question entirely.
That interval is now optional. And we have voted.
I spend most of my week in rooms with senior leaders, and the most valuable thing that happens in a coaching session is usually the silence after a question. Not the answer. The silence. Ten seconds, twenty, sometimes a full minute where a very accomplished person sits with something and does not reach for the nearest plausible response.
Most of them find it excruciating. Watch what people do with that discomfort and you learn everything. They fill it. They answer a smaller, adjacent question they can already answer. They hand it back: "well, what do you think I should do?" Answer-seeking is very often anxiety management wearing a business suit.
Here is what this work has taught me. The question a person brings is almost never the question. It is a proxy. "Should I restructure the team" is frequently "have I lost confidence in myself as a leader." "How do I deliver this feedback" is frequently "what am I afraid will happen if I tell the truth." The first question is real. It is just standing in front of the second one.
The only way through to the second question is to stay in the first one longer than is comfortable.
An answer engine will never do that for you. It will answer the question you asked, beautifully, instantly, and it will never once ask you why you asked it. That is not a flaw in the technology. That is the technology working precisely as designed.
Meanwhile the economics reward the same compression. We hire people who have answers. We promote the fastest credible response in the room. Meetings are graded on decisions per hour. Slack rewards the reply, not the reflection. An executive who says "I need to sit with that until Thursday" runs a real risk of being read as slow rather than rigorous, and everyone knows it, which is exactly why almost nobody says it.
So we have built a social and economic system that pays for output, and we have now handed that system a tool that produces infinite output. The open question is what happens to the capacity underneath.
Which brings me to the analogy I cannot stop turning over.
GLP-1 drugs work. That is not in dispute. People lose significant weight, and for many of them the health outcomes are real and life-changing. I have no interest in the moralizing that has attached itself to this.
But there is a well-documented catch. Lose weight on these drugs without resistance training and adequate protein, and a meaningful share of what you lose is lean mass. Muscle. Bone density. The scale says success. The mirror mostly agrees. And underneath, the structure is quieter and weaker than it was, and you will not find out until the day you need it.
That is healthy-ish. It is the appearance of the outcome without the capacity that normally produces it.
Cognitive fitness works the same way. Your ability to hold ambiguity. To sit in not-knowing without flinching. To notice when a question is a decoy. To build an argument from the studs rather than assembling one from parts already cut to length. That is lean mass. It shows up on no dashboard. It atrophies silently. And it is precisely the capacity you need on the day the situation is genuinely novel and there is no answer to retrieve, because nobody has been here before.
The people who use GLP-1s well still lift. That is the entire point. The drug removed a barrier. It did not remove the training.
So lift.
Pick the questions that actually matter, maybe two a week, and refuse to look them up for twenty-four hours. Write your own bad answer first, then check it, and pay close attention to the gap between the two, because the gap is the lesson. When you catch yourself reaching for the tool, ask what you are reaching away from. And when someone brings you a question, hold your answer for sixty seconds and ask what is standing behind it.
None of this is anti-technology. I would not give these tools up, and neither would you.
The tools will keep getting better. That is not in question, and it is not the risk.
The risk is quieter. Output and capacity feel identical from the inside, right up until the moment they do not. There is no soreness. There is no scale. Nothing flags a thinking muscle you quietly stopped using two years ago. You find out on the day the situation is genuinely new, when there is nothing to retrieve because nobody has been here before, and you reach for something that is no longer there.
Healthy-ish holds until it is tested.
So keep the tools. Use them hard. And then go find the questions worth staying inside of, and stay there longer than is comfortable, on purpose, for no visible return.
That is the training. Nothing has replaced it.