As a hypothetical case, lets assume LLMs get good at coding and someone builds millions of lines of code that exactly duplicate Photoshop or MS Office and its all open source. People start using these LLM clones in masses and stop paying those companies. Let’s see how they react then, because they always claim that’s just how humans learn and that it isn’t copyright infringement. There shouldn’t be any github take downs or copyright infringement lawsuits then, right?
@nixCraft hmm but it isn't about what makes logical sense, it's about power.
poor people don't have the power to stop AI corpos from stealing their images => so it's allowed
rich corpos do have the power to stop the plebs from having an LLM steal their software => so it won't be allowed
@bazkie yes i thought so.
@nixCraft I actually look forward to this scenario. So much fun to watch it.
@nixCraft as I've said before, some people will get very rich out of this. Those people being lawyers.
@nixCraft As an open source dev of 10 years, I've thought about this a lot.
I think copyright was the wrong solution from the start.
What should be protected isn't the final code of a project, it's the work that went into it - the human time (which I am told has inherent moral value).
It's always harder to be the first to do something because you explore dead ends. SW dev isn't about writing code but making decisions and finding out what works and what doesn't. 1/n
@nixCraft I mean that's the reality of a world where LLMs deliver on the promise of "solving" programming. Software no longer has any value. All companies whose main capital is IP are dead in the water.
Good question. As I know some people who are very fond of the idea that in the nearer future one has only to ask a language model to build "on the fly" the software currently needed: office tools, video chat pps, (banking apps?), etc. Of cause this has to take place locally, not in a cloudy environment.
Total freedom, isn’t it?
@nixCraft who/what will ML models learn from if human coders stop coding?
It can only mimic training data, and has no concept of opinionated design or the kinds of insight that humans have into what will be useful, and teaches new generations of developers to avoid original thought.
New devs don't need 50% incoherent and incorrect eco chambers, they need education and actual comprehension of the problems they're trying to solve.
@nixCraft If it's a true clean room implementation I'd say good luck with the lawsuits! And good luck getting it off the the thousands of systems already using it 😂 . The damage is done. The gate has lifted
@nixCraft This is happening today, and you're right this will accelerate. Further let's say a cease and dissist is issued for infringement of x,y, and z. In a few hours you and your AI have refactored the infringement away, Let's go 💪
@nixCraft Interesting. This is actually happening and what's legal right now is that the AI must not see the original source code and it can try to replicate the functionality of the original products. Underlying code will be completely different, and that's why it's legal (for now).
@nixCraft Bet there are a bunch of lawyers rubbing their hands.