Work flagged as plagiarised AI content when using speech to text app

That really is a fantastic result. Well done. Glad it worked out for you. Impressed with the college that they have listened and changed accordingly as well. 

Well done; finally a happy ending..

Thank you for asking, the meeting was a lot of questions from them and I ended up showing them how I use the app to aid my lesrning etc. ANYWAY end result, they are changing their AI policy as a direct result and I do not need to repeat any work! 

It was a lot of talking and it went on over the week, but I think this is a fantastic result for future learners and I’m also happy I didn’t need to repeat my coursework (though they did test my knowledge verbally which I was happy to do so to help show I did the work).

I am very happy they are changing the policy though!

I was completing an assignment today and remembered this post. Just wondered if you'd had your meeting and how it went?

Perhaps you are… thinking about it from my point of view (strange saying) I have never met you. Only interacted with you via an online forum. We could all be AI in each others reality. Cool.

I don’t think I’m interesting enough to be hacked - I’m just an abnormal spaniels sidekick. ‍️

 That just reminded me of this 

I agree. I suppose it’s difficult for them with that side as the last time I did a course AI wasn’t as advanced as it is today. I think we barely had Siri (yes I’m that old )

It seems unethical that they are relying on AI to flag something and without context (other essays).

At the very least, they should cross-check with other essays of yours. If old essays were flagged as "AI generated" that at least should count as counterproof.

 Honestly I've often thought my brain doesn't work dissimilarly to AI sometimes (just with a lot less access to knowledge).

Or you are in the matrix and have been hacked...

Expecto_Patronum said:

There are certain features that AI often uses, it can be as simple as dashes in the writing.

Does that mean—and I'm not being serious here—that I might be an AI? I never looked at myself that way before—it would probably explain a lot.

This is brilliant, thank you so much for sharing 

For future situations, it is probably worth seeing if you can self-assess if your document is flagged as AI generated. There is a decent article on what they look for and why - plus a few links - here: https://www.thesify.ai/blog/how-professors-detect-ai-in-academic-writing-a-comprehensive-student-guide

In particular there is a talk about false positives because the tools look for:

False Positives: Non-native English speakers or students with unique writing styles may have their work incorrectly flagged as AI-generated.

A common issue for AuADHD people is our communication styles are different and this is a good angle to approach the case from.

This can be an issue for definite. It is worth like the others have suggested questioning what their proof is that you used AI. 

It may also be worth asking if they can show you policy that states you cannot use speech to text software. 

The difficulty with this particular issue is that it wasn't previously agreed that you could use the software. In a school any adjustments for coursework/exams have to be previously agreed. I do not know if this changes in college etc. However, if it isn't stated that you cannot use it in policy, this may be a loophole. It gets a bit complicated because it likely does use AI but as the AI isn't used for the content then you may have a case.

See how the meeting goes. And do reach out to student support if you are able to. I hope you manage to get it sorted.

This! Ok so I just looked into the AI checkers and they aren’t foolproof. If you use the same pattern in writing over the answers it flags it, or if you have common words AI like to use. I tested a random email I wrote and 3/4 checkers flagged as 70% or more AI generated content!

It’s my usual style of writing so nothing out of the ordinary. I wouldn't know how to prove it isn’t AI generated, they didn’t accept my option of a verbal question and answer over teams with my ‘tutor’ (who has had no interaction with me other than this) which I said they could record for auditing and fairness. 

squishylea said:

They didn’t say anything like that, they just said ‘because of how it is written it cannot be accepted,’

Before speaking to the examiner again I would find out who in your educational establishment looks after students with special needs and talk this over with them.

They may not realise how much they are required to adapt to cater for autists and I'm sure the support team will be able to clear this up.

If you can prove the date of your voice recording and the date of the transcript then you can prove it came from you. 

Should all this work then you can carry on doing what you are doing, just make sure you get the approval in writing and keep an audit trail of all your recordings and transcripts, generated by the computer and backed up somewhere offsite just in case you ever need to deal with a new examiner again.

If nothing else, you now at least know what you can use even if it is a lesser product. Sometimes life sucks this way but all we can do is understand and adapt. 

Good luck in getting the support you need.

I would request to them a detailed proof, or at least challenge that it is AI generated. "Sounds like AI" is a not a proof; even if it is written like an AI, that's not solid evidence of anything.

However, if this is very different to your usual writing style, then that may be of some evidence.

I would also tell them that I can defend my work, if they have doubts about it.

There are certain features that AI often uses, it can be as simple as dashes in the writing. It may be that although it is your words, the style the app has used to put it into writing flags it as AI.