Developer prefers Jev because constrained classifiers stay controllable
A programmer says they favor Jev over open-ended LLMs because its constrained judgments are controllable and let them write small domain-specific languages on top.
Jev is safe. Jev makes sense. It’s just a classifier. It has so many use cases. You get to use it like a tool in your toolbelt. Remember that feeling? You can constrain Jev! In fact it is only ever constrained. We don’t always need open ended LLMs. Maybe we only rarely need the full power of an LLM. Maybe we can feel safe and special again. Maybe this is all a bubble. I knew this was just a bubble. LLMs just make slop anyways. I like Jev, where I’m in control of the responses. I can’t wait for all the things *I* will code with Jev. It’s called machine learning and it’s a good tool in your toolbelt. I’m a programmer who knows a lot of things. I’m an expert. I’m in control and my future is secured. I like Jev because it’s this side of the stochastic parrot uncanny valley. Jev makes me feel like 2022 again. I can write funny little domain specific languages on top of Jev. Syntax matters! I identify with knowing the arbitrary intricacies of coding, and combining things and being on top of the hottest new tools to add to my toolbelt. I’m an expert programmer and a smart human. The future is just more of the same. With a little bit of machine learning. You should try Jev!