Don't fear the Malthusian trap

2016-12-07 00:00:00 +0100

Scott Alexander, from Slate Star Codex, once uppon a time wrote a fairly long piece called “Meditations on Moloch”. It is fairly hard to summarize, but it is so much worth a read that I am just going to assume here that you have already read it.

I want to write here about the bleak vision of Moloch taking over humanity, stripping us of everything that is worth living for, and converting us into unhappy, supercompetitive beings fighting for ever fewer resources. This is the Malthusian Trap, and it is literally a fairly depressing vision, in the sense that it can make you very unhappy. It certainly tormented me when I was younger.

[...read on...]

How to ruin a great product by adding it to the Internet of Things

2016-11-29 00:00:00 +0100

The kitchen ovens most people have are relatively blunt tools with ample opportunity for improvements. But you can ruin an improved version by trying to convert it into an iot sci-fi contraption.

The oven in question is full of sensors and some logic that might make it a really good oven, but then obscured by a thick crust of AI, IoT, and Big Data that, it seems, makes it hard to use and disappointing.

[...read on...]

The old internet is better than ever

2016-11-26 16:04:59 +0100

For a large number of people, Facebook is the internet, What’s App is how you communicate, and Twitter is how you quip and snicker. I don’t have accounts in these sites, and I don’t plan to have them. I prefer The Old Internet (is it appropriate to call it “Ye Olde Internette?”). And it’s better and more free than ever before.

[...read on...]

Artificial Intelligence is really very, very hard

2016-11-23 09:23:01 +0100

The Register has a fun article on how a very well funded AI research team couldn’t make much of a dent in the ages-old problem of creating an artificial intelligence that can make sense of text. After a lot of cash and smoking heads, the software couldn’t answer very simple questions about the world. Their benchmark base-case was

For example, a well-trained agent should be able to understand and answer questions from simple sentences such as: “Mary picked up the ball. Mary went to the garden. Where is the ball?” It should reply, “garden.”

[...read on...]