I wouldn't call it the new high level language. It's a new JIT. But that's not doing it justice. There's a translation of natural language to a tokenizer and processor that's akin to the earliest days of CPUs. It's a huge step change from punch cards. But there's also a lot to learn. I think we will eventually develop a new language that's more efficient for processing or multiple layers of transformers. Tbh Google is leapfrogging everyone in this arena and eventually we're going to more exotic forms of modelling we've never seen before except in nature. But from an engineering perspective all I can see right now is a JIT.