what do you think ,I think I need use more Scala to save money.
Yeah that’s about what I’ve been expecting.
AI may well be Scala’s killer app in the current world: the expressiveness that’s always been its calling card for human programmers means that it is both more precise and more concise, which means that it should be cheaper for writing and maintaining code in an LLM-centric environment.
(Both because the code is smaller and because idiomatic strongly-typed Scala code tends to require fewer tests than many other languages do: a lot of those tests can be obviated by the type system.)
1 Like
Gordon Cooke’s written an article that makes the same point: Why Scala Is the Best Backend Language for AI-Assisted Development — Scala Teams .
@hepin1989 - do you have a reference for that 40% number?
2 Likes
I’m not sure I accept this take at face value, but it might be persuasive way of getting Java shops to cut over wholesale to Scala - “If you do this, think of all the money you’ll save on LLM fees.”
They might burn some tokens doing that wholesale port, though.
I’d be interested to hear how heavy use of LLMs for commercial work stacks up against having a lot of grunts in terms of outgoing costs. Any bean counters in the group? (I don’t mean Spring developers.)
Given the degree to which it’s all funny-money so far, it’s a little hard to say. But my guess is that it’s a major win if the people using the LLMs are well-trained in how to use them effectively.
Keep in mind, “lots of grunts” don’t actually tend to produce very good code for the long run, at least for interesting problems – the Mythical Man-Month has always been an important consideration. Throwing more people at a problem often results in going slower and getting worse results.
(And even a low-level programmer costs pretty real money.)
Yes, I’m assuming some folks are paying serious money and are prepared to spill some of the aforementioned beans. That’s probably a bit naive of me!