Assorted links

Comments

5. It won't really be mind-blowing until they succeed in writing Turing test novels. (Well, good novels, producing bad novels is probably well within today's capabilities.) Next will come customized immersive entertainment in which plots develop in realtime and then a large chunk of humanity will move into the box.

Well, we've got AI-created video games:
http://www.rockpapershotgun.com/2012/06/04/the-ai-that-designs-its-own-games/

Closer to the subject, there have been previous musical composition projects (C.P.U. Bach, David Cope/Emily Howell, etc.) some of which have been claimed as indistinguishable from human composition.

To my mind, the central question that comes up in computer-generated art is the difference between computer-mediated art (where we assign artistic intent to the programmers or software users) and computer-created art (where we credit the machine, instead). The line between the tool, the artist, and the creator of the tool makes it

Comments for this post are closed

Comments for this post are closed

That Contemporary music sounds like random bursts of noise doesn't reveal anything about the state of artificial intelligence. When a computer generates a pop song, then we will be talking. I'd be surprised if it happens in the next 50 years.

I'd be surprised if it didn't happen in the next ten, particularly if I can stipulate that the computer just needs to compose the piece, and that humans can actually perform it.

Comments for this post are closed

Comments for this post are closed

The humans failed, but this provides no new information. Nor will it help us improve the ratio of great music to terrible. I'd rather the computing power be used to analyze great music that we know is great, rather than generate new, bad music.

Generating a good pop song via computer may be easier than generating a ballet:

http://blog.hooktheory.com/2012/06/06/i-analyzed-the-chords-of-1300-popular-songs-for-patterns-this-is-what-i-found/

When the time is right, modulate up a step.

Comments for this post are closed

Open Letter to GMU Econ Professor and Chess Master T. Cowen: Scott Summer sez: "GDP [in Iceland] is still below previous peak, but I think one could argue, much more so than in say America, that a significant part of that peak involved a Ponzi financial sector that isn’t coming back." A post please on why the popping of the dot-com bubble in 2000-1 involved $11T of wealth evaporating in the stock market but did not cause a systemic collapse in the overall economy while by contrast the popping of the real estate bubble in 2006-7 involved (I think) a mere $3T of wealth evaporating but DID cause a systemic collapse. Is "leverage" in the financial sector the answer? You mean if financial firms are Ponzi that's bad for everybody, but if Joe Blow is Ponzi that's not bad for everybody? Please explain.

If you were a regular reader of Scott Sumner you'd be less likely to refer to the 2000-1 stock slump, but rather the astounding 1987 crash which had NO detectable impact on the economy. The Greenspan years really were different.

From a brief review of the literature it seems that property is digested differently by the economy that stocks. For some reason (and the same was true in the 1920s with FL real estate and indeed a residential boom/bust cycle in the mid 1920s --housing starts went up an incredible 33% in the late 1920s-- *before* the stock market crash of 1929) a housing boom/bust is more prone to bring down the financial system than a stock market boom/bust, even if the latter is bigger. I don't know why and the literature I've seen does not say why. Perhaps because banks historically hold mortgages? In terms of money, the corrected figures are: $10T (not 11T) for the stock bust (wealth lost) vs $3T for the real estate bust. So you see a three-fold increase in stock bust money in 2001 did not dent the economy as much as a real estate bust did in 2007-8.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Why should readers of an economics blog bother about what the scandals in the Vatican?

Yeah, it's not like institutional structures are relevant to the field.

Comments for this post are closed

Comments for this post are closed

The targeted genre is experimental and possibly better understood in terms of trace rather than in terms of product. It could be argued, then, that computers have proven that the human composers are largely succeeding in their project of unbounded experimentation, since that is what the computer is designed to do.

Comments for this post are closed

re #5: Further proof that the Darmstadt school and all its progeny failed.
Many modern 'avant-garde' composers use formal rules to generate music, so no surprise. When a computer writes Nymphes des bois ...

Comments for this post are closed

Comments for this post are closed