This machine makes Mozart

Finally, Cope's program could divine what made Bach sound like Bach and create music in that style. It broke rules just as Bach had broken them, and made the result sound musical. It was as if the software had somehow captured Bach's spirit -- and it performed just as well in producing new Mozart compositions and Shakespeare sonnets. One afternoon, a few years after he'd begun work on Emmy, Cope clicked a button and went out for a sandwich, and she spit out 5,000 beautiful, artificial Bach chorales, work that would've taken him several lifetimes to produce by hand.

I'm going to try to get a hold of some of these computer-generated works and give you, dear readers, the official Bellman review. But in the meantime, I want to focus on what Jason Kottke said when he linked to the article about Cope:

Gosh it's going to get interesting when machines can do some real fundamental "human" things 10,000x faster and better than humans can.

Humanists, take note! This is not a threatening development.

First, let me posit that while Cope's work's works may not rise to the level of human genius, some algorithms will reach that benchmark during our lifetimes (except maybe your lifetime, Oldy McAbouttodie).

This assertion is not one most of my writer, artist, or musician friends agree with, but they (you?) are wrong, wrong wrong. It's inevitable.

Second, this is not a bad thing. Yes, there will be violent change to our idea of what makes a person or a particular work of art valuable or important. But this change will ultimately be for the better. How could being surrounded by the best art ever made, freshly baked each week, possibly be bad? Computers will find ways to reliably speak to our souls in ways that we rarely have.

And we will all be immensely--almost inconceivably--enriched. The future, she kicks ass.
---
(img via Curved White)

Computer, what is the nature of the universe?

WolframAlpha launches tonight. It will turn your computer into the magic "smarter-than-you" box you've been waiting for all these years. Well, maybe not tonight, but it really represents a new way of making computers work for you. A "computational knowledge engine", it won't just look up an answer, but will compute answers and correlating data.

Check out the demo here.

Bots still don't get irony

Researchers recently announced that through the use of data mining techniques, they've taught a computer program to recognize wordplay and simple jokes.

"The ability to appreciate humor is an enormous increment in subtlety," said researcher Tom Mantei from the University of Cincinnati's College of Engineering in a statement. "You need to know a lot to 'get' humor — a computer does not find it easy."

That's what UC doctoral student Julia Taylor and professor Larry Mazlack have discovered in their project on data mining. They reported on their progress with the project at the American Association for Artificial Intelligence conference in Vancouver this week... |New robot thinks knock-knock jokes are funny - Ars Technica|

Which reminds me of the time a priest, a rabbi, and a computer programmer walked into a bar...

First they came for the mice

safety_first by Blu_Matt.

In a thread I can no longer find over at eripsa, I argued that it is foolish to think that we won't soon have true artificial intelligence because soon we'll be able to simulate all of the activity of the human brain, albeit a little slowly.

Now I see via egg that we are one step closer to this nightmare:

Cory Doctorow: IBM researchers have modelled a mouse's brain at 10 percent speed -- and what can be done at 10 percent speed today can be done at 1000 percent in a couple cycles of Moore's Law. Super-intelligent virtual mice ahoy!

Neurobiologically realistic, large-scale cortical and sub-cortical simulations are bound to play a key role in computational neuroscience and its applications to cognitive computing. One hemisphere of the mouse cortex has roughly 8,000,000 neurons and 8,000 synapses per neuron. Modeling at this scale imposes tremendous constraints on computation, communication, and memory capacity of any computing platform.

We have designed and implemented a massively parallel cortical simulator with (a) phenomenological spiking neuron models; (b) spike-timing dependent plasticity; and (c) axonal delays.

We deployed the simulator on a 4096-processor BlueGene/L supercomputer with 256 MB per CPU. We were able to represent 8,000,000 neurons (80% excitatory) and 6,300 synapses per neuron in the 1 TB main memory of the system. Using a synthetic pattern of neuronal interconnections, at a 1 ms resolution and an average firing rate of 1 Hz, we were able to run 1s of model time in 10s of real time!

Gonna need a lot of virtual cheese.