Projection is nine-tenths of the law

Projection is nine-tenths of the law when it comes to AI.

Humans project consciousness and meaning into almost everything.

Our pets. Our cars. Ever said to a broken printer “oh, please just work!”?

Large Language Models benefit from this projection, but to a level we’ve not really experienced before.

They can feel intelligent, but they’re not.

No wants.
No feelings.
No intelligence in the way we imagine it.

Just next-word prediction. It’s absolutely incredible how effective that is.

The illusion of conscious intelligence has some upsides. We can use it easily and naturally. You can use AI as an external thought enhancer, a coach, a sounding board that sometimes feels like someone who’s knowledgeable and who cares.

On the other hand, it’s very risky. We’ll see more people befriending AI, falling in love with chatbots, taking advice on relationships, finances, even how to cope with suicidal thoughts… forgetting there’s no coherent mind behind the answers. That won’t always end well.

Our ability to get meaning from ChatGPT is the same ability that lets us “see Jesus in pieces of toast”. Dangerous if we forget the illusion, or are unaware of it to begin with.

Can we stop projecting more intelligence onto AI than is really there? Or is it just the way we are wired? 🤔