Sub posting myself after having to delete a post which was too mangled to try and edit into submission: is anyone else finding using SwiftKey and seeing it getting massively less accurate in the last few weeks? Half of me is thinking it must just be my perception and I'm blaming the tool, the other half is looking at Microsoft's current release standards and wondering if they've changed how they're doing the recognition... (it happens on device, so I doubt it is actually llm related)
Post