So, as these things go, I said "AI can think, but it cannot feel" on this AIScento podcast (clip via LinkedIn here) and I'd like to clarify the point.

When I say "thinking" I mean pattern recognition and rule-following. Like algebra. AI is great at solving equations this way.

When I say "feeling" I mean something different. Feeling is a leap - a connection that breaks from the pattern, one that defies a clean algebraic proof.

You only get to thinking by following the rules.

You only get to feeling by breaking one.

It's even clearer (to me) when you add time.

Thinking only requires processing time. It can happen at the speed of light. Ask a calculator the square root of 647,249 and you get 804.5303 immediately. Accurate and instantaneous.

Feeling requires lived time. Experience. It can happen fast, but it also evolves in a way a formula doesn't. If I play you a song from your childhood, think about all the years that feeling has been developing. Feeling doesn't have a quadratic equation.

I don't think AI can feel. The human users can - the machine can't.

Which is why when people ask me about trust in an age of AI, I apparently say: "AI can think, but it cannot feel."

Trust takes time. It takes interactions. It takes feelings. No amount of thinking, in isolation, can fully replace it.

Play with the idea. And hey - you look smarter than me. What's wrong with this logic? It came out of my mouth on the spot and I'm really reflecting on it. Lemme know.