alt Hacker NewsYou have good points regarding how copyright works.
> Software is usually obvious.
Hardware and mechanical designs are usually described in CAD programs nowadays, so it comes pretty close to software; it's just that LLMs are not the right tool to "GenAI" them but I've seen plenty of these kinds of design that I know for sure that they are often not any less obvious than a lot of software. Treating software as "obvious therefore not patentable" is not accurate and not fair and is probably not going to help the profession in the AI age. But I agree that patents are bad for innovation.
It is also not fair to claim that an AI-copy is fundamentally different from photocopying.
I mean, in both cases it is like you are picking the worst case interpretation for the field of software engineering.
> I think we'll have to wait for this stuff to shake out before anyone really knows what the rules will end up being.
Yes, but it will help if we think deeply about this stuff ourselves because what law-makers come up with may not be what the profession needs.
> It is also not fair to claim that an AI-copy is fundamentally different from photocopying.
If you clean-room copy it, I think it is different. Eg, first get one agent to make a complete spec of what the program does. And a list of all the correctness guarantees it meets. Then feed that spec into another AI model to generate a program which meets that spec.
The second program will not be based on any of the code in the first program. They'll be as different as any two implementations of the same idea are. I don't think the second program should be copyrighted. If it should, why shouldn't one C compiler should be able to own a copyright over all C compilers? Why doesn't the first JSON parsing library own JSON parsing? These seem the same to me. I don't see how AI models change anything, other than taking human effort out of the porting process.