The drum machine and synthesizers did not replace drummers and real instruments.
In the long run, wavetable synthesis and MIDI did, but only to open the possible options because most instruments are difficult to make in the first place. If you can build a synthesis keyboard that can play every instrument correctly, albeit without the fine analog control of a human, that means "anyone" can play the same music that sounds the same every time. This is why orchestras still exist. Orchestras always sound better than pre-recorded music, but to enjoy you you have to actually BE in the orchestra's audience. There is no way to hear and feel that.
It's the same with theatrical plays versus watching something on blueray/dvd/VHS/film. You do not get the experience of being there. The parts that make it human.
You can argue that computer generated special effects, scenery are better than practical ones, in SOME cases they are when it would put the performers or audience in harms way (example "the phantom of the opera" not only has fireballs, but has a chandelier drop on the audience that is on a cable so that it doesn't actually hit the audience.) You can only do that practical effect in a real theatre. Not only that, there is a risk in the real theatre of being "too close" to these practical effects for comfort. In a small theatre Phantom of the Opera can't do most of these practical effects, so they... use a computer to project those effects.
This is how I see "AI" shaping out in the end. That it brings the ability for people to create works to more people, particularly in locations where the software or people are impossibly priced out of the market. That said, the cost to use "cloud" AI products is far more than paying humans and buying the software that you'd normally make music, animation, television and film with. You will never be able to produce an entire video game, television show or film with the AI we have now because there is no way to maintain a context window longer than a few seconds. This is why so many "AI Slop" things are short incoherent videos. When it comes to music, so far all AI does is "re-autotune" a signer, or predict the next "sound" of a song. It doesn't create anything, so all you can do with AI generated music is create derivatives of existing works. It takes a human to write music. Always. Even algorithm, procedurally driven music (eg portal) ultimately has a human behind it.
This is why people who promote AI, sound like snake oil salesmen. They keep promoting AI to replace human labor or human creativity, but fail to mention that it requires creativity to understand why something IS or IS NOT creative. Many of the AI-slop loving people only use it for making hate/cyberbullying material, because for that stuff to work on it's target it only requires evoking an emotion in the person who sees it. It doesn't require understanding.