Yes! So cool. We should be asking, can we train a model with new algorithms on a limited data set such that an 1800s AI can think itself through discoveries into the 1900s and beyond Also, it might go down tech trees we’ve long abandoned, insights from an alternate timeline.