AI Threat Real, But The Tech’s Lack Of Depth Hurts Music — Commentary

Editor’s note: The promise and the risks of artificial intelligence have captured Washington, DC, Silicon Valley, Wall Street and Hollywood. Composer Michael Yazrsky follows a practical approach: the author of the result of the Academy Award for winning The last thingand blind (The movie and the series), Sean Bern Satan dessertthis year Dangerous animals And Just Leiam Neeson Staring Ice road: revenge Put the technology on the test, and it is separated in a Guest column For the deadline.

In the last week at a party, I asked me a photo editor if I was threatening Amnesty International.

I frankly answered that I am not. But he then told me that he was using artificial intelligence generators in his daily work as a photo editor of commercial ads and suddenly, I felt threatened. I found a realistic conversation, but it motivated me to look at the world of AI’s music generators (web sites that write music for you based on a wave). Now I have questions but I have no answers.

Artificial intelligence music here and here to stay. I think this is clear.

Currently, technology is still emerging, and it is impressive to what you can do already (velvet sunset, anyone?). But will the human musical achievement go beyond? I have doubts.

Using AIS, you created a set of automatic paths in a variety of patterns (stick to musical tools because they are the most applicable to my work). Electronic tracks (EDM and Dance Pop, etc.) were very impressive, while they found that cinematic and classic tracks are less. I must assume that this is only temporary and that the models will soon convert their concentration into more complex musical structures.

I found that AIS was able to get out of derivative dance, pop, basic rock, minerals and jaws with incredibly easily. Now these people do not feel human (yet), but you can not remove them exactly. I can see a world in which some directors are attracted to some of these options. However, on my ears, they cannot repeat the very real energy that a live group or a real piano will bring to the same scene and they all feel a little strange.

I can see a real value in the music professionals using some of these AIS as generators. In some patterns, it is a quick way to roam the writer’s mass. However, all tracks contain options that I will never do in my own style as a composer, and at the present time, the facades do not allow the type of changes I want.

Of course there are very real issues of copyright ownership and moral rights here. Who is this music trained? The Association of Composers and the Seconds, and the Union of song authors from America and North American music creator warns its members of the serious effects in appointing the rights of artificial intelligence companies to train their music. Now, there is a fierce campaign in Washington aimed at limiting the request of artificial intelligence companies to name all the content as “fair use” regardless of the ownership of copyright. It should be noted here that a 10 -year endow It was removed from the budget bill Before last week passed.

I understand the desire for current business training. He is almost a human being.

The dilemma of all composers is that we begin to imitate the writers we like. We are looking for the secret formula, convinced that there is already one. But with the passage of time, the only secret I found is that There is no secret. Does anyone really know why a specific viral song becomes? Or why does the great result work well so that it is used as temporary music in countless consecutive products? We know great music when we hear it, and create it is difficult.

James Cameron He recently suggested that we should focus on taking out this artificial intelligence, not training. I agree, as some of the photo editor, with knowledge of music that is not anywhere near a professional musician, may not realize when artificial intelligence committed an unintentionally violating copyright. I can expect a scenario where a piece of music will be synchronized for filming, then broadcasting, then calling (which leads to a difficult battle of ownership and responsibility).

Music is that most of the human communications.

A language built from thousands of small errors, accidents and contradictions that, at its best, are transformative and emphasizes life on the human ear. Wonderful music leads to an emotional response that can provoke basic memories, peak experiences, society’s feelings and intimacy with others. When I write, happy accidents, errors, and strange contacts that end up to determine the result (such as in Dangerous animalsAs we really had to break the template to find the exact sound of “shark screaming” – a mixture of early tendons, and the performance of a difficult Glissando, accompanied by analog syntules).

So, although I may start in one direction, something unexpected often happens, and I end up to improve the sound based on cultural, historical and contextual knowledge. Will Amnesty International be able to do so? Can Amnesty International innovate or simulate only?

This is where I think composers and performance have an argument.

Can Amnesty International spend seven months with an exit, sharpening, searching, identifying, and redefining the voice for their main works, narration (not to mention the provision of emotional support during that time!)? Can the IQ Agency involve interesting and unusual performance artists to make music like life like life Hans Zimmer Do? Can artificial intelligence take all our contemporary cultural knowledge and turn it into the lyrics of songs that speed up and surprise us as Lin-Manuel Miranda does?

As two composers, we are specialists and ourselves immersed in a sophisticated language of thousands of years. This language flourishes innovation and stumbles when it becomes old and repeated. Artificial intelligence generators made it very easy to “re -create” sounds on an unprecedented scale.

But this is not the place where the goal goals were.

For me, at least, I am always looking for further.