5 Comments

Well done. The music creation stuff is moving really fast! I had a friend who described his experience maybe a month and a half ago, and it seems like you can do a TON more, much more easily right now. This revolution is less about being able to do things we were unable to before, and much more about doing things we could have done in 10 hours in 10 minutes.

Expand full comment

Yeah tell me about it. I wrote the piece "The Current Pace of AI Progress is...Nuts?" (https://www.whytryai.com/p/ai-progress) in October last year, before ChatGPT dropped. The "Music" section there is pretty indicative: All of the models mentioned in today's post, from Riffusion to MusicLM to MusicGen, came after the article.

And the video stuff outlined in the post was something I didn't expect we'd see available to the general public even in a year...and now we have ModelScope and Runway Gen-2 already.

Thanks for chipping in!

Expand full comment

Whoa! GPT 3 really was revolutionary, though. It just took the rest of the world a minute to catch up!

Expand full comment

Indeed. Back then when I told people how magical GPT-3 seemed, I sounded like a bit of a lunatic talking about some voodoo spells. But then ChatGPT came out and everyone could try it for themselves for free, and the world lost its collective shit.

Expand full comment

I'm definitely an early adopter (but not quite an innovator) with most technology. GPT was no exception for me, so I got to live in the tinfoil hat camp for a few months while I told everyone who would listen that this was going to change the world forever.

Of course, then GPT4 came out.

Expand full comment