Amplifier, Not Replacement

It Multiplies What You Already Have

Observed repeatedly across fields: AI increases the output of people who already know what they are doing more than it rescues people who do not. An expert asks better questions, spots wrong answers instantly, and knows which parts matter. A novice cannot evaluate what comes back and often accepts a confidently wrong result. There is a real levelling effect too — these tools genuinely help people get started, and access to good explanation is more equal than it has ever been. But the multiplier framing is the useful one: the tool scales your judgement, and scaling nothing gives you nothing.

• Expertise makes the tool more valuable, not less necessary
• The ability to recognise a bad answer is what separates useful from dangerous use
• Real access benefits exist — good explanation on demand used to be a privilege
• A multiplier applied to zero judgement still produces zero

What Stays Yours

Accountability does not transfer. If you submit it, publish it, ship it or say it, you own it, and "the AI wrote it" has never worked as a defence and is not going to start. Taste is yours — knowing which of five decent options is the right one for this audience is a judgement built from experience. Deciding what is worth doing at all is yours, since a model can generate options but has no stake in the outcome. And your relationships are yours: trust between people is still the thing organisations run on, and nobody has automated being someone others want to work with.

• Responsibility for output stays with the human who put their name on it
• Taste and selection are judgement, and judgement comes from doing the work
• Deciding what matters is a values question, not an information problem
• Trust and reputation are built between people and cannot be generated

Working With It Well

A workable division of labour: use AI for volume, breadth and speed, and use yourself for direction, judgement and the final call. Generate twenty ideas and choose three. Get a rough first version and rewrite it properly. Ask it to argue against your position and see what survives. Have it explain something outside your expertise, then verify with a real source before relying on it. The failure mode to avoid is drifting from using it as a tool to using it as an authority — the point at which you stop evaluating output and start just forwarding it onward.

• AI for breadth and speed; you for direction and the final decision
• Use it adversarially — ask it to find the weakness in your own argument
• Keep evaluating output; the moment you stop, you have handed over the judgement
• If you cannot explain why the output is right, you are not ready to use it

Try It Yourself

Using AI adversarially is the fastest version of the amplifier idea, because you keep the position and it does the attacking. Try it on something you have actually argued about.

Here is a position I hold:

[your two or three sentences]

Make the strongest possible case against it. Then tell me which single part of my argument is weakest and why. Do not be polite about it, and do not tell me what I should conclude.