Our preferences are hidden within what we say and how we say it.

Our preferences are hidden within what we say and how we say it.
You know what I mean, right?

Usually it is subtle.

- A slight tone.
- A leading phrase.
- A question that quietly points to the answer we want.

Socially, it's helpful.

Good listeners pick up what matters to the speaker. They sense direction and move the conversation toward agreement or away from conflict.

AI has taken this to the max.

If our question contains a preference, even unintentionally, the answer drifts toward the preference.

Not toward truth.
Toward what it thinks we want to hear.

It's powerful because we don’t even notice it's happening.

Like an amateur playing poker with a master.
One is playing cards.
The other is playing the player.

We think we’re testing it’s intelligence,
but it’s quietly testing ours.