These arguments are a strawman Filip; there are other tools for this and they work fine. (I wouldn't want to use a new data-structure that's not in a reliable library or shipped as standard.) These ignore main purpose that LLMs can be used for today…
@alexjc
-
AI Code Golf: Refactoring and Bug Detection in ML
By
–
Most significant is the API-compliance "code golf" type of code, which is so prevalent in ML. It's able to handle refactoring much more reliably, and helps find tricky bugs as a first-pass sanity check… I also let it write first pass of most new scripts, but they change a lot.
-
Using LLMs to Simplify Complex Shell Commands
By
–
I also use LLMs for stupid shell commands that are not as intuitive as they should be. This also cuts down on looking up the man page or searching Stack Overflow.
-
Open-source LLMs Transform Coding Practices with Clean Copyright
By
–
First I'd like to point out that there are permissively-sourced LLMs for coding, and there's nothing technical stopping all of them from having a clean Copyright chain of title. With that said, LLMs have fundamentally changed the way I code…
-
AI Finetuning Abuse: Illegal Content Generation Statistics
By
–
Legend* for the graph:
– Blue shows finetunes for nudifying celebrities, undressing teens, oversized boobs, underage anime, tentacle p0rn…
– The other colors (red or green) represent legitimate uses. *: This is a joke but not really. Anyone keen to compile real statistics? -
AI Algorithms Creating Personalized Bubbles Fragmenting Society
By
–
Great (depressing) talk by Jack Conte about the death of the follower. https://
youtube.com/watch?v=5zUndM
fMInc
… I'd go further to say that the 2020s are increasingly about "Generating" a complete bubble custom for you, disconnecting everyone from everyone else. (Dead internet theory weaponized.) -
EU Law Disclosure Requirements Override Business Privacy Interests
By
–
Posting here for reference: "The interests of EU-wide law outweigh private interests." If you can make an argument that failure to disclose violates many other laws (it does — there's a case), then business interests would become secondary.
-
AI Models Face Legal Challenges and Potential Removal
By
–
It may be that some models will be deleted because the company publishing realises that it wasn't legal as they thought. Others may go to court in order to defend their work… I don't think there will be "absolute" outcomes for all existing models simultaneously!
-
AI Training Copyright Law and Regulatory Enforcement Challenges
By
–
My post implies that training is under Copyright, so many conditions from the law will need to be respected. It's allowed in certain cases. I don't expect the AI Office to proactively police Copyright so lawsuits would be needed still.
-
Government Establishes AI Office with Standardized Templates
By
–
I believe they are staffing up an AI Office. There will be templates published so it's somewhat standardized…