4 ms·
Boring is good. If there’s AI in the product, I want it to be 100% accurate. Not 99% accurate. Zero hallucinations. I use MacOS because it’s very predictable an
by chomp 2y ago
Boring is good. If there’s AI in the product, I want it to be 100% accurate. Not 99% accurate. Zero hallucinations. I use MacOS because it’s very predictable and I don’t have to tinker. If this means dialing back to only a subset of LLM capability then that’s perfect.
- wilg 2y agoYeah but there's no evidence to support its going to be 100% accurate. Also accuracy is barely relevant to these features.
- semireg 2y agoSuper ducking accurate.
- jachee 2y agoTell me you haven’t used Autocorrect much since iOS17 without telling me. It really has gotten a lot better. Even lets me lazily type fairly me-specialized things once I made enough manual corrections. Like iRacing iRating RocketLeague carball etc. It sucked at first because it lost all the prior training, but it retrained fairly quickly. I almost never have to hit backspace these days.
- ronsor 2y ago> 100% accurate > Zero hallucinations I hate to break it to you, but you're not getting that from any software, let alone probabilistic AI models