Recent searches
Search options
I still see really smart programmers arguing that LLMs are not useful. I don't get it.
I think there are tons of valid arguments against LLMs, against using LLMs, against the ethics of using LLMs based on their training data, against the unbelievable hype and bullshit around AI, etc., etc.
Even that thinking should not be delegated, or that coding is not the bottleneck.
But that they're not useful? What is going on? Have they just not tried seriously?
The other day, during a CTF, once I'd figured out the stack smashing exploit I wanted in part 1, and how to calculate the ASLR offset in part 2, ChatGPT wrote the payload for me.
I didn't know about `pwntools` before, and being able to write assembly inline like it's 1991 using Turbo Pascal again is :chefs-kiss:
I also have never really internalized x86 or ARM assembly, registers, syntax, calling conventions, etc., so I was glad it handled that.
Just incredibly, astonishingly useful.
Could I have done all that without an LLM? Yes. I've done it before. But it probably saved me an hour, and learning about `pwntools` was fantastic.