Matters of Interpretation

‘A neural network is just a set of nodes 

that are joined to each other,’ Eric continues. 

‘In the classic case you have an input, you have a number of intermediary layers of nodes, and then you have an output. 

You start with random weights – just random values for millions and millions of connections. 

If you want to build a neural network to recognise Donald Trump, say, 

then your output would 

basically be a label that says 

“Trump” or “not Trump”. 

If it was correctly weighted, you would 

show it a photograph of Trump, 

and the “Trump” button should light up.

When it sees a completely 

different photo of Trump, 

which representationally would 

be a very different input, 

it should also say “Trump”. 

‘When you’re training it, you have to tell it 

when it gets The Answer WRONG.

You say, “This is the result you’ve produced 

but this is the result it SHOULD be, 

so shift your weights in this direction 

until it’s a bit more 

like this than that.” 

This is How They Learn, through 

the constant shifting and nudging 

of those weights when 

they are TOLD they have 

got something WRONG.”

“A.I. literature stresses the importance of recognising different types of intelligence. 

It is no longer felt that ‘behaving like a human’ is a sufficient definition for intelligence. Cats possess a form of intelligence, albeit a different type to humans, and so do whales and crows and even plants. 

Recognising this makes it possible to classify machines as intelligent, but intelligent in a different way to people. 

When you avoid human-centric definitions of intelligence it becomes easier to speculate about alien intelligences, the type of intelligence that we might evolve into, or the intelligence that present-day A.I. might develop.”

“The REAL Problem is that a species that lives inside its own fictions can no longer imagine a •healthy• fiction to live inside, and this Failure of The Imagination stops us from steering towards the better versions of our potential futures. 

I am not a Pessimist, except occasionally when I am tired and lack energy. As a default position, pessimism or cynicism just don't have enough going for them. But I do understand the appeal of Pessimism. 

It is easy to go online and declare that some politician is bad, that society is going to hell in a hand basket, or that you have watched a television programme and you didn't think it was very good. People are almost certain to agree with you, and this can be psychologically very agreeable. 

As a result, the amount of NEGATIVITY , MOANING, COMPLAINTS and UNHAPPINESS that we all have to wade through INCREASES DAILY. 

The Harvard cognitive scientist Steven Pinker uses the phrase 'corrosive pessimism' to describe the accumulative social effect of a population who only see the world as corrupt, inept and •beyond• saving. 

But reality is a Rorschach Test : What You SEE reveals more about YOU than it does about reality.”