Using AI to correct spelling mistakes
- Thread starter nodle
- Start date
Yes, this is possible if you closely audit the first batches until perfect and create a skill. You'd probably want an adversarial agent double checking before commit.
Do mind that AI injects fingerprint into text by default and this needs to be avoided. The agent itself is unaware. Do a check of the output for hidden unicode, etc.
Do mind that AI injects fingerprint into text by default and this needs to be avoided. The agent itself is unaware. Do a check of the output for hidden unicode, etc.
I must admit I've struggled for a long time trying to decide the best way to go with this. In my ideal world people would use a spellchecker at the time of posting and perhaps re-read their post before submission but...Keep the spelling mistakes. It helps show the content is legitimate.
On the one hand I don't like to mess with member's posts while on the other a perceived lack of literacy might unfairly define a member.
I'd be interested to here other opinions on this, especially as set out by the OP.
I've made it very clear to my members that we want to hear their "voice" and their opinions, not something sanitised by an AI system. What they write doesn't need to be perfect - it needs to be genuine.
The spelling mistakes and quirks of the way they write is what identifies a post as uniquely theirs. If everyone started sounding the same - then we're just reading a textbook rather than having a discussion.
Remove the humanity (the imperfections, the quirks, the mistakes) from the discussion and what have you actually got? Not something worth keeping in my opinion.
The spelling mistakes and quirks of the way they write is what identifies a post as uniquely theirs. If everyone started sounding the same - then we're just reading a textbook rather than having a discussion.
Remove the humanity (the imperfections, the quirks, the mistakes) from the discussion and what have you actually got? Not something worth keeping in my opinion.
