2. Generative models for video games - https://aidungeon.io/ is barely scratching the surface. Story, art, animation, music, gameplay, it will all be generated by models in the future.
3. New direct drive robotics actuators such as https://www.google.com/search?q=peano-hasel+actuators I think actuators are holding robotics back more than software now. Breakthroughs are needed. No general purpose robot will ever be practical with electric motors and gearboxes.
4. Self-driving cars are still happening, despite delays. I think discounting Tesla's approach is a mistake, but Waymo is still in the lead.
5. NLP is finally starting to work. The potential for automation is huge. Code generation is very exciting as well.
6. I was excited for Rust but I now believe it's too complex. I'm looking for a much simpler language that can still achieve the holy grail of memory safety without GC pauses or refcounting. But I'm not holding my breath. If ML models start writing a large part of our code then the human ergonomics of programming language design will matter less.