Gongol.com Archives: May 2021
Decide for yourself whether this is a mere idle thought: There exists a really fascinating platform called Aiva that generates original music using artificial intelligence. It is targeted at those who need to compose large quantities of soundtrack music -- especially for video productions. Some of what it produces now is surprisingly appealing to the ear. With some human nudges, the user can make some of its compositions very good indeed. Because it's based on AI, it gets better with time. ■ We are likely not that far from a time when the Aiva AI won't need many human nudges to come up with some compositions that are really outstanding. Other programmers working in other labs are undoubtedly coming up with similar technologies, too. ■ This is more than merely putting a thousand monkeys in front of a thousand typewriters for a thousand years and expecting them to compose Shakespeare. Artificial intelligence, by definition, is supposed to learn how to work better and better. ■ Here's where it gets weird: Given enough time and training input, it's not impossible to imagine that someday an AI could produce the Singular Best Piece of Music Ever Composed. Seems like a stretch now, but it's no longer an impossibility. ■ Obviously, calling a piece of music the "singular best" is a subjective label -- but by the same token, it wouldn't be hard to find fairly broad agreement among experts that some music is objectively better than other music -- Beethoven's Fifth Symphony is objectively better than the Kars 4 Kids jingle, and anyone who says otherwise is a dolt. So for the sake of argument, let's suspend our disbelief just a little and suppose that a specific individual piece could in fact be named the best. If it helps, pretend that it's merely the best in a particular genre. ■ Suppose, then, that an individual user of Aiva (or a comparable AI tool) clicks "compose" and the AI creates the song, but then the user never actually listens to it, never downloads it, and simply clicks "delete". This means it is possible that someone could entirely unintentionally and unwittingly destroy the Best Song Ever Composed. ■ Certainly, great works are created and lost all the time. There are tremendous stories of professional musicians going to great lengths to try to capture a fleeting moment in song -- Prince built an extensive recording studio to capture those moments of inspiration. "Gimme Shelter" was recorded in the middle of the night. Paul McCartney heard "Yesterday" in a dream. ■ Now, if you were to imagine the Best Song Ever in a dream, then you may not be able to record it anywhere or share it with anyone. But at least you would have dreamt it and experienced it, if only subconsciously. But if an AI composes the song but a human deletes it without ever listening, then it was never heard at all -- an existence even more ethereal than a dream, since it in fact existed in the known, conscious world, but was never experienced. ■ It is hard not to find it peculiar that, thanks to artificial intelligence, it is now technologically possible to create and then accidentally destroy a phenomenal piece of art without ever knowing that it existed. This isn't just a question of "if a tree falls in the forest and nobody hears it" -- it's a matter of something potentially perfect being created and then never being replicated (because, ultimately, each and every AI-generated song ought to be in some way unique). ■ This bizarre ability to potentially destroy something amazing without ever knowing it existed (beyond, perhaps, reading a file name) seems almost worse than the problem of works created and recorded in the analog world that may not ever make the leap into the digital domain. At least there are conscientious people working at projects like the Internet Archive who are trying to rescue analog materials and digitize them so that they aren't accidentally lost. ■ Historians look to the destruction of the Library of Alexandria as one of humanity's great unforced errors -- a colossal waste that almost certainly did some measurable damage to the progress of human knowledge. How strange it is, indeed, that anyone with access to our increasingly powerful tools of artificial intelligence may entirely without intention and without malice destroy great works with no more thought than a keystroke.
On the east side of the Des Moines metro, Facebook is going to build its 9th and 10th buildings at a vast data center. Your files have to be stored somewhere, you know.
An interesting perspective on being a trial participant for an unproven vaccine after proven vaccines become available: "What if you stay in the placebo group and you get infected and you infect somebody else and that person dies?" Let us be perfectly clear: These kinds of conflicts are exactly why people with a strong orientation towards the practical and scientific schools of thought need to be trained in the "soft" sciences of the humanities, and why people who study the humanities need to be science-literate. We have to be able to communicate across conventional boundaries if we are to have any hope of addressing complex issues.