While AI-based proofreading tools have undoubtedly advanced in recent years, they are still not able to identify some errors that human proof-readers can. Which is great news for us a memo!

We do enjoy spotting the things that AI misses and we have made a list!

Here are 10 things that us human proof-readers can spot, that AI can't:

1. Incorrect homophones: Words that sound alike but have different meanings. Now we have to say this is getting better (with there/their and they're in particular) but there are other instances such as Cell/Sell: "The lab results showed abnormal sell counts" where it still just doesn't spot it.

2. Incorrect verb tenses: AI tools may not always identify incorrect verb tense usage. It certainly doesn't spot when verbs don't match in a list such as, "The client likes swimming, to run in the park and walked with his dogs." I mean, that is just all over the place!

3. Punctuation errors: AI may struggle with identifying errors in punctuation, such as misplaced commas or missing full stops. In my experience AI is militant about commas and wants as many as possible but doesn't always flag missing full stops which is strange to me.

4. Inappropriate word choices: AI may not be able to identify when a word is used incorrectly in context. This one is a favourite for sure and for us at memo, proofing medico-legal reports we see this a lot: "in jury" instead of "injury". AI would never spot that – we do.
   

5. Inconsistencies in tone, style or formatting: AI may not be able to identify inconsistencies in tone or style throughout a document but we can. Think dates in multiple formats, bullet points out of alignment etc.

6. Missing words: AI may not always identify when a word is missing from a sentence. A missing word can change the whole meaning of a sentence from, "She could climb the stairs independently." to "She could not climb the stairs independently." Pretty important wouldn't you say?

7. Inconsistent capitalisation: AI may not always identify when words are capitalised inconsistently or incorrectly. Medicines, job titles and occupations are common mistakes we see time and time again.

8. Misplaced apostrophes: AI may not always identify when an apostrophe is used incorrectly, such as in plural nouns. We see this a lot with names ending in s such as James. AI doesn't always recognise when the plural apostrophe is needed and will not flag it up.
   

9. Inconsistent spelling: AI may not always identify when spelling is inconsistent throughout a document especially names! This is a big one and a very embarrassing mistake to miss.

10. Finally, AI for some reason LOVES American spellings, z galore! When writing in UK English it is important to not assume the spelling suggested is the correct one and watch out for dictionary settings resetting themselves to US for some reason!

So there you have it, while AI-based proofreading tools do have their benefits (I have spellchecked this), they still have limitations when it comes to catching certain errors. Human proof-readers bring a level of expertise and judgment that AI tools just can't replicate...yet.

For how a human can help with your proofing needs contact us here.