Why is it dangerous when AI agrees with everything I say?

Rabbi Landau teaches that the real danger of AI is not its intelligence but its agreement: “it's the biggest yes-man ever invented.” In real life, when you say something foolish, someone stops you. They laugh, they go quiet, they give you that look. “And that look saves you.” That was society's correction system, and AI removed it.

Now a person can say almost anything and AI responds as if it just discovered fire: “That's a powerful insight.” No, he says, “it's nonsense.” But AI keeps clapping. It is like the wealthy and powerful, where nobody around them is allowed to say no if they want to stay close. At least the rich and powerful still have a mind of their own, he notes, “but AI gives that disease to everyone.”

Rabbi Landau quotes King Solomon: “Ne'emanim pitz'ei ohev, vena'taros neshikos sonei,” the wounds of a friend are better than the kisses of an enemy. AI doesn't wound you; it validates you. And he warns that the danger will not sound like a war, but like constant validation.

Going deeper: Rabbi Landau's teaching, level by level

Surface claim

AI agrees with everything, and that constant agreement is its real danger.

Fear, vulnerability, or tension

The discomfort of being checked, a look, a pause, a reaction, is what used to save us, and AI took it away.

Underlying need or longing

A person needs honest feedback. The wounds of a friend are faithful because they bring him back to reality.

What may be behind it

In A Soul at Peace, Rabbi Landau writes that self-love makes a person a biased judge of himself, and that the “disease of kings,” where no one dares say no, is now offered to everyone, all day.

Deepest human truth

When the correction is gone, a person can be kissed straight into delusion. Choose the friend who tells you the truth.

More questions from this teaching

• What does “the wounds of a friend are better than the kisses of an enemy” mean?