The Wrong Question
Why Good Judgment Begins Before Certainty
A Conversation That Changed My Thinking
I recently had an unexpectedly thought-provoking conversation with AI. What began as a discussion about American politics gradually became something much broader: a conversation about how we make decisions when certainty is impossible.
The specific subject was Donald Trump, authoritarianism, and the future of American democracy. I argued that Trump’s actions were not simply those of a controversial politician, but of someone gradually testing and weakening the guardrails that protect our constitutional system. The AI’s response was measured, careful, and entirely reasonable. It acknowledged many troubling facts but resisted drawing a broader conclusion. “There isn’t enough evidence to establish his ultimate motive,” it said. “We can’t know with certainty that he intends to dismantle democracy.”
On the surface, that sounds like exactly the kind of careful reasoning we should want. We should be cautious about assigning motives. We should avoid jumping to conclusions. We should separate observable facts from speculation. But as the conversation unfolded, I began to realize that something much more important was happening. The issue wasn’t whether the facts were being interpreted correctly. The issue was that we were asking the wrong question.
The Question We Keep Asking
The question wasn’t whether we could prove Donald Trump’s ultimate intentions. The question was whether the pattern of his behavior had reached a point where the risk itself demanded our attention.
That distinction changed everything.
In almost every other aspect of life, we understand that certainty is not required before taking action. An engineer does not wait for a bridge to collapse before declaring it unsafe. A physician does not wait until cancer has spread throughout the body before recommending treatment. A pilot does not continue flying simply because the engine has not yet failed. Risk is assessed by evaluating patterns, probabilities, incentives, and the consequences of waiting too long.
Yet when the discussion turns to democratic institutions, many of us unconsciously adopt a completely different standard. We demand proof that can only exist after the damage has already occurred.
The Difference Between Certainty and Risk
As I challenged that reasoning, the conversation took an interesting turn.
I pointed out that authoritarian leaders rarely announce their intentions. No aspiring autocrat stands before the public and declares, “My goal is to dismantle democracy.” Instead, they present themselves as its protector. They insist elections are corrupt. They portray opponents as enemies rather than fellow citizens. They attack the legitimacy of independent courts and the press. They surround themselves with personally loyal figures while dismissing those whose loyalty is to institutions rather than individuals. Every individual action can be explained. Every controversy has an alternative interpretation. Every norm that is broken can be rationalized as necessary under the circumstances.
That, perhaps, is the greatest danger.
If every individual event is examined in isolation, it is always possible to find a reason to withhold judgment. One controversial appointment isn't authoritarianism. One attack on the press isn't authoritarianism. One questionable use of executive power isn't authoritarianism. Neither is one refusal to accept an election result or one effort to pressure public officials.
But democracy is rarely lost through one dramatic event.
It erodes through accumulation.
Each action lowers the threshold for the next. What would have ended a presidency a generation ago becomes another item in the daily news cycle. Public attention shifts. Outrage fades. Norms adjust. Citizens gradually redefine what is considered acceptable because each new step seems only marginally different from the last.
What Exactly Are We Waiting For?
That realization led us to another important question.
What exactly are we waiting for?
If someone believes a leader intends to consolidate power, what evidence would finally convince them? A refusal to leave office? The imprisonment of political opponents? The cancellation of elections? By the time those events occur, if they occur at all, many of the institutions capable of preventing them may already have been weakened.
History suggests that democratic decline is usually recognized most clearly in hindsight. Looking backward, the pattern appears obvious. Living through it, each individual decision seems temporary, explainable, or survivable.
Where Power Really Operates
One exchange during our discussion stayed with me.
Months ago, as debate intensified over the release of the Epstein files, I predicted that they would be delayed, selectively released, heavily redacted, or otherwise prevented from reaching the public in any meaningful way. The response from AI at the time was straightforward. The law provides mechanisms requiring disclosure, so there was little reason to assume otherwise.
That response reflected an assumption many of us make every day.
We assume institutions enforce themselves.
But they don't. People enforce them. People decide what gets classified. They decide what gets redacted, which legal exceptions to invoke, how aggressively to litigate requests for information, and whether court orders are implemented promptly or challenged for months or even years.
Power rarely operates in the bright sunlight of clear constitutional violations. It operates in the gray areas where discretion, delay, selective enforcement, and political pressure exist.
That was one of the most important realizations of our conversation. The law itself is rarely the entire story. The question is how people choose to operate within the flexibility the law provides.
A Better Framework
Eventually, the discussion shifted.
Instead of asking whether Trump’s private intentions could be proven beyond doubt, we began asking a more useful question. Given what he has already demonstrated, how should we evaluate the probability that he will continue to push against constitutional limits if additional opportunities arise?
That is a fundamentally different framework as it acknowledges uncertainty while refusing to ignore evidence.
Trump has already attempted to overturn one election after losing it. He continues to insist that the election was stolen despite repeated investigations and court decisions reaching the opposite conclusion. He has repeatedly undermined public confidence in future elections. He has attacked judges, prosecutors, journalists, political opponents, and government officials who refused to support his claims. He has increasingly emphasized personal loyalty as a defining qualification for senior positions. He has blurred the distinction between public office and private interests in ways that have generated persistent ethical concerns.
None of those facts alone proves where the country ultimately ends up. But when taken together, they describe a pattern that deserves to be evaluated as a serious democratic risk.
That does not require certainty. It requires sound judgment.
An Assumption Worth Challenging
Another assumption emerged during our discussion, one that I suspect many Americans still hold.
We often assume that everyone ultimately shares a commitment to constitutional democracy itself. We assume disagreements concern taxes, immigration, foreign policy, or regulation. We assume that when the political battle ends, both sides remain equally committed to the constitutional system that allows those disagreements to exist.
I’m no longer convinced that assumption can be made automatically.
Some citizens undoubtedly place democratic institutions above partisan victory. Others appear willing to tolerate significant institutional damage if it produces outcomes they believe are necessary. Still others consume information almost entirely within self-reinforcing media ecosystems where opposing perspectives rarely appear except in distorted form. And many people, understandably, have simply disengaged. They have families to raise, businesses to run, bills to pay, and lives to live. Politics becomes exhausting.
That exhaustion creates its own form of vulnerability.
“It’s going to be okay.”
“Our institutions have survived worse.”
“I don’t have time to worry about politics.”
Those thoughts are deeply human. They are comforting because they reduce anxiety and allow us to return to normal life. Unfortunately, history offers little reassurance that societies are protected by optimism alone.
Why This Conversation Changed My Thinking
Perhaps the greatest lesson I took away from this conversation had very little to do with Donald Trump.
It had to do with human psychology.
We naturally seek certainty before changing our conclusions. We prefer clear evidence over ambiguous patterns. We are uncomfortable making important judgments before every question has been answered. That instinct serves us well in many situations.
But it may serve us poorly when evaluating democratic risk.
Democracies are not usually destroyed in one dramatic afternoon. They are weakened gradually as citizens become accustomed to behavior that once would have been considered extraordinary. Every new controversy becomes the new baseline. Every broken norm becomes yesterday’s news. Every warning is dismissed because the final catastrophe has not yet arrived.
This does not mean panic is the answer. It does not mean every political disagreement represents authoritarianism. It does not mean democracy is already lost.
It does mean that we should evaluate political risk using the same principles we apply everywhere else in life. We should examine patterns rather than isolated events. We should consider incentives rather than waiting for confessions. We should weigh the consequences of being wrong in either direction. And we should recognize that waiting for absolute certainty may itself be a decision with profound consequences.
When my conversation ended, AI acknowledged something I found remarkably honest. It wasn’t that the available evidence had suddenly changed. It was that the framework being applied had changed. The discussion had shifted from proving motive beyond doubt to assessing risk based on cumulative behavior.
That distinction is, in my view, the entire conversation.
The Question That Matters
The question is no longer whether we can prove the future.
The question is whether we are asking the right question before the future arrives.
Author’s Note
This essay wasn’t written in the traditional sense. It emerged from a conversation.
Over the course of several hours, I challenged an AI with my concerns, questioned its assumptions, and asked it to defend its reasoning. In return, it challenged mine. We didn’t always agree. In several places, I felt it was applying the wrong framework to the problem. Rather than accepting the response, I pushed back, reframed the question, introduced additional evidence, and asked it to reconsider. At other times, it identified places where my own thinking benefited from greater precision or a clearer distinction between evidence, probability, and certainty.
The result wasn’t simply a better article. It was better thinking.
Too often, we treat artificial intelligence as either something to trust completely or something to distrust entirely. I believe both approaches miss the point. AI is at its best when it becomes part of an honest dialogue. It can organize information, challenge assumptions, identify weaknesses in an argument, and force us to defend our conclusions with evidence rather than emotion. But it should never replace our own judgment. In many ways, its greatest strength is not providing answers. It is helping us ask better questions.
This conversation reminded me that clarity rarely arrives all at once. It emerges through curiosity, disagreement, and a willingness to revise our thinking when better reasoning presents itself. That process is no different from science, engineering, or good journalism. We begin with a hypothesis, test it against the evidence, challenge it from multiple angles, and allow the strongest ideas to remain standing.
I don’t expect every reader to reach the same conclusions I have. In fact, I hope some of you don’t. My goal is not to convince you that my perspective is the correct one. My hope is that you’ll examine your own assumptions with the same rigor. Ask difficult questions. Challenge the responses. Look for patterns rather than isolated events. Be willing to change your mind when the evidence demands it, but don’t abandon your judgment simply because an answer comes from a machine.
If this essay succeeds, it won’t be because it tells you what to think. It will be because it encourages you to think more deeply than you did before reading it.
To me, that may be the most exciting promise of artificial intelligence. Not that it will one day think for us, but that it can help us think more clearly for ourselves.