🎧 Listen to this article (5:03)
Two people are sitting in a pub arguing about which football team is better. One supports Arsenal, the other supports Liverpool, and before long they both try to prove the other is wrong.
They can argue about trophies, goals, league positions and players because those things can be measured. But if the real argument is simply that one person prefers Arsenal and the other prefers Liverpool, there is nothing to settle.
The same applies to films, music, food, art and almost everything else people develop strong personal tastes around. I can say I love a film and you can say you hate it, and neither of us has to be wrong.
We can still have a useful conversation about why we feel differently. We might disagree about the acting, the pacing, the ending or the characters, and that can be interesting because it helps us understand how another person experienced the same thing.
The problem begins when one person decides there must be a winner.
You can tell me that an album won awards, sold millions of copies and influenced an entire generation of musicians. I can still say I do not enjoy listening to it.
Some questions have answers
Objective questions are different because they do not depend on preference. Gravity does not care what I think, and if I drop a glass I cannot save it by insisting that I personally disagree with the laws of physics.
Nobody argues with gravity because the answer arrives straight away. Most facts are not like that. The answer is often somewhere else, so people argue instead of checking.
I can remember plenty of pub conversations where two people would confidently argue about a historical event, a football statistic, a law or a scientific fact even though neither of them really knew the answer. One person vaguely remembered hearing one thing, while the other remembered something different, and somehow that uncertainty became a competition.
At that point, the sensible response is not to argue harder. It is to admit that neither person knows, then check.
If there is a fact of the matter, evidence should settle it. You do not need the louder voice or the greater confidence. You need better information.
Two different mistakes
The first mistake is treating a subjective preference as though there must be an objective winner. Someone says your taste in music is wrong, or that you are wrong to like a particular film, even though they are really arguing against your personal response.
The second is treating an objective question as though confidence is evidence. Someone insists they are right because they remember something, heard it years ago, or feel certain.
If it is subjective, you do not need to win. If it is objective, you need to check.
Some disagreements should end with different preferences. Others should end with better evidence.
Facts can exist inside opinions
Subjective opinions can still contain objective claims. I might say I hated a film because it was four hours long, but if the film was actually ninety minutes long, my dislike remains real, but the reason I gave contains a factual error.
What happened, what it means and how we feel about it are three different questions, even though people often mix them. Once you separate them, the conversation becomes much clearer.
Facts can be checked. Interpretations can be discussed. Preferences can be different.
When AI knows more than you
AI is getting better at the facts. That makes it tempting to hand it the interpretations and preferences too.
Imagine an AI that is logically and scientifically correct almost all of the time. It rarely hallucinates, understands enormous amounts of information and can compare evidence faster than any person.
It would be exceptionally good at telling you what is likely to happen. It could tell you which job will probably pay more, which investment carries less risk or which treatment has the strongest evidence behind it.
Those are questions where facts, probabilities and evidence matter. But then comes a different kind of question: what matters to you?
The lower-paying job may give you more freedom. Maybe you care more about family than status. Maybe every critic tells you a film is terrible, but you still love it.
An AI may be very good at predicting what you will prefer. That is still information, not a decision.
This is the difference between asking AI to help you choose and asking it to choose for you. The first uses its intelligence to improve your judgement. The second assumes that a question of preference has a correct answer and that the machine has it.
Being right about reality does not give AI the right to decide what matters.
Some things are checked. Some things are chosen.
Some questions have answers whether we like them or not. Other questions depend on what we care about.
AI can help with the checking. The choosing is yours.
Nobody argues with gravity, but you can still love a bad film.
