Remix.run Logo
▲ xmodem 5 hours ago

"I worked on the physics engine, tests, via widely accessible libraries." is a nonsensical sentence as written, but a colleague who told me they had done this would be telling me a vastly different thing to if they had told me they had compiled the source code to a physics engine.

There is a massive difference between prompting "security audit this codebase, make no mistakes", performing an in-depth agent or LLM-assisted focused search for security relevant code, or otherwise manually reviewing code as we did in the pre-LLM era. It behooves us to be honest about our own personal contribution to our output - both towards others and to ourselves.

▲tkz1312 4 hours ago | parent [-]

I don't disagree with you necessarily, and I do think it a good practice to disclose how much of a contribution was llm generated, but regarding the difference an individual makes, I see a huge gap between different colleagues in how well they are able to use llms and the quality of the results they are able to produce with them, so I am increasingly inclined to view them as something closer to a compiler or SMT solver than a contributor in of themselves.

▲boie0025 2 hours ago | parent [-]

I am also finding this disparity. I hand off the agent skills I use daily, to produce the results I want in the way I want them, to some people and I still get poor results, and others I get good results. There's something about a combination of tools, prompting and FTGs about the outcomes that combine into these different results. I should imagine at some point that will disappear?