Seven myths holding you back

In every lecture I give, the same questions come back to me in slightly different words, and after some years I ended up with a list of seven. Not one of them is entirely wrong, and that is exactly what makes them so hard to shift: every myth carries half a truth inside it, and the other half is the part that holds you back.

1"It will take my job"

The half that is true: some tasks really will be automated. The half that holds you back: a job is not a single task. What actually gets displaced is a person performing one repetitive task that calls for no judgement, no relationship and no responsibility. The remedy is not to resist the tool, but to rise above it.

2"It understands what I say"

It does not understand in the sense you mean by that word. What it builds is a mathematical representation of the relations between words, and then it produces the most probable one. The result resembles understanding closely enough, until it gives itself away at the first question that needs lived experience rather than text.

3"It is neutral because it is a machine"

The machine is neutral; the data it learned from is not. It inherited whatever bias those texts carried, and then phrased that bias in confident language that hides where it came from. Neutrality is not a property you can assume. It is a result you have to test for.

4"What I type into it stays private"

This is the most dangerous of the seven in practice. What you type leaves your screen for a server you do not own, under terms of use you have almost certainly not read, and it may well be stored or reviewed by someone. The rule is simple: no patient data, no draft contract and no trade secret goes into a public system.

5"Technology is complicated, and I am not one of those people"

Building the model genuinely is complicated — and it is not your business. Mastering the use of it, and judging what it gives you, is a linguistic and methodical skill, and if you are an expert in your own field you are better placed for it than the programmer is. The value of what comes out is set by the quality of the context you put in, not by the quality of the code.

6"The next version will fix the problem"

Some flaws are structural, and no upgrade will treat them: hallucination, bias, and the absence of any explanation. Their rate improves with each version; their nature does not go away. This is precisely why trust is never built on the model itself, but on the system you build around the model.

7"It is not the right time yet; I will wait"

Waiting is not a neutral position — it is a decision to let the gap between you and the people who have already started widen until you can no longer close it. Someone who starts today is not ahead of you by one tool they have learned, but by a hundred failed experiments that their judgement now rests on. And judgement cannot be bought; it is earned with time.

Fear of the technology and fascination with it are two faces of the same coin: both of them excuse you from having to understand it.

Why does a myth survive?

Because half of it is true. If a myth were entirely false it would collapse in the first conversation. A conviction that carries a portion of the truth resists correction, because the person holding it finds fresh evidence for it every time it is argued. So you do not take a myth apart by denying it. You take it apart by detail: by showing which portion of it is true, and which portion is the one holding the person back.

The most useful thing you can do in this unit is to write down your own myth — the one I did not mention, because it belongs to you alone. It is the sentence you repeat to yourself to justify putting off learning. Write it down honestly, and then read it again once you have finished the third degree.

Where to after this unit? The myths have fallen. What comes next is measurement: exactly where you stand, on which degree, and in which dimension.