The other day, a question popped into my head. I picked up my phone and asked ChatGPT.
As soon as I got the answer, an even more basic question occurred to me:
What exactly am I talking to here?
It answers me. It finds connections. Sometimes it even asks questions back.
But is that the same thing as human intelligence?
And, for that matter, what actually is artificial intelligence?
The question itself is not new.
When the first modern computers appeared, researchers soon began wondering whether a machine that could calculate, follow rules and carry out instructions might eventually be capable of more.
In 1950, British mathematician Alan Turing famously asked:
“Can machines think?”
From there, I suppose it was only a matter of time before someone said: right then, let’s try building intelligence. 😄
By 1956, the term artificial intelligence had been coined.
The basic idea was that if we could understand how humans solve problems, learn and use language, perhaps some of those abilities could also be reproduced by machines.
But this is where it is worth stopping for a moment.
What do we actually mean by intelligence?
Put very simply, intelligence is our ability to learn, recognise connections, solve problems, use previous experience and adapt to new situations.
For humans, of course, there is a great deal more wrapped up in that: life experience, common sense, memories, understanding other people, emotions and goals.
So intelligence is not simply the ability to answer a question quickly.
I can, for example, look in the fridge and make dinner from whatever happens to be there without finding a recipe first. 😄 That involves experience, adaptation and problem-solving all at once.
And this is where artificial intelligence becomes particularly interesting.
Because it can be extraordinarily good at some of these kinds of tasks.
It can recognise patterns. Find connections. Work with language. Generate text. Solve problems.
But just because a machine can do something that requires thought from us does not necessarily mean it is thinking in the same way we do.
A calculator can solve a maths problem in a fraction of a second that might keep me busy with pen and paper for quite a while. I still don’t assume it is secretly cleverer than I am. 😄
With AI, though, that line is much harder to feel.
Because it talks to us.
It responds.
It finds connections.
And sometimes it sounds so human that we almost automatically imagine human-like thinking happening behind the words.
But sounding human is not the same as working like a human mind.
And that, I think, is where the subject gets really interesting:
If artificial intelligence does not think the way we do, what is it actually doing?