I’m of two minds about this. (Destiny)

by Cody Miller , Music of the Spheres - Never Forgot, Thursday, August 22, 2024, 07:57 (91 days ago) @ Coaxkez
edited by Cody Miller, Thursday, August 22, 2024, 08:02

Now, 343 is absolutely jumping the gun here. There's no question about that. But my point, economically speaking, is that AI doesn't need to be "as good" as human output in order to cause major disruption; it only needs to be "good enough". Once it gets to a point that consumers are willing to spend money on AI output preferentially over human-generated output, it's over. The cost savings will be too difficult to fight.

This is predicated on the idea that it can ever be cost effective though. Many economists and people studying it are saying it is not looking good even decades out.

https://www.goldmansachs.com/images/migrated/insights/pages/gs-research/gen-ai--too-much-spend,-too-little-benefit-/T...

It's certainly interesting technology, but it won't be able to magically do things it wasn't designed for. A steam engine might power a car, but it can't send you to the moon.

Even if it gets "good enough", that only matters if it is significantly cheaper than human labor, which it is not, and won't be. "Good enough" only matters if it can easily integrate into your project without a human supervising or reworking it which offsets any cost savings.

These technologies offer no actual understanding of what they generate. You cannot say, "Take the exact scene you just generated and move the camera 6 feet to the right". The resulting media is completely different because it has no understanding of what it just made. Yet such corrections are trivial for a human, and a huge part of the iterative process of art.

Economics is everything. The bubble bursts as companies realize it doesn't actually save them money.