Today i am sharing some early stage work for a short guidebook on Generative AI: i am still trying to find the language and style of this piece, and indeed still unsure whether this work will ever see the light of day. So i’m sharing this extract in that spirit, as part of #WorkingOutLoud. Specifically i am calibrating my writing to find the level and feel of the work – it’s playful, but developing a language and ‘feel’. I cannot articulate it clearly, but i will know it when and if i find it! This extract is in the third section, exploring what ‘intelligence’ is – and as you will see, not from a detailed scientific perspective, but a more ethnographic or exploratory one.
What is Intelligence?
For an intelligent species, we struggle to precisely articulate what intelligence is.
Our definitions will tend to focus on problem solving, on our ability to reason, the creation of knowledge, our ability to project ideas from one space to another: to conceptualise and be creative.
It all sounds rather grand and exciting.
You could probably legitimately argue that our intelligence is also illustrated by our ability to lie, to cheat, to deceive, to steal in creative ways, to to have invented tax systems of infinite complexity.
Arguably our intelligence is best indicated by our ability to play, create, invent, and learn.
All these things may be true, but what is probably most true is that we tend to define it by what it is not.
We define intelligence by absence, in that we look down upon those things that do not have it.
It’s quite a long list.
Starting with stones. Stones, by most people’s standards, are not intelligent. Unless they are just thinking really really hard.
Tulips, similarly, are not intelligent. Although they do have the ability to turn their heads to the sun, which beats stones hands down.
Ants, well individually they are not strong contenders, but they do seem to achieve quite a lot together. So maybe not intelligent per se, but slightly clever?
Squirrels? Famously clever, within a narrow and nut related domain. Nobody thinks that a squirrel is going to steal their job. So it’s a very narrow capability.
Monkeys? Slightly tricky as some of them use tools (one measure of our cleverness), and they fight, and care for each other, both of which seem quite human. But there is no great concern about their capacity for collective action to overthrow us.
Babies? Well, Meadow aged ten months once crawled into a corner and was unable to reverse herself out, presumably resigned to the fact that the universe had become smaller and more angular than it used to be. But not obviously world beating intelligence. At least not yet.
My iPhone? Clever, but in a different way. Not intelligent. But very useful.
Generative AI? Well, that’s the open question. Clever in it’s conception, for sure. And very handy. But is it intelligent?
Generative AI feels like a leap-ahead advancement that brings AI closer to human-like intelligence.
However, in practice, this isn’t entirely true. Generative AI still relies on pattern recognition and doesn’t truly comprehend. That’s why some have referred to them as “stochastic parrots.”
Clever, but nothing to worry about?