We could technically do it now, but you’d need a ton of quality control.

A basic workflow:

Use an image generator to generate start (and optionally end) frames of each scene.

Use a video generator like Runway to create the scene.

Use something to generate the speech, either a realistic text-to-speech engine or a record the dialog yourself and use AI voice changing software.

Use a lip synch AI to match the mouth flaps to the audio.

Generate music using something like Udio

Job done.

All these technologies exist and in some capacity are available to use. The only issue at the moment is consistent quality. This is coming on in leaps and bounds, however.