I only find this is a problem when it writes code in a language I deeply understand. I can’t help but look at it, get the LLM to change it, etc. But when it writes me a Cloudflare worker in Typescript that I don’t understand, I couldn’t care less. It works, that’s good enough.
So I suppose we just have to care less and let it get on with it. Not easy tho, for sure.
Perhaps BigCorp is only just starting to use agentic dev properly. Plus they’ll have loads of bureaucracy still to get code out of the door. Might take another year or two.
> They talk to me in this grating LLM-voice, an uncanny valley of talking to a real human
It’s entirely possible to configure an LLM to talk to you in a different voice of your choosing, eg. “when responding, use a professional, formal, business tone of voice”.
I think the use case here is for people who have an iPhone as well as an iPad they use occasionally for reading/watching stuff, as well as business people / execs.
Now you can (sort of) have both in one device. Reading the rumours, I was completely disinterested. Now I’m actually tempting to ditch the iPad Pro and iPhone Pro and have a Duo replace both. I’m excited to have a go with one in the Apple Store.
I feel like we're not talking about this too much. This phone-o-tablet is the same price as base Macbook Pro. Let me reiterate: a foldable phone costs as much as Pro line computer.
This is just crazy
Yes, LLM code gen and human review seems to be the universal standard... today! that may well change soon though as trust in the agents increases and they get better. Fable, for instance, is so good for what I do (Rails) that I pretty much trust it implicitly. I still like to check though :)
So I suppose we just have to care less and let it get on with it. Not easy tho, for sure.
reply