AIVA - the Intelligent Composer

We saw this kind of program back in the 1960's. It pumped out endless Bach fugues. Nobody claimed it to be intelligent. It was clever. The cleverness belonged to a human programmer, not a machine.
AIVA is a musical instrument being presented by a musician and a guy who sells musical instruments. It's more like a piano than a Beethoven. Speaking of it as somehow being "creative" is a serious error logical error. It's a sales pitch by a thief who wants to sell you the work of other human beings. Anyone who plays a musical instrument knows how they seem to have a "life of their own", but nobody seriously thinks of a Stratovarius violin as being creative or alive in its own right.
We know that a composer can be sued for copying a single phrase or chord progression from another composer and claiming it as his own. But if the composer copies hundreds of thousands of pieces, has he created a creative MIND? Or just a big pile of loot?
We cannot expect this guy to actually pay the copyright fees associated with all the music he has mashed up to sell as an "intelligent machine". Like Google and Facebook, AIDA is a machine that is designed to publish, devalue, steal and re-sell the works of human beings. By speaking of it as an intelligent being, he escapes the moral and legal liability that would arise if he did all this by a transparent "human" process. He's just invented the theft platform. He holds no responsibility apart from collecting the money he makes by re-selling the works of thousands. Not his fault. It's just an "algorithm".
Of course, his AI composer will compete with human composers, drive their incomes down and put many out of work. That's the common feature of these algorithms posing as persons. It's the real agenda that drives the value of real human creativity to zero. Where does the profit of this enterprise go? Let me know when his "intelligent machine" opens a bank account.
This essay owes a great deal to the ideas of Jaron Lanier. If you want to understand what AI is and what the impact will be, Jaron has the inside information and a perspective you need to know.
There is some insight to be gained from this presentation. Jaron's perspective tells us about the deep and sinister trend to hijack language in the interest of storytelling (and sales pitches). But you can also get some insight into the nature of "deep learning" from the great visuals. These days, "AI" is almost entirely this kind of technology.

There are lots of great TED presentations that describe how deep learning works*. Once you know the trick, you can be impressed. In some way, we are learning the secrets of how the visual cortex works (and probably the cortex in general). If you think we are on the way to creating a machine "mind", it's good to know exactly what it is we are inventing/discovering. Personally, I think it's best to think of these algorithms as extensions of perception - tools that allow us to "see" patterns that we would not otherwise see. They are like the wings of an aircraft that allow us to fly higher and faster than a bird ever could. But they do not "see" any more than a camera or a pair of binoculars "see". They do not "create" any more than a hammer or a guitar creates.
-----
* For example, this a mind-blowing and intuitive presentation of how the trick works "behind the scenes".  More detail on the images from "Deep Dream" here. If you want to buzz on an old-fashionedco Mandelbrot zoom, try this.

Comments

Popular posts from this blog

Facebook and Bing - A Killer Combination

A Process ...

Warp Speed Generative AI