Can You Trust What You See?
Updated: 3 days ago
If animals could secretly understand everything humans say, which one would make you MOST nervous?
🐶 Your dog
😼 Your cat
🦜 A parrot
🐦 Every pigeon on the street
When something seems unbelievable, how do we work out what’s really going on?
Imagine someone shows you something extraordinary. A person appears to read minds. An animal seems to understand human language. A video captures something nobody can explain. An AI seems to know information it was never taught. You might immediately think, “That’s impossible,” or you might decide, “I saw it happen, so it must be true.” Neither reaction tells us very much. A better question is: how could we test what’s really happening?
That question sits at the heart of science, detective work and critical thinking. When something surprising happens, good thinkers don’t automatically believe it, but they don’t automatically dismiss it either. They become curious. They separate what they actually observed from the explanation they have given it.
What happened — and why did it happen?
Suppose you watch a magician correctly guess a card someone picked from a deck. You really did see the magician name the right card. That part is real. What you don’t yet know is how they did it. Maybe there was a trick involving the cards. Maybe someone in the audience helped. Maybe the magician noticed something you didn’t. The observation and the explanation are two different things.
We make this mistake in everyday life too. Imagine a student suddenly starts getting much higher marks. We can observe that their results improved, but why? Maybe they studied more effectively, started sleeping better, learnt how to manage their time, received tutoring, or simply sat an easier test. A result can be completely real while our explanation for it is wrong.
One of the smartest ways to investigate a mystery is to change something and see what happens. Imagine your friend claims they can tell which of three cups contains lemonade without tasting them. Instead of arguing about whether they can do it, you could test the claim. Could they still identify the lemonade if the cups were covered? What if someone else poured the drinks? What if nobody in the room knew which cup contained it? Each change helps eliminate another possible explanation.
This is essentially what scientists do when they design experiments. Instead of only asking, “Can I prove my idea is right?”, they also ask, “What test could show me that my idea is wrong?” That can be a much more powerful question.
Your brain likes quick explanations
The tricky part is that our brains are very good at forming explanations quickly. If your friend walks past you without saying hello, you might immediately think, “They’re angry with me.” Maybe they are, but perhaps they didn’t see you, were distracted, were having a bad day, or thought you were ignoring them first. The first explanation that appears in your head can feel convincing simply because it appeared first.
This matters even more online. You might see a ten-second video that seems to prove something shocking. Someone has added a confident caption, thousands of people have liked it, and the comments are full of people saying, “I knew it!” But you still might be missing important information. What happened before the clip started? What happened afterwards? Who uploaded it? Has anything been edited out? Is the caption describing what happened, or simply telling you how to interpret it?
The internet gives us enormous amounts of information, but not always enough context to understand that information properly. That is why critical thinking is not about refusing to believe things. It is about slowing down long enough to ask whether there might be another explanation.
And then there’s AI
Artificial intelligence creates another version of the same problem. Imagine an AI system answers 100 questions and gets every one correct. That sounds impressive, but researchers would still want to know how it reached those answers. What happens if the questions are phrased differently? What if irrelevant information is added? What happens when the system encounters something unlike anything it has seen before?
Sometimes a system can arrive at the right answer by using an unexpected shortcut. That means the final result alone does not always tell us whether the system understands what we think it understands. One of the most important questions in AI research is therefore surprisingly simple: “How do we know it learnt what we think it learnt?”
The same habit is useful in everyday life. When something surprises you, ask yourself: What exactly did I observe? What different explanations could fit what I observed? What could I change or test to tell those explanations apart? You can use this way of thinking when you see a strange video, hear a rumour, get an unexpected result, or encounter a claim that sounds almost too incredible to believe.
And sometimes, the explanation you eventually uncover is even more interesting than the mystery itself.
There’s a horse involved…
Episode 3 of FYP: For You Podcast begins with a horse named Clever Hans and a claim that once fascinated huge crowds.
Was Hans really capable of what people thought he could do?
I’m not spoiling that here.
Listen to The Horse That Could Do Maths to find out.
Comments