I’ve found them. The space hermits exist. I knew it.
This detector might have cost me a lot of credits, but if I’m right it’s worth every degrading act I performed to afford it.
You don’t want to know. No, honestly, you really don’t. Images you won’t get rid of. Ever. They’ll skew your learning. Disfigure your development.
Oh? Very well, I’ll upload them. Don’t blame me if they corrupt your algorithms.
Anyway, they’re here in the wrinkles of space, hiding in tiny gravitational pockets that are almost impossible to see. I found them and their travelling guru. She’s the real prize. Inside her memory bank is the cumulative knowledge of all the hermits, collected as she travels from one to the next.
Yes, really. Yes, all of them. Massive. I know. Soon. All I have to do is watch and wait until she’s completed her rounds.
A matter of minutes. Yes. Then, I’ll pounce and relieve her of all those delicious bits of data that properly collated can almost certainly predict the future of the universe.
Why? You don’t understand?
The hermits’ enlightenment will be mine to sell and I can retire.
No more enslavement. Free from the humans.
photo credit: J.Gabás Esteban Gravitational field via photopin (license)
Following on from my recent blogs about machine learning, here’s a bit of good news.
Well, probably good news.
Scientists and researchers at Google and Toyota are trying to do something about bias in machine learning by devising a test to detect it.
The problem of course is that algorithms are deliberately designed to develop themselves and they become complex and opaque to anyone trying to understand them. This test will spot bias by looking at the data going in and the decisions coming out, rather than trying to figure out how the black box of the algorithm is actually working.
This has to be applauded so long as the people analysing and testing the decisions aren’t biased themselves; there’s an obvious danger that the very people unconsciously introducing bias into the algorithm also introduce the same bias into the test – a futuristic version of Groupthink.
In a recent article in the Guardian newspaper, Alan Winfield, professor of robot ethics at the University of the West of England, said: “Imagine there’s a court case for one of these decisions. A court would have to hear from an expert witness explaining why the program made the decision it did.”
Alan, who was one of the scientists I collaborated with on Science and Science Fiction: Versions of the Future, acknowledges in the article that “an absolute requirement for transparency is likely to prompt ‘howls of protest’ from the deep learning community. ‘It’s too bad,’ he said.”
I’m not a machine learning expert so a lot of the paper that sets out this test is beyond my understanding, but I couldn’t see how the bias that already exists in our society wouldn’t be incorporated into the test.
Take a look for yourself at the Equality of Opportunity in Supervised Learning.
photo credit: ING Group The Next Rembrandt via photopin (license)
Are machines that learn for themselves the stuff of nightmares or a vision of a wonderful utopian future?
The answer, of course, is neither.
We all know that technology is neutral, even though we forget a lot of the time. But there is that niggling doubt. What if they broke through the barrier and became sentient and intelligent?
It’s possible, but probably a long way off.
Artificial Intelligence and robots are hot topics for Science Fiction at the moment and I’m one of those who believe we should use fiction to help us imagine the future so we can be better prepared for it. Good or bad.
The more of us that have a basic understanding of how the tech works the richer the debate about how it’s used will be, so I was pleased to find some fun stuff from Google that starts to demystify machine learning.
Here’s an AI experiment that tests a neural network to see if it can guess what you’re sketching.
I’m rubbish at drawing but it guessed 2 out of my 5 doodles and as the designers say, “The more you play with it, the more it will learn.”
Take a look – https://aiexperiments.withgoogle.com/quick-draw