AI Considerations #256
Description
We've always referred to netnet as an "AI-TA" (artificial intelligence teaching assistant)... but that was before the rise of LLMs... now AI has all sorts of connotations we need to contend with.
wherever possible we need to start differentiating netnet from these other AIs (like ChatGPT), first by explaining netnet is "classical AI", meaning coded by hand (something like an expert system?) rather than an ML trained neural-network, && thus does not suffer from the same issues like bias, hallucination, etc.
we also want to stress that netnet is different from code editors that have LLMs built in (like Copilot in VScode, tabnine, replit, cursor, etc), netnet will NEVER write code for u, it will only ever help u write ur own code (ie. edu info, annotated examples u can remix) even when using a widget like the Color Widget or a CSS code generator, it's always about small snippets u can understand && piece together urself.
should we ever decide to integrate an LLM it should be for educational purposes. for example, we might consider creating some sort of tutorial where netnet teaches u about LLMs && maybe shows u how u can super-charge it by plugging into some AI-API urself (if not by integrating ur own local LLM), we could use something like hugging face's inference API. the goal being that creatives understand what LLMs are, what their strengths/weaknesses are && have as much agency as possible in how to integrate that into their coding process.