I’m a historian with a PhD and twenty years of experience in teaching a d research as a professor. You are making some big assumptions about what software can do in the future based on a flawed premise about how technology develops over time. Sure there might be more sophisticated software in
Five years. But there also might not be enough difference to justify that kind of software in economic terms. Another perspective is that the productivity gains made by first hardware and software over the past thirty years have reached a Pareto limit. All the easy gains have been achieved, more
incremental gains in productivity have become more expensive. At some point the value is not going to be there and the next technological revolution is going to happen in some completely different field. For now the TechBros have unlimited stacks of money they can set on fire to make billions.
I also have degrees up the wazoo (too many) and years of experience in the software field. The AI we have now is nascent. I'll bet you lunch on it and I'm not talking about LLMs specifically, that's only one small specific application of AI, I'm talking about the field in general.
Nothing grows forever. Everything regresses to the mean sooner or later. It’s the law of thermodynamics.
Yes, eventually, but I don't think we're there yet. There are so many AI applications in the sciences that are in progress or waiting to be coded.