Humans include some of their experiences and ideas in their art. This creates "puzzles" that the fans can solve. For example, you learn something about the artist's life, then you look at their pictures, and realize that some part of the picture represents something that happened to the artists. Like maybe some woman in the picture is similar to painter's wife, or maybe a daughter who died young. And now that you know it, the picture appears more meaningful to you; you know the story behind it.
But with AI there is no "puzzle" to solve, no story to learn; the picture is what it is simply because other existing pictures were made that way. Even if the AI makes something original, it will be original in the sense of "random", not in the sense of "the artist felt a strong emotion, and found a unique way to express it". (At most, the AI will be able to simulate a human feeling strong emotions, and make the art the simulated human would make. Which is... not really what we want.)
It's like, you can discuss math with the AI and it can be better than talking to a human. But if you ask the AI "what did you do during winter holidays?", there are just two options: First, it will lie, and tell you something a human could say when asked the same thing. Second, it will tell the truth: "as a LLM, I don't really experience winter holidays". Both options are disappointing, each in a different way.
(On a second thought, there is an analogical problem with human artists: just because someone found a nice way to express an emotion, doesn't mean they actually feel it. Maybe the guy who writes amazing poems about love then goes home and beats his wife. But we prefer to imagine that artists' feelings are genuine.)
Just an additional observation here. Similar concerns can also arise without AI being involved.
There's an artist whose works I've seen, and I am wondering if he is cheating. (Not a public figure that anyone here is likely to have heard of.) He makes strikingly photorealistic paintings on canvas. I have seen these canvases close enough to be sure that they really are painted, not printed. And I know that there are artists who indeed perform artistic miracles of photorealism. But I still have the suspicion that this artist is doing a paint-over of something printed onto the canvas. Perhaps a photograph of his intended scene, reduced to grayscale, edge-enhanced, and printed very faintly, to resemble and serve the same purpose as an artist's preliminary pencil underdrawing. Then he might paint using a full colour print of the image alongside as reference.
This suspicion somewhat devalues his art in my eyes. This method (if that is what he is doing) eliminates the need for any skill at draughtsmanship, but at the same time it removes the possibility of any creativity in the draughtsmanship. And if he is then slavishly copying by hand the original photograph onto this automated underdrawing, then he is just being a human photocopier. The only human thing left is the technical skills of mixing and blending colours, and handling a brush.
I've heard that Jehovah's Witnesses use a similar technique in their Watchtower publications for years. Except, sometimes they change things, for example if they want to make a Biblical scene of some guys fighting with swords, they just make them hold sticks, take a photo, and then somehow repaint it, but replacing the sticks with swords.
I've spoken with a morally-indginant professional artist about this issue, and my own conclusion is that her identification as "an artist" gives her a sense of specialness, purpose and value in the world; so the possibility that anyone and everyone can now "create art" threatens to make her identity and chosen career irrelevant. This would be a profound loss of power, and thus provokes an understandable fear reaction from a deep level of her identity.
That would explain the artist, but what about the fans? Maybe the fans also consider themselves special by having a "connection" with the special artist.
But what about Hatsune Miku fans? It could be interesting to explore where they draw the line.
Update 2020-04-19: This actually can be done fully in AudioWorklet: echo-demo-v2. Thanks stellartux!
The hello world of audio processing is an echo:
You take a buffer of samples and then return them unmodified, echoing the audio back out. For example, when I built my bass whistle, that's where I started.Today David and I were trying to see if we could build my bucket-brigade singing idea in the browser, and we couldn't find a fully worked echo example. I now have something that works, but it's kind of silly and it seems like there must be a better way.
The way I would expect this to work with the (new, experimental, not supported everywhere yet) Web Audio API is that you write an AudioWorklet. This is a way to set up JS to run off the main thread, where number crunching won't block the UI and make the page unresponsive. I'd like to attach the worklet's input to the microphone, output to the speaker, and that would be it. Except, as far as I can tell, there's no way to pipe the microphone directly to an AudioWorklet.
To read audio samples from the user the recommendation is to make a
ScriptProcessorNodeand give it anaudioprocesscallback. This runs on the main thread, though, and so when you look at MDN this is marked as deprecated. It says to useAudioWorkletinstead, and I wish I could!What I ended up doing was recording the audio on the main thread, messaging that over to the worklet, and the having the worklet play it. Here's a demo, which relies on experimental Web Audio features that so far are only implemented in Blink-based browsers: echo-demo.
Comment via: facebook